Q firflfia GI: Li'C‘f’7 ABSTRACT EFFECTS OF C(MPANY ORGANIZATION STRUCTURE ON THE ACCOUNTING SYSTEM BY Kenneth Yale Rosenzweig A fundamental assumption of accountants and accounting authors is that internal accounting systems must be designed in accord with the organization structures of their companies. Another fundamental assump- tion is that an important role of internal accounting systems is provid— ing information for the control of their operations. Yet surprisingly, these assuntpions have not been extensively examined. In fact, little research has been done on the structure of accounting systems themselves or their links to the organization structures of their companies. More- over, hardly any research has been done on how the control function of accounting relates to other company control systems. Thus much needs to be learned about whether characteristics of accounting systems relate to those of organization structure or other control systems. Furthermore, the nature and strength of these relationships should be studied. The purpose of this dissertation is to investigate the associa- tion of properties of the accounting system and those of the overall com- Pan)’ organization. In order to guide the inquiry, a basic model is con— Structed which incorporates some suppositions as to the effects of differ- ent types or levels of organizational characteristics on the accounting System. These levels are structural complexity, control systems (other than the accounting system), and process. Structural complexity refers ~4- . Kenneth Yale Rosenzweig to the extent to which an organization is divided into parts on various dimensions. For instance, one dimension of structural complexity is the nunber of departments at various levels of the organization. Control systems include structures of an organization, other than its accounting system, which help control and coordinate the organization's operations. An example is standardized procedures by which management controls opera- tions by establishing authorized ways of doing things. Process refers to the level of technology of the production operations of the organization. The directions, positive or negative, of effects on the account- ing system of the three levels of organizational variables are predicted in the model. These predictions are incorporated in three hypotheses which are as follows: 1. Structurally complex organizations tend to have more fully developed accounting systems to contribute to the resolution of greater control and coordination problems. 2. The stage of development of the accounting system is nega- tively related to that of other control systems since con- trol systems are partial substitutes for one another. 3. The more sophisticated the production process of an organi- zation, the more developed must be the accounting system to provide more and better information for management decisions. The objectives of the research are: (a) to test the three hy- POtheses and thereby substantiate the model; (b) to determine the strength and direction of influence of characteristics of process, structural com- Plexity, and control systems on the accounting system; (c) to determine how much the overall organization influences different accounting system characteristics; (d) to determine if accounting systems can be conveniently classified into types on the basis of their properties and accounting- related organizational properties. ,0. Kenneth Yale Rosenzweig In this field study, a sample of eighteen small manufacturing companies was selected. The controllers or chief financial officers of the companies were interviewed about their accounting systems, structural complexity, control systems, and process. Over one hundred measurements were collected for each company. Since the basic model included about twenty organizational variables and about ten accounting system variables, there were several measurements for each. The multiple measurements of each variable were combined into a single measurement with the statistical technique principal components analysis. The relationships between each of the accounting system variables and those of the overall organization were calculated with the statistical technique stepwise multiple regression analysis. This technique finds subsets of the organizational variables which best explain each variable of the accounting system. For example, the technique may find three or four of the twenty organizational variables which, taken together, are most associated with accounting system size. The output of the regression analysis was used to calculate the measures "explanatory power" and "explainability," which were developed in the course of this dissertation. Explanatory power is defined as the ability of each organizational variable to explain accounting system vari- ables, taken together. Explainability is defined as the ability of each accounting system variable to be explained by organizational variables, taken together. Another measure developed in the course of this dissertation was "consistency with the hypotheses." This is defined as the extent to which found relationships conform to the directions, positive or negative, Kenneth Yale Rosenzweig predicted in the three hypotheses. Consistency with the hypotheses was calculated for accounting system variables and for organizational vari- ables. For example, a structural complexity variable such as number of departments is predicted by hypothesis one to have positive relationships to accounting system variables. Consistency with hypothesis one for num- ber of departments is the extent that any such relationships that are found are positive. In addition to the consistency measures for vari- ables, tests of the overall conformity of the research findings with each of the three hypotheses are performed. The basic model and the three hypotheses were revised in accord with the research findings of the types discussed in the preceding three paragraphs. In the course of the model revision, the respective roles of the three organizational levels with respect to the development of the accounting system were refined. Furthermore, many of the variables within the three levels were reinterpreted. Similarly, the accounting system variables were reinterpreted with respect to the extent and manner they are explained by organizational variables. The important findings of this dissertation are as follows. Structurally complex organizations do have more fully developed account- ing systems, as predicted by hypothesis one. Process is even more im- portant in determining the stage of development of the accounting system. Companies with more sophisticated processes have much more fully devel- oped accounting systems, as predicted by hypothesis three. The stage of development of the accounting system is negatively related to that of some control systems, as predicted by hypothesis two. However, it is positively related to the stage of development of other control systems. Kenneth Yale Rosenzweig These positive relationships have been interpreted as resulting from complementarity: information.produced by the accounting system helps these control systems function better. The research findings apply to the accounting system as follows. Overall organization variables primarily influence the output of account- ing systems (the nature of reports and where they are sent in the organi- zation). They do not have as much influence on the structure of account- ing systems. The only key feature of accounting system organization structure is the distinction between centralized and divisionalized accounting systems. In sum, the research findings show that it is im- [mssible to design a new accounting system or even to adapt an existing one without understanding the organization of which it forms a part. EFFECTS OF COMPANY ORGANIZATION STRUCTURE ON THE ACCOUNTING SYSTEM BY Kenneth Yale Rosenzweig A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Accounting and Financial Administration 1976 ® Copyright by KENNETH YALE ROSENZWEIG 1976 I dedicate this dissertation to my parents, Morris and Freda Rosenzweig. Without their guidance and support through the years, this dissertation would not have been possible. ACKNOWLEDGMENTS The eclectic nature of this dissertation, spanning three fields of study (accounting, organizational sociology, and statistical research design), created significant prOblems of integration of the diverse sub- ject matter. iMy dissertation committee provided knowledgeable, under- standing, and cooperative support during the development of the disser- tation proposal and.its implementation. Harold Sollenberger, chairman of the committee, helped assemble the dissertation sample. Furthermore, he was the primary impetus fOr organizing and clarifying early drafts of the dissertation. Harry Perlstadt, committee member from the Sociology Department, contributed much to the development of the interview ques- tions and to the tie-in of the dissertation with sociological literature. Maryellen.Mc3weeney, committee member from the Department of Counseling, Personnel Services, and Educational Psychology, assisted greatly in the selection and use of statistical methods and the development of the over- all research design. I would also like to express my appreciation to Doug Johnson, my colleague at Arizona State University. He provided valuable suggestions fOr handling various prOblems with the research design. I am gratefu1 for the patience and understanding of my wife Bonnie during the trying time when my dissertation was in progress. iv TABLE OF CONTENTS Page LIST OF TABLES ................................................... ix LIST OF FIGURES .................................................. xi Chapter 1. OVERVIEW OF THE STUDY .................................... 1 Purpose ................................................ 1 Objectives ............................................. 3 Research Design ........................................ 4 Revision of Medel ...................................... 5 Findings ............................................... 5 Organization of the Dissertation ....................... 6 2. REVIEW OF LITERATURE AND DEVELOPMENT OF BASIC MODEL ......................................... 8 Accounting Literature .................................. 10 Simon et al. ......................................... ll Golembiewski ......................................... 15 Caplan ............................................... 18 Sociology Literature ................................... 20 Pugh et al. .......................................... 22 Blau and Schoenherr .................................. 26 Development of Basic Mbdel ............................. 30 Desirable Features of the Model ...................... 31 The Levels and Their Interrelations .................. 33 Chapter Page Structural Complexity and Control System ........... 33 Context ............................................ 37 variables within the Levels .......................... 43 Structural Complexity and Control System Variables ........................................ 46 Context Variables .................................. 52 Accounting System Variables ........................ 53 Hypotheses as Incorporated in the Basic Mbdel ........ 57 3. RESEARCH DESIGN .......................................... 59 Source of the Research Data ............................ 60 The Sample ........................................... 61 The Respondents ...................................... 67 Refining the Data ...................................... 68 Collection and Disposition of'Measurements ........... 70 Principal Components Analysis ........................ 72 Example of Derivation and Interpretation of First Component ............................... 73 Derivation of Second Component fer Some Variables ........................................ 81 Rotation of First and Second Components ............ 84 Component Scores ..................................... 88 Analyzing the Data ..................................... 89 Stepwise Multiple Regression ......................... 90 Example of Multiple Regression ..................... 91 The Stepwise Procedure ............................. 98 Explanatory Power and Explainability ................. 100 vi Chapter Page Description of Frequencies and R Square Increase Methods ................................. 105 Comparison of Frequencies and R Square Increase Methods ........ . ........................ 114 Consistency with the Hypotheses ...................... 116 Frequencies Method ................................. 116 Consistency of Individual Components ............... 122 R Square Increase Method ........................... 124 Limitations .......................................... 127 4 - INTERPRETATION OF THE RESEARCH FINDINGS .................. 133 [Relationships within the Accounting System ............. 136 Classification as Input and Output Components ........ 137 Explanatory Power and Explainability of .Accounting Components .............................. 140 Path Analysis ........................................ 146 Output Component Cluster ............................. 149 Relationships among Input Components ................. 150 Relationships between Input and Output Components ......................................... 152 Alternative Types of Accounting Systems .............. 155 Relationships of the Accounting System to the Overall Organization ................................. 158 The Extent of Influence of an Organization on Its Accounting System ........................... 159 The Influence of the Explainer Levels on the Accounting System .............................. 161 The Influence of the Explainer'Variables on the Accounting System ........ . ......... . ........ 167 vii Chapter Page The Influenceability of Accounting Variables by Organizational Variables ........................ 175 S . INTEGRATION OF THE RESEARCH FINDINGS ..................... 181 Generalizations from the Research Findings ............. 181 Revision of the Basic Model ............................ 183 Reconsideration of the Organizational Levels and Variables ........................................ 185 Reconsideration of the Accounting System Variables ............................................ 194 Revision of the Hypotheses ............................. 197 6- SIMARY AND CONCLUSIONS .................................. 200 Review of the Steps in the Inquiry ..................... 200 Important Findings ..................................... 202 Implications for Accountants and Managers .............. 205 Proposed Future Research ............................... 208 APPENDICES A' QUESTIONS FOR INTERVIEWS AND MEASUREMENT RULES ........... 212 13“ INTERPRETATION OF COMPONENTS ............................. 239 C - OTHER INFORMATION ........................................ 273 SELECTED BIBLIOGRAPHY ............................................ 302 viii LIST OF TABLES Tab 1e 1- Characteristics of Sample Companies ...................... 2. Populations of Cities from Which Sample Companies Were Selected .......................................... 3. 4' Component Description .................................... 5 ° Regression Coefficients for the Stepwise Regressions 0f the Thirteen Accounting System Components on Nineteen Potential Organizational Explainer Components ............................................. 6 ' l=recluencies of Regression Coefficients of Accounting Ystem Components on Organizational Explainer Components ............................................. 7' R SQUare Increase of Organizational Components Admitted to Regression ................................. 8 ’ FrECmencies of Coefficients Consistent with otheses ............................................. 9 . . . . Pr(Sportion of R Square ConSIStent With the Hypotheses .............................................. 10 . Intercorrelations of Organizational Explainers ............ 11. . . . . . Regressmn Coeff1c1ents for the Stepmse Regressmns of Each of the Thirteen Accounting Components on the Other Twelve Accounting Components .................. 12' Frequencies of Regression Coefficients of Accounting Components on Each Other ................................ 13' R Square Increase of Accounting Components Admitted to Regressions on Each Other ............................ 14. Explanatory Power and Explainability of Accounting Components with Respect to Other Accounting Components . . . ix Titles of Respondents .................................... Page 62 66 69 74 101 106 108 117 125 132 138 141 142 143 Tab 16 Page 15 . Explanatory Power and Consistency with the Hypotheses of Explainer Levels ........................... 162 16 . Explanatory Power and Consistency with the Hypotheses of Explainer Components ....................... 169 17 . Explainability and Consistency with the Hypotheses of Accounting System Components .......................... 176 18- Regression Coefficients for the Stepwise Regressions of the Thirteen Accounting Components on Nineteen Potential Organizational Explainer Components—— Rows and Columns Rearranged per the Revised Basic Model .............................................. 186 19 - Assignment of Measurements to Principal Components Procedures ............................................... 274 20 ' Measurements Included in Components ........................ 293 21' Zere-Order Pearson Correlation of Accounting Components 011 Each Other ............................................ 300 22 ‘ z'eI‘CP-Order Pearson Correlations of Accounting Components Wlth Other Components .................................... 301 LIST OF FIGURES Figure Page 1 . The Basic Model ........................................... 44 2 . Path Analysis of Accounting System Interrelationships ...................................... 148 3. The Revised Basic Model ................................... 184 xi Chapter 1 OVERVIEW OF THE STUDY .A.fundamental assumption of accountants and accounting authors is that internal accounting systems must be designed in accord with the 0I'ganization structures of their companies. Another fundamental assump- ton is that an important role of internal accomting systems is provid- ing information for the control of their operations.1 Yet surprisingly, these assumptions have not been extensively examined. In fact, little reseamh has been done on the structure of accounting systems themselves or their links to the organization structures of their companies. More- Over, hardly any research has been done on how the control function of accounting relates to other company control systems. Thus, much needs to be leél‘rned about whether characteristics of accounting systems relate to those of organization structure or other control systems. Further- “10 re, the nature and strength of these relationships should be studied. PURPOSE The basic thesis of the dissertation is that the nature of the a . . . Ccomting system 15 a result of the influence of the overall company \— (1i 1These assumptions may be inferred from the general tone of the .SCUSsion in Charles T. Horngren, Cost Accounting, A Managerial Empha- 8“ (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1972), specifically PP- 157-58 on the first assumption, and pp. 5 and 157-59 on the second. organization. The purpose of the dissertation is to investigate this influence. In order to guide the inquiry, a basic model is constructed which incorporates some suppositions as to the effects of different types or levels of organizational characteristics on the accounting system. These levels are structural complexity, control systems (other than the accounting system), and process. "Structural complexity" refers to the extent to which an organization is divided into parts on various dimen- sions. For instance, one dimension of structural complexity is the number of departments at various levels of the organization. "Control SYStems" include structures of an organization, other than its accounting System, Which help control and coordinate the organization's operations. An example is standardized procedures by which management controls opera- tions by establishing authorized ways of doing things. "Process" refers to the level of technology of the production operations of the organiza- ti0n. The directions, positive or negative,1 of effects on the account- 111g Sys'tem of the three levels of organizational variables are predicted 1n the model. These predictions are incorporated in three hypotheses whlch are as follows: 1. Structurally complex organizations tend to have more fully developed accounting systems to contribute to the resolution of greater control and coordination problems. \‘_—__ 1"Positive direction" means that, when the stage of development of the organizational level is high, that of the accounting system tends to be high. "Negative direction" means that, when the stage of develop- ment of the organizational level is high, that of the accounting system tends to be low. The stage of development of the accounting system is nega- tively related to that of other control systems, since control systems are partial substitutes for one another. The more sophisticated the production process of an organi- zation, the more developed.must be the accounting system to provide more and better information for management decisions. This is a diagram of the predictions in these hypotheses: Structural Complexity Sophistication Control of Process + Systems \ / + _ \ / Stage of Development of the Accounting System OBJECTIVES The objectives of the research are: 1. 2. To test the three hypotheses and thereby substantiate the model. To determine the strength and direction of influence of characteristics of process, structural complexity, and control systems on the accounting system. To determine hOW’mUCh the overall organization influences different accounting system characteristics. To determine if accounting systems can be conveniently classified into types on the basis of their properties and accounting-related organizational properties. RESEARCH DESIGN In this field study, a sample of eighteen small manufacturing (:(3nq1anies was selected. The controllers or chief financial officers of the companies were interviewed about their accounting systems, struc- tural complexity, control systems, and process. Over one hundred meas- urements were collected for each company. Since the basic model included about twenty organizational variables and about ten accounting system variables, there were several measurements for each. The multiple measurements of each variable were combined into a single measurement With the statistical technique principal components analysis. TFhe relationships between each of the accounting system variables and thOSe of the overall organization were calculated with the statisti— Cal teamique stepwise multiple regression analysis. This technique finds subsets of the organizational variables which best explain each Variable of the accounting system. For example, the technique may find three 01‘ four of the twenty organizational variables which, taken to- Esther, are most associated with accomting system size. The output of the regression analysis was used to calculate the meaSUres "explanatory power" and "explainability," which were developed ill 1318 course of this dissertation. "Explanatory power" is defined as t&“3 ability of each organizational variable to explain accounting system 'variables, taken together. "Explainability” is defined as the ability of each accounting system variable to be explained by organizational Variables, taken together. These definitions of terms are used con- 5.. siStently throughout this dissertation. Another measure developed in the course of this dissertation was "consistency with the hypotheses." This is defined as the extent to which fOund relationships conform to the directions, positive or nega— tive, predicted in the three hypotheses. Consistency with the hypothe- ses was calculated for accounting system variables and for organizational variables. For example, a structural complexity variable such as ”number of departments" is predicted by hypothesis one to have positive relation- ships to accounting system variables. Consistency with hypothesis one fOr "number of departments" is the extent that any such relationships that are found are positive. In addition to the consistency measures fOr variables, tests of the overall conformity of the research findings with each of the three hypotheses are performed. REVISION OF MODEL The basic model and the three hypotheses were revised in accord with the research findings of the types discussed in the preceding three paragraphs. In the course of the model revision, the respective roles of the three organizational levels with respect to the development of the accounting system were refined. Furthermore, many of the variables within the three levels were reinterpreted. Similarly, the accounting system variables were reinterpreted with respect to the extent and manner they are explained by organizational variables. FINDINGS The important findings of this dissertation are as follows. Structurally complex organizations do have more fu11y developed accounting systems, as predicted by hypothesis one. Process is even more important in determining the stage of development of the account- ing system. Companies with more sophisticated processes have much more fully developed accounting systems, as predicted by hypothesis three. The stage of development of the accounting system is negatively related to that of some control systems, as predicted.by hypothesis two. How- ever, it is positively related to the stage of development of other control systems. These positive relationships have been interpreted as resulting from complementarity; information produced by the account- ing system helps these control systems function better. The research findings apply to the accounting system as follows. Overall organization variables primarily influence the output of account- ing systems (the nature of reports and where they are sent in the organ- ization). They do not have as much influence on the structure of ac- counting systems. The only key feature of accounting system organiza- tion structure is the distinction between centralized and divisionalized accounting systems.1 In sum, the research findings show that it is impossible to design a new accounting system or even to adapt an exist- ing one without understanding the organization of which it forms a part. ORGANIZATION OF THE DISSERTATION The first chapter is a brief nontechnical outline of the disser— tation. Literature relevant to the topic of this dissertation is 1Centralized accounting systems have a single accounting office at the company headquarters, while divisionalized accounting systems also have accounting offices at division headquarters. mwmined in Chapter 2. From this examination emerged ideas for the basic nwdel and hypotheses which are developed in the latter part of Chapter 2. The research design is elaborated in a step-by-step manner in Chapter 3. The statistical techniques are described using simple examples from the analysis of this dissertation's data. The emphasis is on the way data are manipulated rather than mathematical complexities. The detailed findings of the study are in Chapter 4. These are of two types: the interrelations among accounting system variables and the relationships of organizational to accounting system variables. Chapter 5 attempts to integrate the detailed research findings of Chap- ter 4. .A revised model is developed which incorporates the expected and unexpected findings. The steps of the inquiry and the important findings are reviewed in Chapter 6. Also, implications for company accountants and managers are suggested. Finally, follow—up research is proposed. Chapter 2 REVIEW OF LITERATURE AND DEVELOPMENT OF BASIC MODEL The purpose of this chapter is to review the literature in both management accounting and the branch of sociology devoted to the study of organization structure which was instrumental to the development of the research design of this dissertation and to formulate the basic model utilized in this dissertation. In Chapter 3, the research design for testing this basic model will be elaborated. In the first section of the chapter, three studies from the accounting literature are reviewed.which were vital to the development of the basic thesis of this dissertation. This thesis is that the nature of the accounting system is a result of the influence of the overall organization. Golembiewski fOrmulated this thesis, but his arguments for it were defective in many respects.1 Though Simon et al. were more concerned with the accounting system than the overall or- ganization, their study suggested the idea of measuring the structural characteristics of the accounting system for the purpose of relating 1Robert T. Golembiewski, "Organization Structure and the New.Ac- countancy: One Avenue of Revolution," The Quarterly Review of’Eeonomics and Business, 111 (Summer, 1963), 29-40; Robert T. Golembiewski, ”Accoun- tancy as a Function of Organization Theory," The Accounting Review, XXXIX (April, 1964), 333-41. them to characteristics of the overall organization.1 Caplan was con- cerned with another influence of the overall organization on the ac- counting system——that of the attitudes of managers (both accountants and nonaccountants) about how organizations operate. Caplan's study provided some important concepts that were used in developing the basic model.2 In the second section of the chapter, two empirical sociological studies of samples of organizations are reviewed. These studies were essential to the development of both the research design and the basic model. Though neither study involved the accounting system, the primary contribution of the two studies was the idea that an empirical study of a sample of organizations could provide evidence of the validity of the basic thesis that the nature of the accounting system is a result of the influence of the overall organization, MOreover, the two studies fur- nished some indispensable methodological and conceptual ideas to the dis- sertation, the most important of which follow. The Pugh et al. studies concentrated on levels of variables instead of individual variables and applied data-combining techniques to organizational variables.3 The lHerbertA. Simon, George Kozmetsky, Harold Guetzkow, and Gordon Tyndall, centralization vs. Decentralization in Organizing the controller's Department (New York: Controllership Foundation, Inc., 1954), pp. 1-10. 2EdwinH. Caplan, "Behavioral.Assumptions of Management Account- ing," The Accounting Review, XLI (July, 1966), 496-509; Edwin H. Caplan, “Management Accounting and the Behavioral Sciences," Management Account- ing, L (June, 1969), 41-45; Edwin H. Caplan, Management Accounting and Behavioral Science (Reading, Mass.: Addison-wesley Publishing Company, 1971), pp. 7-46. 3D. s. Pugh, D. J. Hickson, c. R. Hinings, K. M. Macdonald, c. Turner, and T. Lupton, "A Conceptual Scheme for Organizational Analysis," 10 Blau and Schoenherr study developed the concept of structural complexity and helped refine the concept of control. These concepts were vital to the completion of the basic model.1 The basic model is developed in the third section of the chapter. It is composed of an accounting system level of variables and three levels of organizational variables. The nature of the relationships between the accounting system level and the three organizational levels is proposed in the model. Furthermore, the variables to be included in the four levels are developed and defined. Finally, three hypotheses are proposed for testing. A CCOUN TI N G LITERATURE The three studies in this section2 deal with various ways in which the overall organization relates to the accounting system. Fun- damental ideas for the development of this dissertation came from the Simon et al. study and the Golembiewski work. The Caplan work furnished some concepts which were useful in developing the basic model. Administrative Science Quarterly, VIII (December, 1963), 289-315; D. S. Pugh, D. J. Hickson, C. R. Hinings, and C. Turner, "Dimensions of Or- ganization Structure," Administrative Science Quarterly, XIII (June, 1968), 65-105; D. S. Pugh, D. J. Hickson, C. R. Hinings, and C. Turner, "The Context of Organization Structures," Administrative Science Quar- terly, XIV (March, 1969), 91-114. 1Peter M. Blau and Richard A. Schoenherr, The Structure of Dr- ganizations (New York: Basic Books, Inc., 1971). 2Though most of the authors whose research is reviewed in this section are behavioral scientists, they are included because their work was published in accounting-oriented publications and concerns the ac- counting system explicitly. The only accounting researcher is Caplan. 11 Simon et al. investigated the structure of the accounting system and tried to relate it to accounting system effectiveness. They contrib- uted to this dissertation the idea of measuring structural characteristics of accounting systems as well as several specific measures of those char- acteristics. Golembiewski explored the relationship between the overall or- ganization structure of a company and the role of its accounting system. He discussed the effect of alternative organization structures on the degree that the accounting system is a field of conflict within organi- zations. He furnished this dissertation with the idea that the nature of the accounting system may be determined by the overall organization in which it operates. Caplan examined another way that the overall organization impacts on the accounting system. He considered the effects of the attitudes of executives and accountants about how organizations operate on the meas- urements produced by the accounting system. He classified those attitudes into two models of the finn: the traditional management accounting model of the firm and the modern organization theory model of the firm. Though Caplan's work did not contribute a fundamental idea for the dissertation, it suggested some important supportive concepts which were used in the model development section of this chapter. Simon et al. In the early 1950's, The Controllership Foundation (now the Financial Executives Institute) sponsored a groundbreaking study of the organization of controllers' departments in seven relatively large 12 and complex companies. The study was undertaken by four behavioral scientistsa Simon, Kozmetsky, Guetzkow, and Tyndall. They investigated the relationship between the structure of the controllers' departments, along five dimensions of centralization and decentralization, and the controller's department effectiveness, as measured by three perform- ance measures.1 The effectiveness measures were: 1. Providing information services of high quality. 2. Performing services at minimum cost. 3. Facilitating the long-range development of competent account- ing and operating executives. The five dimensions of decentralization were: 1. The structure of accounts and reports. 2. The geographical location of accounting functions. 3. Fonmal authority relations. 4. Loyalties. 5. Channels of communication. Decentralization of the account structure has to do with whether finan- cial infOrmation is developed for subordinate units of the company. Geographical decentralization means locating controllership personnel ‘within operating locations away from the home office. Decentralization of fOrmal authority relations has to do with whether accounting units are attached to operating units (as opposed to being responsible only to the controller). Decentralization of loyalties has to do with whether accounting personnel regard themselves as part of the operating team at 1Simon et al., pp. v-ix, 1-10. 12 and complex companies. The study was undertaken by four behavioral scientists; Simon, Kozmetsky, Guetzkow, and Tyndall. They investigated the relationship between the structure of the controllers' departments, along five dimensions of centralization and decentralization, and the controller's department effectiveness, as measured by three perform- ance measures.1 The effectiveness measures were: 1. Providing information services of high quality. 2. PerfOrming services at minimum cost. 3. Facilitating the long-range development of competent account- ing and operating executives. The five dimensions of decentralization were: 1. The structure of accounts and reports. 2. The geographical location of accounting functions. 3. Formal authority relations. 4. Loyalties. 5. Channels of communication. Decentralization of the account structure has to do With whether finan- cial infbrmation is developed for subordinate units of the company. Geographical decentralization.means locating controllership personnel ‘within operating locations away from the home office. Decentralization of formal authority relations has to dO‘With whether accounting units are attached to operating units (as opposed to being responsible only to the controller). Decentralization of loyalties has to do With whether accounting personnel regard themselves as part of the operating team.at 1Simon et al., pp. v-ix, 1-10. 13 their respective locations. Decentralization of communication channels means the extent of horizontal communication between accountants and operating personnel on a given level. Some comments on the measures of decentralization are appropriate here. First, none of the dimensions is directly involved with the in- ternal organization of the controllership function, though some have implications for internal organization. Instead, all of the dimensions have to do with the relations to or attachment to operating units of controllership personnel. Second, none of the five dimensions has to do explicitly with the classic definition of decentralization-authority to make decisions at lower company levels. The only tie-in with the classic definition is that decentralization of the controllership func- tion along all these dimensions can facilitate lower-level operating decision-making. In other words, if infOrmation and support are not provided to lower-level operating personnel by the controllership de- partment, it is unlikely they will be able to make decisions. On the other hand, the fact that the infOrmation and support is provided does not assure that decentralize decisions will be made by operating employ- ees. That could be influenced by company policies and other factors. In light of the above, Simon et al. might better have labeled their variable "dispersion of accounting services and resources" rather than decentralization. In spite of the poor labeling, the Simon et al. study contributed significantly to the development of this dissertation in two ways. It suggested the idea of measuring the structural charac- teristics of accounting systems. In addition, their dimensions of decentralization suggested three accounting system variables used in 14 this dissertation. These were decentralization of accounts and reports, decentralization of geographical locations, and decentralization of for- mal authority relations.1 The research design of the Simon et al. study was relatively unsystematic. Much qualitative as well as quantitative infOrmation was collected from company officials in loosely structured interviews. The effectiveness measures are quite vague, and the authors did not describe the statistical methods they used to relate them to the centralization measures. Consequently, the authors' opinions about the companies were difficult to distinguish from their research findings. The research really boils down to an extensive case study of the seven companies. The "findings" of Simon et al. have to do with the separation of and the appropriate degree of centralization of three basic functions of controllership. These functions are: l. Recordkeeping and preparation of accounting reports. 2. Assistance to operating departments in analysis of account- ing infOrmation. 3. Participation in special studies to solve problems in oper- ating departments. The authors advocate the separation of these functions so they can eaCh be placed at the appropriate organizational level. The assistance func- tion must be completely decentralized to the operating departments. In that way, accountants can be located near operating officials in order to develop their trust by communicating regularly with them. The 1The relationships of these variables to the model for this study are discussed below on pages 54-56. 15 special studies function should be at a high level since it involves recruiting specialists from different disciplines to attack a problem on a team basis. The recordkeeping fUnction should be at some interme- diate level depending on the strength of two countervailing influences. A decentralized location near the operating departments where documents are produced promotes reliability of the system and allows access to the documents where it is needed. .A more centralized location facilitates clerical specialization and mechanization which reduce the cost of recordkeeping. Golembiewski The basic thesis of my dissertation (see above, page 1) was sug- gested in the early 1960's by Golembiewski. He maintained that the task and.prOb1ems of management accounting were determined, to a great extent, by the organization structure of the productive sector of an organization. He based this conclusion on analysis of two alternative organization structures rather than empirical research.1 He called one of these the traditional theory of organization, and diagrammed it as follows: Manager M[!|\BC SuperV1sors SA SB SC . I Product 1 Operatives a + b -+ c Product 2 Product 3 1Golembiewski, "Organization Structure and the New Accountancy," pp. 29-40; Golembiewski, "Accountancy as a Function of Organization Theory," pp. 333-41. 16 There are three production departments, each headed by a supervisor (S , SB, and SC). Three production processes are required to make each prod— uct, and the three processes are assigned, respectively, to the three production departments. Thus each production department has only one (or possibly a few) type of operative (production worker). He called the alternative organization structure the emerging theory of organization, and diagrammed it as follows: Manager MABC Supervisors 81 52 53 ////i\\\\ ////i\\\\ ////i\\\\ Operatives a + b + c a + b + c a + b + c 4 I 4 Product 1 Product 2 Product 3 All three productive processes and their respective types of production workers are in each department. The departments may produce the same or different products. Note that little integration of the activities of the depart— ments is required with the emerging theory since the departments produce products independently, while extensive integration (scheduling, etc.) is required with the traditional theory since the output of one department is input to the next. Golembiewski maintains that the accounting function must play a dominant role in forcing integration of the.activities in the traditional structure and is thus on the "firing line." In addition, there are no natural standards of perfOrmance fOr the process departments since each contributes only partially to the production of a product. 17 Thus arbitrary standards and arbitrary allocations of the costs of inte- gration (i.e., idle time in one department caused by a slowdown in an- other department) are necessary. The accounting function suffers because it must force integration with necessarily imperfect instruments. This is alleviated by the emerging theory of organization where forced integration is not necessary. In this system, the accountant, relieved of his forced integration role, can concentrate on providing helpful infOrmation to managers. Golembiewski's arguments about the superiority of the emerging theory over the traditional theory are defective in two respects. First, they cannot be proved without empirical verification. It is hoped that this dissertation may contribute in this area. Second, they are defec- tive analytically. The two organization structures are not economically viable alternatives for most organizations. Economies of scale with respect to processes determine how much they must be concentrated in a single organizational unit. For example, if process A (under supervisor A) were an automated assembly line, it would be impossible to break it up between three product departments. Admittedly the emerging theory structure poses less coordination prOblems for the accountant and for management in general. But the accountant must work with whatever or- ganization structure is mandated by the company's technology (i.e., economies of scale). Furthermore, much modern manufacturing enterprise is highly integrated and technologically sophisticated. It would be impossible to break up such enterprises into product units so that par- tial contributions to the production of the product need not be meas- ured. In a high-technology environment, a primary challenge to accountants 18 is to learn to measure such partial contributions rather than avoiding the prOblem.by insisting on a simpler organization structure and thus a simpler level of technology. Though the actual relationship between the organization struc- ture and the accounting system may be much more complex than the over- simplistic and naive model that Golembiewski proposed, his postulation of it was a key steppingstone in the development of the idea fOr this dissertation. Caplan In the late 1960's, Caplan considered, from a different perspec- tive, the effects of the organization on the nature of management ac- counting. Specifically, he considered the effect of various assumptions by management and accountants about the way organizations operate on what is measured by the management accounting system. Caplan developed two models which he maintained represent the mainstreams of management thought. These were the traditional management accounting model of the firm and the modern organization theory model of the firm. For each model, Caplan developed assumptions about organization goals, the be- havior of participants (employees), the behavior of management, and the role of management accounting. He hypothesized that adherence to one of the models by management and accountants determines what accounting meas- urements are produced for management.1 1Caplan, Management Accounting and Behavioral science, pp. 7-46. The material in this reference was extracted by Caplan from two prior jour- nal articles: Caplan, "Behavioral Assumptions of Management Accounting," 19 The traditional management accounting model of the firm is ori- ented chiefly around the exclusive organization goal of profit maximiza- tion. The roles of both management and.management accounting are to maximize profits. The chief function of management accounting is to subdivide the overall profit goal into subgoals, assign responsibility fer the subgoals to managers (departmental budgets), and hold the manag- ers responsible fOr accomplishment of the subgoals (departmental per- fOrmance reports). The modern organization theory model is oriented to decision- making at various levels of the organization and is based primarily upon the writings of Barnard and Simon. Its assumptions attempt to describe how organizations actually operate. The key aspect of the modern model is taking into account human limitations: limited rationality, limited cognitive ability, limited knowledge, and limited commitment to the or— ganization. Caplan suggested that the acceptance by accountants of the tra- ditional model has greatly restricted the ability of accounting systems to respond to the actual needs of management. He recommended empirical research to determine the relative effects of acceptance of the two models by company accountants on company accounting measurements and on company functioning. In contrast to Golembiewski, Caplan's analysis seems conceptually sound. He avoids postulating relationships which must be sUbject to pp. 496-509; and Caplan, "Management Accounting and the Behavioral Sciences," pp. 41-45. 20 empirical verification. The influence of the basic assumptions of ac- countants (and of course management) on the nature of accounting meas- urements, though subject to empirical verification, seems logical. But Caplan's models omit consideration of a vital factor: cue ganization structure. The management of organizations would be an impossible task were it not for relatively permanent structures created to channel organizational activities. Some of these structures are located within the accounting system. .A key role of management, in addition to directly controlling operations, is creating such structures which fOster effective and efficient operations, facilitate the control of operations, and thus allow management to pursue other activities such as planning. SOCIOLOGY LITERATURE The two organization structure research studies reviewed in this section heavily influenced the conceptual foundations and the re- search design of this dissertation. The idea that relationships between the accounting system and.the overall organization ought to be examined evolved from the studies dealing with the accounting system in the last section. But none of the three studies made effective use of empirical research techniques to document their conclusions. Only the Simon et al. study used a sample of companies. But the way they generalized from the data collected was not clear. Though the two empirical organization structure studies in this section did not address the subject of the accounting system, their use of empirical research techniques suggested the idea of applying those 21 techniques to a study involving the accounting system. Mere specific— ally, their use of the statistical technique multiple regression analy- sis to isolate relationships between variables was duplicated in this dissertation. ‘MOreover, the studies suggested important implications of organization structure for the accounting system. Both Pugh et al. and Blau and Schoenherr were interested in the relationships among characteristics of the context (environment and un- changeable aspects of the organization), organization structure, and functioning of organizations. Though Blau and SChoenherr were concerned with relationships among individual variables, Pugh et al. emphasized relationships among levels of variables, specifically the relationship between the context level and the organization structure and functioning level.1 The levels approach was adopted in this dissertation. Pugh et al. also contributed to this dissertation the idea of combining measure- ments into a single measurement of a variable by means of such techniques as principal components analysis. Hence, the greatest contribution of Pugh et al. was in method- ological areas. In contrast, the most significant contribution of Blau and Schoenherr was in conceptional areas. They were very concerned with building a theory that explains how organizations develop. Consequently, they carefully defined concepts and tried to explain how relationships found among those concepts came about. .A vital concept in their study was differentiation of organization parts which leads to complexity of 1The rationale for the levels approach is that variables within the levels have a common influence on other levels of variables. 22 the organization structure. They also dealt with the problems of con- trol and coordination in organizations. Each of these concepts contrib- uted to establishing a role for the accounting system and thereby an explanation of the processes which may cause accounting systems to develop. These matters are discussed in the model development section of this chapter. Pugh et al. Much of the research design of this dissertation was suggested by the study by D. S. Pugh and his colleagues of forty-six organizations in the English Midlands during the late 1960's.1 Rather than emphasizing the relationships among individual variables, they explored the relation- ships among different levels of variables in organizations, specifically the levels of contextual variables and organizational structure and func- tioning variables. The contextual level includes relatively unchangeable factors such as the history of the organization, the nature of its owner- ship and control, its size, its mission, its technology, its geographical dispersion, and its dependence on other organizations. The organiza- tional structure and functioning level includes such factors as special- ization, standardization, formalization, centralization, and configuration. Pugh et al. hypothesized that the contextual variables were prime determi- nants of the organizational structure and functioning variables and 1Pugh et al., "The Context of Organization Structure," pp. 91- 114; Pugh et al., "Dimensions of Organization Structure," pp. 65-105. 23 investigated that relationship.’ They planned to investigate, in turn, whether these two levels of variables may be prime determinants of two other levels: group composition and interaction, and individual person- ality and behavior.2 Thus the Pugh et al. model might be diagrammed as follows: Contextual variables 1 Organizational Structure and Functioning Variables 1 Group Composition and Interaction variables 1 Individual Personality and Behavior variables Pugh and his colleagues hypothesize that the contextual variables influ- ence the development of organizational structure and functioning. In turn, organizational structure and functioning variables influence the development of group composition and interaction variables, and these, in turn, influence the development of individual personality and be- havior. Pugh and his colleagues recognize that the hypothesized direc- tion of causation (as indicated by the arrows) may be reversed in some 1Pugh et al., "The Context of Organization Structure," pp. 91-114. 2Pugh et al., "A Conceptual Scheme fOr Organizational AnaIYSiS," pp. 289-315. 24 cases.1 For example, specializing of roles (a structural variable) may require more people (a contextual variable). Pugh and his colleagues also deal extensively with the problem of aggregation of measurements. They make a clear distinction between measurements and concepts and use data-reduction techniques (principal components and item analysis) to merge different measurements felt to be associated with a concept into a single measurement of the concept.2 The principal components technique of merging measurements was incorporated into this dissertation and is discussed below on pages 72-88. Pugh et al. recognize the cost of such aggregation in lost individual relation- ships but feel that the greater conceptual clarity outweighs the cost.3 Using the data-reduction techniques indicated in the prior para- graph, Pugh et al. consolidated their multitudinous contextual measures to eight dimensions, and their organization structure and functioning measures to three dimensions. Their dimensions of context were: age of the organization, size of the organization, size of the parent organiza- tion, operating variability, operating diversity, workflow integration, number of operating sites, and dependence.“ Their organization structure 1PUgh et al., "The Context of Organization Structure," p. 112. 2Pugh et al., "Dimensions of Organization Structure," p. 70; Pugh et al., "The Context of Organization Structures," p. 93. 312nm, p. 93. l’Size of the parent organization, for an agency of government, is the size of the government of which it is a part. For an independent business organization, it is the same as organization size. For a sub- sidiary, it is the size of the parent company. Operating variability is the extent to which the organization does not produce a standardized good or service. Operating diversity is the number of different outputs 25 and functioning dimensions were: structuring of activities, concentra- tion of authority, and line control of workflow.1 Pugh et al. used stepwise regression to explain the dimensions of organization structure and fUnctioning with the dimensions of context. They found that a large proportion of the variance of each of the three organization structure and functioning dimensions could be accounted for by respectively one or two contextual dimensions and that further con- textual dimensions added nothing significant to the explanation.2 From the Pugh et al. study came the idea of using stepwise regression in this dissertation. The stepwise technique, as applied to the dissertation, is discussed below on pages 90-100. The most important conclusion of Pugh et al. is that context is a key determinant of organization structure and functioning. Their specific findings were that organization size and workflow integration were key determinants of structuring of activities; dependence and (i.e., products) produced by the organization. WOrkflow integration is the rigidity and integration of production operations. Dependency is the degree to which the organization is constrained by other organiza- tions in its environment such as labor unions, suppliers, customers, parent companies, governments, etc. Three of these dimensions were incorporated into this dissertation; these are size of the organization, operating diversity, and number of operating sites. They are discussed below, respectively, on pages 49, 52, and 48. See Pugh et al., "The Context of Organization Structures," pp. 94-109. 1Structuring of activities is a combination of measures of the extent the organization is bureaucratic; i.e., has standardized proce- dures, specialized roles, etc. Concentration of authority is roughly centralization of authority. Line control of workflow is the extent the organization is not dominated by a staff superstructure. See Pugh et al., "Dimensions of Organization Structure," pp. 85-88. 2Pugh et al., "The Context of Organization Structure," pp. 109- 11. 26 number of operating sites were key determinants of concentration of au- thority; and operating variability was the key determinant of line con- trol ofworkflow.1 Blau and Schoenherr A study by Blau and Schoenherr of fifty-three state employment security agencies, published in 1971, exerted a predominant influence on the research design of this dissertation.2 Their extremely well documented study influenced significantly the conceptual foundations, the statistical research design, and the measurement techniques of this dissertation. The employment security agencies administer unemployment insur- ance programs and provide employment services to the public. Blau and Schoenherr collected a vast amount of data on each of the fifty-three state employment security agencies (fifty states, the District of Colum- bia, and two territorial possessions) by means of interviews with agency officials and documents supplied by the agencies. Mest of the infOrma- tion involved structural characteristics of the agencies, and.much of it was obtained from agency organization charts. The authors examined the interrelations among these agency characteristics to determine if sig- nificant patterns were observable. The conceptual foundations of this dissertation are rooted in concepts developed in the Blau and Schoenherr study. Structure was ‘Ibad. 2Blau and Schoenherr, op. cit. 27 defined in their study, per the dictionary, as something composed of parts.1 I would add two clarifying aspects to the definition. Struc- ture is composed of "interrelated" parts.2 The second clarifying aspect is not in the dictionary definition but, to my mind, is an es- sential aspect of organization structure: Structure is something com— posed of interrelated parts which persist with relative permanence; i.e., they do not have to be re-established constantly. It follows that a key aspect of organization structure is the subdivision of the overall objectives of the organization into roles or areas of responsibility which individuals, departments, or levels of the organization can accomplish. Often these roles and areas of responsi- bility are different from one another. The general process of develop- ing new parts of the organization which are often different from the old ones is referred to by Blau and Schoenherr as "differentiation." As organizations differentiate more and more parts, they become more complex. Differentiation can occur along several dimensions, but there are two basic types: Ihorizontal and vertical. ‘vertical differentiation is the number of authority levels in the organization structure (between the chief executive officer and the lowest-level employees). Horizontal differentiation includes several other variables involving number of or- ganization parts: number of different jobs, average number of departments ‘Ibid. , p. 300. 2Webster's Third New International Dictionary of the English Language, Unabridged (Chicago: Encyclopaedia Britannica, Inc., 1971), p. 2267. 28 under a given level of managers, average span of control of managers on a given level, and number of local offices. The overall conclusion of the Blau and Schoenherr study is that organizations differentiate more and more parts, along all dimensions, as they become larger in size. Large organizations have more levels, more departments, more jabs, etc. However, the rate they differentiate new parts with increases in size levels off. Large organizations differ- entiate less new parts fOr a given increase in size than smaller organi- zations. A key point made by Blau and Schoenherr and incorporated in this dissertation is that organization structure can be studied apart from the behavior of humans within the structure. It is Obvious that humans (man- agers) create the structure and it is also obvious that humans are af- fected by the structure. Nevertheless, Blau and Schoenherr maintain that, given the scope of the organization's responsibilities, structural conditions exert constraints on the decisions of managers such that they tend to create structures with certain regularities and thus assessing the behavioral inclinations of managers is not necessary.1 For example, the Appollo spacecraft could not have been constructed in someone's backyard, regardless of the amount of money invested in it. An organi- zation structure (governmental and industrial) of a minimum degree of complexity was necessary to marshall the resources to get the job done.2 1Blau and Schoenherr, p. 300. 2This is my own illustration. 29 Blau and Schoenherr also emphasize the importance of control and coordination of activities in complex organizations. They interpret many of their research findings in terms of control and coordination. For example, they explain why the rate of differentiation of organization parts with increases in size levels off by citing the additional coordi- nation that is required in large organizations. For example, the number of departments in an organization increases with increases in company size. However, when the organization becomes large, coordination of the activities of the numerous departments becomes difficult. Consequently, pressure builds up to resist further differentiation. This pressure slows down the rate of differentiation of departments. Furthermore, Blau and Schoenherr recognize that operations are controlled by "impersonal mechanisms of control." Personal control or supervision is the oldest form of control in organizations. MOst other control mechanisms are impersonal to some degree. Impersonal mechanisms of control include automation and standardization of procedures, both of which control operations without human (or personal) intervention. For example, automation in the form of an assembly line controls the actions of workers by forcing them to adhere to the pace of the assembly line.1 The statistical research design of this dissertation was heavily influenced by the Blau and Schoenherr study. Theirs was a cross-sectional study of the interrelations of organization characteristics at a single point in time. Yet their explanations of the relationships were devel- opmental. They were not satisfied with saying that one characteristic lBlau and Schoenherr, pp. 300-26. 30 happened to be associated with another. They suggested why the associa- tions might have come about. The cross-sectional research design is not adequate to verify such explanations. Yet proposing the explanations is necessary fOr theory development.1 Like Pugh et al. and this disserta- tion, Blau and Schoenherr used multiple regression analysis to determine the effects of variables on each other.2 The measurement techniques of this dissertation were based to a large extent on those used by Blau and Schoenherr. They described the specific measures they used for variables in appendices to their book.’ These measures were used extensively in developing the interview ques- tionnaire fOr this study. The specific uses of their measures will be noted later. DEVELOPMENT OF BASIC MODEL In order to direct the attempt to discern relationships between the structure and the accounting systems of organizations, a model is developed in this section. The model focuses attention on the key over- all relationships and provides a means of classifying variables of the accounting system and those of the overall organization which influence the accounting system. This model incorporates many of the concepts of the studies that were discussed in the prior section. The first step in developing the model is to set out the desir- able features that it should have. Second, the broad concepts of the 1179id., pp. 326-29. 213m, pp. 23-27. 3mm, pp. 373-407, 422-35. 31 model which are represented by levels of variables are defined. At the same time, the conjectured relationships between the levels are elaborated. Next, the variables which are to be included in each level are designated and defined. Finally the hypotheses which incorporate the assumed relationships among the levels of the basic model are fOrmulated. Desirable Features of'the Model The model must have certain characteristics in order to be use— ful. The major features that are strived for in the model development are simplicity, causality, and intuitive meaning fOr the major parts of the model (levels). First of all, the model must be simple. Why should a model that is designed to explain highly complex relationships among multitudinous variables be simple? Because, otherwise, the human mind is unable to deal with it. Simplicity requires that relationships among only a few (no more than five) basic elements be examined at any one time. Conse- quently it is necessary to classify the numerous variables that must be considered in a study involving the structure of organizations into a few basic levels.1 The cost of this simplification is that attention is diverted from.many interesting and important individual relationships be- tween variables. The advantage is the conceptual clarity gained. In 1The idea for the levels approach came from the previously cited works of Pugh et al. See above, pages 22-23. Their distinction between the contextual level of variables and the organizational structure and functioning level was incorporated in the basic model of this disserta- tion. See the discussion of context below on pages 37-41. 32 fact, the "art" of theory building in any field is the simplification of highly complex relationships. .A second characteristic necessary to make the model useful (and a basis fOr distinctions among levels) is that it be causal. The levels should be related logically to each other. It seems obvious that such logical relationships among the levels would appear over time, and so a time dimension is a necessary element of the explanation. For in- stance, both Blau and Schoenherr and Pugh et al. have concluded that contextual level variables "cause" the development of various character- istics of organization structure. Blau and Schoenherr discussed the issue of whether theories should explain why associations develop logically or merely predict that in certain circunstances they do appear. They concluded, as I have, that the usefulness of a theory is greatly enhanced by an explanation of the logical connections that explain the relationships that develop in or- ganizations.1 Consequently, causal relationships are described.between the levels in this dissertation. The status of variables in the higher level necessitates that variables in the lower level take on a certain pattern. A third characteristic necessary for the model to be useful is that the levels have intuitive meaning. If levels include seemingly random collections of variables, it is difficult either to establish the meaning of the level or to explain the logical relationships among the levels. Consequently, the variables in each level should have 1Blau and SChoenherr, pp. 328-29. 33 some common characteristics which determine the meaning of the level as a whole. The Levels and Their Interrelations In this section, the four levels of the basic model and the theorized relationships among those levels are introduced. The four levels are the accounting system, the control system, structural com- plexity, and context. In the first subsection, the expected positive relationship between structural complexity and the stage of development of the accounting system is developed. In addition, the expected nega- tive relationship between the stage of development of the accounting system and that of the control system is advanced. In the second sub- section, the expected positive relationship between the stage of devel- opment of the accounting system and that of context is elaborated. Structural Cbmplexity and COntrol system The Blau and Schoenherr study suggested one common characteris- tic that can define a set of variables in organization structure analy— sis: structural complexity. As was discussed above on page 27, struc- tural complexity is the degree the organization is divided into (usually different) parts along various dimensions. Not only does structural complexity have conceptual clarity, but it suggests some important logical relationships to the accounting system that should be tested. Since accounting has often been considered to be focused on control and 34 1 the question naturally arises whether an coordination in companies, organization with greater structural complexity needs a more developed accounting system to force goal congruence.2 But is accounting the only control and coordination system in organizations? In small organizations there may be no accounting or any other observable control system. Yet all organizations must control their operations. Blau and Schoenherr refer to supervision as the tra- ditional fOrm of control in work organizations with the threat of punish- ment or termination of employment as the support for it. They point out that modern organizations do not rely as much on supervision. Instead, they use less offensive control systems, such as standardization of procedures and qualified personnel. Standardization of regulations and procedures limits the discretion of employees. The technical knowledge and self-discipline of certain employees is a control mechanism in that suCh employees need less supervision, procedures to guide them, and other fOrms of control.3 This variable will be called "personnel qual- ity" in this dissertation. Another form of control is the retention in the hands of top management of the authority to make key decisions. If lower-level 1Caplan's traditional management accounting model of the firm clearly views the role of management accounting as primarily a control device. Its chief functions are to assign responsibility fOr perform- ance to managers by means of budgets and hold them responsible for ac- complishment by means of performance reports. Caplan implies that the traditional model is widely accepted among accountants. See above, page 19 for a description of Caplan’s traditional management accounting model. 2This question was initially raised by Golembiewski; see above, page 15. , 3Blau and Schoenherr, pp. 348-50. IIIIIIIIIIIII-—______________________________________ c________n_ 35 managers do not have such authority, they do not have the power to dis- rupt company operations. This retention of authority is referred to in this dissertation as centralization of authority (as opposed to decen- tralization of authority). The development of each of these control systems, like the development of the accounting system, may respond to increasing structural complexity. These relationships can be diagrammed as follows: Structural Complexity ///1\.\. . Standard- Central- Accounting Su ervision ization of Personnel ization of System p Quallty . Procedures Authority If control systems are related positively to structural complex- ity, how are they related to each other? A supposition of this disser- tation is that they are negatively related to each other. The logic behind this supposition is presented in the following paragraph. Blau and Schoenherr asserted that automation (or mechanization) acted as a control system. It took the fOrm of the extent of computer usage in employment security agencies. Blau and Schoenherr felt that computers exert a constraining influence on employees since their per- fOrmance must conform to the computer setup.1 But of more importance for this dissertation, they found an interesting interaction effect lIbid. , p. 126. 36 between computerization and centralization of authority. Increasing organization size created pressures to decentralize operations in order to decrease the decision-making load of top management. But such de- centralization tended to take place only in the presence of computeriza— tion. Apparently computerization restrained operations to the extent that top management felt more comfortable about decentralizing decision- making.1 This finding suggested the possibility of similar interaction effects between other control systems. Is it possible that, in general, as one control system develops (i.e., automation), another (i.e., cen— tralization) need not develop as far? If so, negative associations between the stages of development of the control systems can be ex- pected. The two basic relationships discussed so far can be diagrammed as follows: I Structural Complexity .e—-————-—"i:::::”” i T“\‘::::_‘*~-—————____.~ Stan-' Account- dard- Per- Central- ' Super- - - 1zat10n Ing -neg» ViSion ncgs 1zat10n neg sonnel -.neg.. of Au- System of Pro— Quality . thor1ty cedures neg neg neg neg neg neg 1Ibid., pp. 321-22. 37 The above diagram presents the core of the model used in this dissertation. Though structural complexity will be broken into differ- ent dimensions and other control systems will be added, it incorporates the primary relationships that are examined: the positive relationship between structural complexity and the accounting system, and the negative relationships among the control systems (particularly the negative rela- tionships between the accounting system and other control systems). All the variables taken together constitute organization structure, the per- sistent aspects of the organization.1 context Though the model proposed so far includes the key relationships to be examined in this dissertation, it is advisable to expand it some- what fOr two reasons. First, it is necessary to control for some factors which may affect the relationships between overall structural complexity and the control system. Second, an alternative explanation of the devel- opment of the accounting system is possible. 1Neither Pugh et al. nor Blau and SChoenherr subdivide organiza- tion structure between structural complexity and control systems as this dissertation has. They were not compelled to make such a subdivision because their primary Objective was to explain organization structure in terms of context. If context induces the development of structural com- plexity and structural complexity induces the development of control systems, then positive relationships will be fOund between context and both structural complexity and control systems. In contrast to Pugh et al. and Blau and Schoenherr, the primary Objective of this dissertation is to explain a single control system——the accounting system——in terms of organization structure. The subdivision of organization structure is necessary because negative relationships are expected between the accounting system and other control systems, while positive relation- ships are expected between the accounting system and structural com- plexity. See the discussion of the definition of structure above, page 27. 38 Though the context of the organization has been left out of the model so far, it is important to include it for several reasons. Many organization researchers, including Pugh et al. and Blau and Schoenherr, have examined (and documented) the ways that context, the relatively unchangeable aspects of the organization, induces the development of organization structure.1 To ignore that relationship would leave open the possibility that any relationships found within the organization structure are accountable solely by the common influence of context. For example, context may influence the development of the control system and structural complexity, as follows: Context Structural r;?::§:::fi: ¥_ Control Complexity found I) '" System 1Both Pugh et al. and Blau and Schoenherr distinguish variables that are given (context) from those which can be altered. Given vari- ables cannot be changed by the current management of the organization. Blau and Schoenherr split the given variables between environmental and parameter variables (characteristics of the organization itself which cannot be changed). On the other hand, Pugh et al. create one group called contextual variables, which includes both environmental and parameter variables. Pugh et al. include in.the contextual group suCh variables as size of the organization, its mission, and its technology (nature of its productive operations). Blau and Schoenherr include suCh variables as size of the organization and extent of civil service regu- lations governing the organization (employment security agency) in the parameter classification and Characteristics of the local population in the environmental classification. See above, pages 22 and 24. 39 If context were not included in the study, structural complexity would be fOund to be positively related to the control system due to the com- mon influence on the two by context. That relationship would be erro- neous, since there is no causal connection between the two. Consequently, the context of the organization is included in the model as a control on this possible relationship. A second reason for the inclusion of context in the model is that context, rather than structural complexity, may actually be the primary determinant of the development of the accounting system. The logic behind this alternative explanation is the additional decision- making infOrmation that may be required for companies with involved contexts . .A significant body of accounting thought in recent years has assumed that the primary function of both management and financial ac- counting is the provision of infOrmation for decision-making. For ex- ample, Charles T. Horngren, in his widely used and respected cost accounting textbook, asserts that providing infOrmation for managerial decision-making is the basic reason for management accounting.1 Caplan adopts a decision-making orientation in his modern organization theory model and contrasts this with the primarily control orientation of the traditional accounting model.2 If providing information for decision- making were actually the primary function of management accounting in 1Charles T. Horngren, Cost Accounting, A Managerial Emphasis (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1972), pp. 2-3. 2Adiscussion of Caplan's alternative model appears above, page 19. 40 organizations, then it might be expected that the stage of development of the accounting system would be closely related to the demands placed on the organization by its context. It seems logical that companies facing a changing and intricate environment would need more and.better infOrmation for decision-making. This relationship could be diagrammed as fellows: Context 1 Need for InfOrmation for Decisions 1 Accounting System As was discussed on page 38, the relationship between context and organization structure has been extensively examined by organization researchers. .A direct, prObably causal, relationship between context and organization structure has been verified by these researchers.1 In a general sense, it is evident that a changing and intricate environ- ment as well as a sophisticated teChnology and other aspects of context necessitate the development of a complex organization structure. This general relationship is assumed to be true in this dissertation and is not tested. Context is added to the model on page 36 as follows: 1The specific variables of context and organization structure that are found to be related are not always the same. 41 Context —————— assumed A Structural Complexity V I >- Control System To this point, the control system as a whole has been presented as the bottom level of the model. The model suggested the desirability of examining the relationship of the control system to two higher levels of variables in the model: structural complexity and context. But since the concern of this dissertation is the accounting system rather than the control system as a whole, it is desirable to elaborate the model somewhat further. The control system can be divided into two parts: the accounting system and other control systems. Furthermore, the relationships, indicated by arrows, from structural complexity and context are redirected to the accounting system. This elaboration is diagrammed as follows: Context assumed v Structural __fi Complexity Other Control Systems . . . Control examined V examined * 1 examined ——— System Accounting System 42 The accounting system is now on the bottom level and thus sub- ject to explanation by three higher levels of variables: cnher control systems, structural complexity, and context. But the relationship of the accounting system to other control systems is quite different than the relationships to the other two levels. The accounting system is to be explained by other control systems because it is the central focus of the research, not because the other control systems cause the accounting system to be developed. Hence the relationship to other control systems is reciprocal rather than causal. In summary, the model includes three levels of variables which are possible determinants of the state of development of the accounting system. This dissertation will examine these relationships in a sample of companies. The relative strength and direction (positive or negative) of the relationships to the three levels will suggest implications as to the nature of the processes1 which cause the accounting system to be de- veloped. For instance, if the complexity of the organization structure is found to have a strong positive relationship with the stage of devel- opment of the accounting system, when the effects of context and other control systems are held constant, then support is provided for the thesis that management accounting is primarily control-oriented and must develop in response to the differentiation of the organization into parts along various dimensions. If, in addition, the stage of develop- ment of the accounting system is fOund to be negatively related to the 1"Process" is used here in the general sense, not as the organi- zational level. 43 stage of development of the other control systems, when the effect of context and structural complexity are held constant, then support is provided for the thesis that control systems are at least partially sub- stitutes in contributing to the resolution of control and coordination problems created by structural complexity. If, on the other hand, the stage of development of the accounting system is found to be more strongly related to context than to structural complexity, when the other levels are held constant, then support is provided for an alternative thesis that the accounting system is primarily decision-oriented rather than control-oriented and that the quantity and quality of infOrmation needed by management for decision-making is dependent on the intricacy of the context facing the organization. variables within the Levels Though some of the variables of the control system level were developed in the previous section in order to describe the relationship between the structural complexity level and the control system level, the purpose of this section is to fill out the levels with variables which are to be their respective elements. In order that the reader may have a guide to follow as various variables are discussed, the basic model used in this study is presented at this point with variables in- cluded. Figure 1 illustrates the basic model. The three major groupings of variables-—process (a condensation of context discussed on pages 52-53), overall structural complexity, and control system——are in the three outer boxes. Within each box are variables or subgroups of variables of 44 Process Sophistication Output Diversity Materials Input Diversity I Overall Structural Complexity Job . Divisional Divisional . . * Size Structure Ggggrzggigfil Differen- Special- 632466313111 2331:1218 rs Complexity p tiation ization mp I Control System Centraliza- . Direct Staff Au- Personne*l tion .of * Standardiza- . Quality Authority tlon of Super- Support thor1ty H' I. I Pu P d * vision of Line Levels lgh' ow- n- r- roce ures Level Level vest- chas- Jobs General ment ing Accounting System Structural Complexity Size* Job Geo- Struc- graph- unit Differ- Au- InfOr- Re- . . . . ture lcal entlatlon thor- matlon source C _ D' _ V _ H ._ ’t Output Input om 15 er orl 1 y plex- per- tlcal zontal Levels ity sion System . Report Co lexit Sophls- Personnel Differ- Degntralf tication Mechan- Quality" entia- ization of of Tech- ization Edu- Gen- tlon Accounts nlques cation eral Figure l. The Basic Medel *These six variables were respectively split into the two vari- ables indicated in the data refinement stage (below, pages 81-88). In zgis Chapter on model development, they are discussed as a single vari- le. 45 the respective groupings. For example, "sophistication," "output diver- sity," and "materials input diversity" are variables of process. "Direct supervision" and "staff support of line" are variables of the control system. The accounting system, a subgroup of variables within the con- trol system, is itself broken down into four variables and two subgroups of variables. The two subgroups of variables within the accounting sys- tem are labeled "structural complexity" and "system complexity." There are four levels of variables in the basic model. Since the focus of interest of the study is on characteristics of the account- ing system, the variables included in that box are considered criterion (dependent) variables and, taken together, constitute the first level of variables in the basic model. The rest of the levels in the model are composed of explainer (independent) variables which are frequently re- ferred to in this dissertation as organizational variables. The rest of the variables in the control system box, taken together, constitute the second level of variables. The overall structural complexity and process variables respectively constitute the third and fourth level of variables. In the first subsection, variables are assigned to the struc- tural complexity and control system levels. In the second subsection, the technology of the production process is set forth as the only aspect of context that it is necessary to measure for the sample of companies examined in this dissertation. Consequently, the context level is re- named "process" and three variables of process are elaborated. In the third subsection, the variables of the accounting system are developed. 46 Structural Cbmplexity and Control System Variables In this section, variables are assigned to the structural com- plexity and control system levels. The first step is to incorporate into the control system level fOur control systems that were developed to demonstrate the nature of the control system level (above, page 34). Then the variables of differentiation are allocated between the structural complexity and control system levels. Finally, "size" and "mechanization" are ascribed to the structural complexity level, and "staff support of line" to the control system level. The four control systems (other than the accounting system) orig- iinally discussed above on page 34 are recapitulated and redefined formally fOr this study here. "Direct supervision" is the extent to which activi— ties of employees are controlled through the issuance of directives by a supervisor. "Standardization of procedures" is the proportion of organ- izational situations for which there are explicit rules and regulations to govern employee actions. "Personnel quality" is the extent of tech- nical knowledge and self-discipline among employees which alleviate the necessity of having other control systems. "Centralization of authority" is the prevention of deviation from top management plans by restricting the authority to make key decisions to high-level executives. Each of these control systems is incorporated in the control system level of Figure 1. Like the concept of structural complexity itself, many of the variables which are elements of the structural complexity level were derived from Blau and Schoenherr. As was discussed above on page 27, 47 differentiation, the developing of new parts of the organization, is the process which creates structural complexity. Blau and Schoenherr divide differentiation into two types: horizontal and vertical. ‘Vertical dif- ferentiation is the number of authority levels in the organization struc- ture (between the chief executive officer and the lowest-level employees). Horizonal differentiation includes all the other variables involving num- ber of organization parts. These include jOb structure complexity, geo- graphical dispersion, and divisional differentiation. All of these variables of differentiation were incorporated in the model in some way. "Authority levels" was included as a control system, while the other differentiation variables were included as structural complexity variables. The reasoning for not including "au- thority levels" in structural complexity follows. Blau and Schoenherr fOund that the number of authority levels in an agency is inversely related to the number of divisions, a measure of horizontal differenti- ation. They speculated that increasing agency size leads initially to horizontal differentiation, specifically more divisions. But the in- creasing number of divisions overburdens top management with adminis- trative work (controlling the divisions). This forces top management to create a new level of superdivisions between itself and the old divisions. This decreases top management's administrative load but increases the number of authority levels.1 The implication is that vertical differentiation (authority levels) is a pressure-relief mechan- ism rather than a control-problem-producing aspect of differentiation. 1Blau and Schoenherr, p. 321. 48 vertical differentiation (authority levels) has to do With the administrative superstructure above the operating level. It is devel- oped to manage the operating level and enable it to perform its func- tion. For example, a greater number of authority levels permits more first-level supervisors to be in the organization structure and thus more operatives to be controlled without increasing the span of con- trol of the first-level supervisors. Rather than contributing to structural complexity and its consequent problems, it alleviates those problems. Thus it is classified as a control system in the basic model. The horizontal differentiation variables were assigned to the structural complexity level because they were perceived to increase control and coordination problems. "JOb structure complexity" is de- fined for this dissertation as the subdivision of the process into tasks which individual employees can accomplish. It is roughly the number of different jobs or types of work in the organization.1 "Geographical dis- persion" is the degree to which the organization is divided into separate locational units and the degree these units are spread out.2 "Divisional differentiation" is the number of company divisions. 1This variable was suggested by Blau and Schoenherr's variable, "number of different jobs." .A secondary aspect that was incorporated into the variable was the distribution of employees among jObs. This has to do with whether most employees are in a few jObs or the employees are uniformly dispersed among the jobs. The idea for this came from James L. Price, Handbook of Organizational Measurement (Lexington, Mass.: D. C. Heath and.Company, 1972), pp. 70-71. 2Itwas suggested by the Blau and Schoenherr variable, "number of local offices" (see above, page 28), and the Pugh et al. variable, "number of operating sites" (see above, page 24). 49 In addition to divisional differentiation, divisional specializa- tion was included in the structural complexity level. "Divisional spe- cialization" is the degree the responsibilities of divisions are different from one another. For example, a company with three divisions-—manufac- turing, sales, and administration——is highly specialized since the re- sponsibilities of the divisions are completely different. Another company with three divisions——West Coast, East Coast, and South——is very unspecialized since the three divisions have the same responsibilities. In general, differences among the parts of an organization can be as- sumed to increase the complexity of the organization structure and thereby control and coordination problems. In other words, given the number of divisions is the same, differing division responsibilities make the job of coordinating their activities more difficult. The definition of "organization size" in this dissertation is the quantity of resources at the disposal of the organization. The most important resource for most organizations is employees, and thus the number of employees is considered the major aspect of organization size. Even though organization size is not a dimension of structural complexity as has been defined (differentiation of parts), "size" is as- signed to the structural complexity level in this dissertation.1 Two criteria have been used for assigning variables to the structural com- plexity level. First, the variable must develop in response to some 1The operational definition of "structural complexity" has thus been broadened. 50 aspect(s) of context. Second, the variable must icnrease control and coordination problems of the company. "Size" qualifies on both criteria. It is Obvious that quantities of resources, especially employees, enable an organization to cope with its context. For example, manufacturing companies must hire employees to accomplish their primary objective which is to manufacture product. Furthermore, it is clear that large quantities of resources create control and coordination problems for organizations.1 "Mechanization" is defined for this dissertation as the utiliza- tion of sophisticated.machinery and equipment, and is assigned to the structural complexity level. It should be noted that the "mechanization" variable in this dissertation is much broader than that of Blau and Schoen- herr, since it includes "automation of production" (assembly lines, etc.) as well as "data automation" (computerization). On page 36 above, it was indicated that Blau and Schoenherr considered automation in general, and specifically data automation (computerization), to be control devices.2 1Both Pugh et al. and Blau and Schoenherr assigned "organization size" to the context level. The rationale was that organization size is mandated by the scope of the organization's responsibilities, and not controlled by current management. For example, employment security agen- cies must have a certain number of employees to accomplish the responsi- bilities laid on them by state and federal laws. In contrast, the busi- ness organizations in the sample selected fOr this dissertation are free to expand or reduce the scope of their operations and.make tradeoffs between human and nonhuman (machinery and automation) resources which management perceives will maximize the accomplishment of company goals. Thus a key difference between this dissertation and the studies of Blau and Schoenherr and Pugh et al. is that "organization size" is a variable that is under managerial control. 2In fact, the basic idea of negative relationships among control systems came from Blau and Schoenherr's discussion of the interaction between centralization of authority and computerization. 51 Though the control impact of computers is apparent in certain circum- stances, the traditionally accepted role of computers is an aid to decision-making rather than a control device. For example, computers can efficiently produce detailed sales infOrmation which is vital for decisions on pricing and development of new product lines. Thus the computer enables the company to respond effectively to a volatile mar- keting environment which is one aspect of context. Similarly, automation of the production line enables a company to efficiently produce goods fOr the market. In general, mechanization, like company size, enables an organization to satisfy requirements of its context. Furthermore, companies can trade off other aspects of structural complexity for mech- anization. For example, computerization can reduce the numbers of em- ployees involved in clerical activities, thereby decreasing company size. In addition, the organizational arrangements necessary in a computerized operation (data processing department, systems analysis staff, etc.) can create control and coordination prOblems themselves. In view of the above, the "mechanization" variable is assigned to the structural com- plexity level. "Staff support of line," as measured by proportion of staff personnel, was assigned to the control system level.1 Supervisors in modern organizations face exceedingly complex control and other respon- sibilities. Support personnel may be necessary to accumulate informe: tion and perfOrm routine functions for supervisors. Without such 1Blau and Schoenherr used "staff support" in their study, but it had little impact on the theory they attempted to develop. See Blau and Schoenherr, pp. 86-87. 52 support, supervisors may do a poor jOb of controlling and other control systems may have to take up the slack. It is expected that, to the ex- tent the staff component is developed, control systems (particularly the accounting system) need not be as highly developed. context variables .AS‘was discussed above on page 38, context includes character- istics of the environment (local community, etc.) which affect the or- ganization as well as committed characteristics of the organization itself. Contextual characteristics are not under the control of current management. Since the key interest of this writer was in the relation of organization structure to the accounting system, a conscious effort was made to mini- mize the importance of context, though some characteristics of context did have to be included. The importance of context was minimized pri- marily by selecting a sample of organizations with similar contexts; i.e., small manufacturing companies in small Nfichigan communities (see below, page 61, for a description of the sample characteristics). It was felt that the context of these companies differed chiefly in the technology of their productive operations. Due to the control by sample selection, no other aspects of context are measured in this study. The contextual level is consequently renamed the "process level" to indicate that vari- ables of the technology of the productive process are included. Even the characteristics of process can be extremely complex. This study has taken a very rudimentary approadh to the measurement of process. One of the variables was derived from Pugh et al.: "diver- sity of output," or the number of different products or services produced 53 by the organization.1 Other factors being equal, the greater the number of outputs (products), the greater is the difficulty of producing them (the process). Pugh et al.'s "diversity of outputs" suggested the idea of "diversity of inputs" (different types of raw material or components purChased) as a variable of process. It seems logical that, other fac- tors being equal, the greater the number of materials inputs, the more highly developed must be the process in order to assemble them. The third variable, "sophistication of the process," is just a convenient label fOr a variety of measures suggested by various people who read the dissertation proposal. These measures included such characteristics of the companies as length of the operating cycle, proportion of expenses which are for research and product development, and average value added (sales price minus materials costs). Accounting System Variables In this section, variables of the stage of development of the accounting system are elaborated. Many of these variables are mirrors of variables for the organization as awhole.2 In the first part of the section, variables of the structural complexity of the accounting system are developed. Then size of the accounting system is defined. Next, a set of nonorganizational variables, having to do with the nature of the lPugh et al., "The Context of Organization Structures," p. 103. Also see above, page 24. 2The Simon et al. study suggested to me the idea of measuring the structural characteristics of the accounting system, but more of the specific accounting system variables were influenced by the Blau and Schoenherr study than by any other source. See above, pages 11-15 and 26-30. 54 output (reports, etc.) of the accounting system, are developed. Finally, three further aspects of the accounting organization are described. These are personnel quality, mechanization, and centralization of au- thority. The entire set of accounting system structural complexity vari- ables is mirrored after variables of the same name for differentiation of various types. "Accounting system structural complexity" is thus defined as the degree the accounting system is divided into parts on various dimensions. "Job structure complexity" is the extent of sUbdi- vision of the accounting system function into employee tasks (see above, page 48). "Authority levels" is the difference between the authority level of the top-level controllership executive and the authority level 1 "Geographical dispersion" of the lowest-level controllership employees. is the number of different sites at which controllership employees are located.2 The variable "size of the accounting system" was suggested by the previously discussed variable "size of the organization as a whole." "Size of the accounting system" is defined as the quantity of organiza- tion resources (primarily personnel) allocated to accounting activities. 1Though the variable "authority levels" was distinguished from the other fOrms of differentiation as a control system fOr the overall organization (see above, page 47), no such distinction is made for the accounting system. Whether accounting system authority levels help to create or alleviate control problems within the accounting system is not important fOr this study. The question is whether they change in response to other levels of variables. 2Simon et al. used an almost identical variable which they called "decentralization of geographical locations" (see above, page 12). 55 Since accounting resources are allocated out of total organization re- sources, the variable should perhaps be called "proportion of organi- zation resources allocated to accounting activities," and some of its measures reflect this proportional interpretation. "Decentralization of accounts" is defined as the extent to which accounting reports are prepared for and presented to lower-level depart- ments.1 Stated another way, it is the depth in the organization struc- ture for which departmental accounting reports are developed and thus the extent the data are collected in detail for lower-level departments. "Decentralization of accounts" suggested the idea of measuring other nonorganizational aspects of the development of the accounting system. These characteristics, along with "decentralization of ac- counts," were subsumed under the label "system complexity," to distin- guish them from structural complexity. Both were suggested.by account- ing faculty who read and commented upon the dissertation proposal. "Re- port differentiation" is the elaboration of different types of accounting reports for different purposes. "Sophistication of techniques," which could perhaps have been called accounting technology, is the use of ad- vanced accounting tools such as standard cost, break-even analysis, etc. "Personnel quality" of the accounting system is defined in the same way as personnel quality of the organization as a whole (see above, page 46): the technical knowledge and self-discipline of accounting employees. However, the function it serves for this dissertation is 1Simon et al. developed the variable, and it is used in much the same way in this dissertation as it was in their study. See above, page 12. 56 somewhat different for the accounting system than for the organization as a whole. For the overall organization, it is assumed to alleviate the necessity of other control systems. For the accounting system, it is assumed to be a.measure of the human technology of the system; i.e., the more qualified the accounting personnel, the more effective the out- put of the system, other things being equal. Thus it is a variable of the stage of development of the accounting system.1 "Accounting system.mechanization" is defined as the utilization of equipment to perfOrm accounting functions. Though there was some question whether mechanization of the overall organization creates con- trol prOblems or alleviates them (above, page 51), it seems logical that mechanization of the accounting system contributes to the control func- tion of accounting. Naturally, mechanization of the accounting system involves data processing equipment, including computers, and not produc- tion processing equipment. The accounting system "centralization of authority" variable was derived from the Simon et al. study. As was mentioned on page 13, their "centralization of authority" variable does not have anything to dO'With the classic definition of centralization of authority: the restriction of 1Of course, "personnel quality" probably serves both functions for both the accounting system.and the overall organization. "Person- nel quality of the overall organization," like the variables of "over- all structural complexity," furthers directly the accomplishment of the organization's goals, and "personnel quality of the accounting system" helps to alleviate control prOblems within the accounting system. HOw- ever, this study hypothesizes that the predominant effect of "personnel quality of the overall organization" is as a control system, and.the pre- dominant effect of "personnel quality of the accounting system" is as a dimension of the stage of development of the accounting system. 57 the right to make decisions at lower organization levels. Instead, it involves the extent the company has accounting units which are attached to operating units and not to the central controller's department. A very centralized accounting function would have all lower-level account- ing units and accounting personnel in a direct authority line to the controller and none reporting to operating officials. .A decentralized accounting function would have numerous accounting personnel and units attached to various operating departments with no reporting responsibil- ity to the controller. The Simon et al. definition was adopted for this study. Hypotheses as Incorporated in the Basic Model Though the basic theses to be tested in this dissertation have been stated at various points in the preceding sections and at various stages of development of the basic model, it is advisable to recapitu- late them here in their final form. The three hypotheses to be tested in this dissertation are: 1. Structurally complex organizations tend to have more fully developed accounting systems to contribute to the resolution of greater control and coordination problems, given the proc- ess and stage of development of other control systems is held constant . 2. The stage of development of the accounting system is inversely related to that of other control systems, when process and structural complexity are held constant, since control sys- tems are partial sUbstitutes for one another. 3. The more sophisticated is the production process of the or- ganization, the more the accounting system must be developed since it must provide more and better infOrmation for manage- ment decisions. 58 The meaningfulness of hypothesis two is dependent on the valid— ity of hypothesis one. If organizations do not have control problems, it is not meaningful to speak of substitution of control systems. How- ever, hypothesis two could be false even if hypothesis one is true. One implication of this would be that accounting is the only control system in organizations. It is possible that both hypothesis one and hypothesis three could be true. The stage of development of the accounting system could depend partly on the control needs induced by structural complexity and partly on the decision needs induced by the process. Chapter 3 RESEARCH DESIGN The purpose of this chapter is to lay out a program for testing the basic model developed in Chapter 2 and for determining specific re- lationships between characteristics of an accounting system and charac- teristics of the organization in which it operates. Techniques for processing the data up to the point of interpretation are described in this chapter so that the research findings chapter is not cluttered with research design material. The research findings are not interpreted in this research design chapter. However, most of the data used for the interpretation of the research findings are developed and presented in this chapter. Those data are summarized and often used in a different order in the research findings chapter. The order of presentation in this chapter is by steps in data development. The order of presentation in the research findings chapter is that which facilitates interpretation of the research data. The three sections of this chapter are Source of the Research Data, Refining the Data, and Analyzing the Data. Source of the Research Data comprises the determination of the sample of companies and charac- teristics of the company employees who provided the infOrmation for this dissertation. Refining the Data covers basically principal components analysis which was used to reduce a larger number of measurements to a 59 60 limited number of variables of the accounting system and the overall or- ganization. Analyzing the Data includes multiple regression analysis which was used to determine the relationships between the accounting system variables and organizational variables. It also covers tech- niques of using the multiple regression data to determine the strength and direction of influence of organizational variables and levels of variables on the accounting system. Techniques fOr determining the in- fluenceability of accounting system variables by organizational variables are also discussed. The two parts of the research findings chapter analyze, respec— tively, relationships among the characteristics of the accounting system and relationships between accounting system Characteristics and overall organization characteristics. The same research design techniques were used for each of these sets of relationships; specifically, multiple regression, explanatory power, and explainability.l In order to keep the discussion of these techniques in this research design chapter sim- ple, only data for the relationships between the accounting system char- acteristics and those of the overall organization are presented and discussed. SOURCE OF THE RESEARCH DATA In this section, the factors which were considered in determin- ing what type of sample of organizations to select and the general 1The technique "consistency with the hypotheses" only applies to relationships between accounting system characteristics and characteris- tics of the overall organization. 61 Characteristics of the sample companies are examined. Then the charac- teristics of the individuals within the companies who were interviewed to obtain the research data are discussed. The sample The objective of the sample selection process was to get a rela- tively homogeneous sample of companies which would not differ signifi- cantly due to environmental factors. Little environmental infOrmation was fOrmally collected in the study, and thus it was impossible to ex- plicitly control for environmental factors. Using a homogeneous sample means that the conclusions of the study cannot be generalized beyond companies of the same type. It must be left to other studies to deter- mine if the conclusions of this study apply to other types of organiza- tions. Table 1 lists some characteristics of the sample companies. It was decided that the entire sample would be composed of profit- making organizations. The study is a new application of comparative or- ganization structure research to accounting systems. Such a new research application should first be applied to traditional subjects. Accounting systems are most highly developed.in profit-making organizations, and thus relationships with the overall organization structure might be stronger than in nonprofit organizations. The sample was restricted to manufacturing companies without significant selling operations. This restriction helped satisfy the homogeneous sample objective stated above. Also, numerous manufactur- ing companies were available within reasonable travel distance of the researcher's home base, Lansing, Michigan. A further justification for 62 Table 1 Characteristics of Sample Companies gégy City Eggib;::s Line of Business 1 Jackson 210 MOtor vehicle parts 2 .Albion 650 welded wire and sheet metal 3 Lansing 245 Tools, dies, jigs, fixtures 4 Kalamazoo 385 Pumps, compressors 5 Jackson 525 Surgical, orthopedic appliances 6 Battle Creek 810 Paperboard 7 Kalamazoo 500 Machine tools 8 Grand Rapids 1,302 Hardware 9 Grand Rapids 700 Paint 10 Grand Rapids 730 Wire products 11 Grand Rapids 800 fabric finishing 12 Grand Rapids 869 Hardware 13 Jackson 105 Games and toys 14 Grand Rapids 305 .Aluminum processing 15 Saginaw 600 Sugar refining 16 Jackson 600 Mbtor vehicle parts 17 Lansing 186 Machine tools 18 Jackson 1,041 Automobile service tools Average 587 63 this restriction was that selling operations tend to have quite differ- ent organization structures from manufacturing operations, and it would have been difficult to control for this variability. Also, manufactur- ing operations are likely to have more mature accounting information systems and thus the role of the accounting systems within the control system.might be clearer than for selling operations. The researcher intended for all the companies to be autonomous. All companies which were subsidiaries of other companies were excluded from the sample except one company which was discovered to be a subsid- iary only after the interview was completed. This company was retained in the sample since it was managed separately from the parent company, a conglomerate. The reason for excluding subsidiaries was that it was felt that only autonomous companies would have complete sets of control sys- tems. Subsidiaries might be controlled in part by control systems in the parent organization, such as parent company internal audit staffs, inter- change of executive personnel between parent and subsidiary, and imposed Charts of accounts and standardized reports. Second, the primary inter- est was in control systems that were spontaneously developed as the re- sult of the condition of the company. Subsidiaries might have control systems which were imposed by the parent company rather than spontane- ously developed. As Table 1 indicates, the companies ranged in size from about 100 to about 1,300 employees. Several factors influenced the decision as to the size of companies to be included in the sample. First, it was felt that companies must reach a certain size before their control systems were mature enough to develop patterns required by the company's 64 complexity. For example, a small family-controlled business might need few control structures apart from the family grouping. Only when the company reached a size that the family could not control the business directly would significant control structures develop. Consequently, a floor should be placed on the size of companies permitted in the sample. Second, a ceiling was placed on the size of companies in the sample by the fact that few central offices of large companies were within practical driving distance of Lansing. Third, large companies would often require the participation of numerous respondents within the company. Such contacts with several employees of a company would have been time-consuming and not consistent with the data-collection design. Another factor restricted the range of size of companies to be included in the sample. Though variability in size was desired to explicitly measure the effect of size on the control systems of com- panies, the measurement of characteristics of companies of vastly dif- ferent sizes on the same scale might have been extremely difficult. The control systems of very large companies might be so different from those of small companies that a completely different interview instru- ment would.have been necessary. Though such comparisons of large and small companies would be a valid research study, they were considered outside the scope of this new research application. For all of the above reasons, it was decided to concentrate on small to medium-sized companies of at least 100 employees. An important characteristic of'most of the companies in the sam- ple is that they are suppliers of the large automObile manufacturers. This was inevitable due to the dominance of the auto industry in southern 65 Nfichigan. Even though all but one of the companies were not owned by another company, most were heavily dependent on the automobile manufac- turers for business. The effect of this dependence on the control sys- tems of the companies could not be measured in this study but should be considered in interpreting the conclusions. Identification of most of the candidates for inclusion in the sample was made from The Directory of Michigan Manufacturers 1istings fOr cities whose metropolitan areas had populations of from 100,000 to 500,000 within about 100 miles of Lansing,iMichigan.1 Detroit-area companies were purposely excluded as possibly being significantly differ- ent from the companies in the smaller cities and thus introducing varia- bility into the sample that could not be controlled. The following Nfichigan cities were selected by the above criterion: Jackson, Lansing, Kalamazoo, Battle Creek, Grand Rapids, and Saginaw; A few of the candi- dates were identified from Chamber of Commerce directories for those cities. Table 2 lists the cities from which sample companies were selected. .A letter requesting participation in the study was sent to the top financial executive of each of the companies (below, page 273). An attempt was made to select companies that had top financial executives who were members of the local chapters of the National Association of Accountants, as indicated by their listings in the NAA chapter director- ies fOr the respective cities. It was thought that NAA membership would 1The Directory ofKMichigan Manufacturers (Detroit: Manufacturer Publishing Co., 1971). 66 Table 2 Populations of Cities from Which Sample Companies were Selecteda Population City Within Metro- City politan Limits Area Jackson ........................................... 44,500 132,500 Albion ............................................ 12 , 112b 20 ,000 Lansing ........................................... 133,000 311,000 Kalamazoo ......................................... 86,000 214,000 Battle Creek ...................................... 38,200 110,700 Grand Rapids ...................................... 196,500 455,000 Saginaw ........................................... 91,000 184,500 aBritannica Atlas (Chicago: Encyclopaedia Britannica, Inc., 1972), pp. 122-123. bThe Official Associated Press Almanac 1975 (Maplewood, N.J.: Hammond.Almanac, Inc., 1975), p. 774. 67 increase the probability of agreement to the interview and active co- operation with the interviewer. One company from the smaller town of Albion, Michigan, was selected for the sample exclusively on the basis of a personal contact through NAA. The sample is a selective one and therefore no conclusions can be drawn statistically from the research results about companies other than those in the sample. Nevertheless, various tests of significance are performed throughout the data analysis. These can be justified in two ways. First, the test of significance is a convenient and Objective criterion to distinguish large associations, worthy of attention, from small associations, not worthy of attention. Second, according to the Cornfield-Tukey argument for inference, the statistical results derived from a selective sample can be generalized to a population from which the sample might have been drawn (similar to the sample). The charac- teristics of the sample should be described and the reader of the re- search report can judge whether the universe of similar companies is of interest to him.1 The Respondents InfOrmation was collected by means of interviews with the con- trollers, treasurers, or other financial personnel of the sample com- panies. For fOurteen of the companies there was only one respondent, but for the other four companies two persons shared the answering of the 1Jerome Cornfield and John W. Tukey, "Average values of Mean Squares in Factorials," The Annals of'Mathematical Statistics, XXVII (1956), 912-13. 68 questions. For eleven of the companies the chief respondent was the top official in the finance and accounting function. For six of the compa- nies the chief respondent was on the level below the top official in the finance and accounting function. For one company the president was the respondent. Table 3 lists the titles of the respondents. REFINING THE DATA The essential elements of the statistical research design are principal components analysis fOr reducing a voluminous set of measure- ments to a manageable set of variables, and multiple regression analysis for analyzing the relationships among the variables.1 Multiple regres- sion analysis is discussed in the next section. This section is concerned essentially with principal components analysis. But first the questions addressed to the respondents are examined, various manipulations of the answers to those questions are outlined, and the assembly of the measure- ments which are the raw data for the principal components procedures is described. The important features of principal components analysis are conveyed in a nontechnical manner by means of two illustrations, one of the combination of measurements into a component and the other of 1The idea fOr the combination of principal components and multi- ple regression analysis came from Green and Tull. They suggested that principal components analysis can be used to reduce the number of ex- plainer variables (while retaining as much as possible of the original information) and also to reduce the multicollinearity among the explainer variables. Both of these factors facilitate the interpretation of mul- tiple regressions. See Paul E. Green and Donald S. Tull, Research for Marketing Decisions (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1970), pp.424-26. 69 Table 3 Titles of Respondents Position CONETHY Titles SETSOHZ Finghce Function 1 Treasurer 1 l 2 Controller; Accounting Manager 2 l 3 Treasurer l 1 4 VP-Finance l 1 5 Sec/Treas; Asst Treasurer 2 l 6 Controller/Secretary l 2 7 Credit Manager 1 2 8 Mgr of Finance; Gen Acct Mgr 2 l 9 Accounting Manager 1 2 10 Head of Finance 1 1 11 Controller 1 2 12 Controller 1 l 13 President 1 0 l4 VP-Finance l 1 15 Controller/Asst Treas; Asst Gen Fact Mgr 2 2 16 Financial Manager 1 1 l7 Secretary/Treasurer 1 l 18 Controller (Tool Div) 1 2 70 rotation of components. Then the output of the principal components procedures, the component scores, are characterized. Collection and Disposition of'Measurements One hundred thirty-eight questions were directed to the respond- ents. These questions are listed in the central column of Appendix A (pages 212-38), along with the measurement rules used to interpret the responses to questions. Since 16 of the 138 questions were multiple-part questions, the total number of measurements was 166.1 The measurements were recorded on an SPSS2 system tape file for use in further processing. The SPSS calculation capabilities were used to calculate 117 new measure- ments by mathematical combinations of the original measurements. These new measurements were usually calculated by subtracting one measurement from another, dividing one measurement by another, multiplying one meas- urement by another, or adding one measurement to another. In some cases more than two measurements were combined to fOrm the new measurement and more than one of the above mathematical functions were used in the cal- culation. The 117 new measurements were added to the SPSS system tape 1One of these measurements, question 33, was dropped from the analysis at this point since seven of its responses were missing and the rest exhibited no variability. Three of the questions (R1, R2, and R3) pertained to requests for general infOrmation from the respondents, such as organization charts, financial statements, and personnel lists. The answers to these questions, though coded on the SPSS system tape file, were not used in the subsequent analysis. zAll of the data manipulation and statistical data processing in this dissertation were performed with SPSS, which is a very versatile statistical program prackage. See Norman H. Nie, C. Hadlai Hull, Jean G. Jenkins, Karin Steinbrenner, and Dale H. Bent, Statistical Package for the social Sciences (San Francisco: McGraw-Hill Book Company, 1975). 71 file.1 .At this point, the total number of measurements on the SPSS sys- tem tape file was 284 (including the responses to the original questions [166], the calculated measurements [117], and company interview order). These measurements are listed on Table 19, pages 274-92. The numbers in measurement names correspond to question numbers in Appendix A, pages 212-38. The names of original measurements begin with an R (for raw), and calculated measurements begin with a C. The fOrmula used to calcu- late the new measurements is listed under the caption "Description and Calculation." One hundred two of the raw (uncalculated) measurements were discarded once they were used to calculate a new measurement, since the calculated measurement was more clearly related to a variable in the basic model. Several of the measurements collected for the eighteen companies had.missing responses.2 These missing responses, when used in calculating the new measurements, resulted in missing responses for them also. .Measurements with more than three missing responses were dropped from the analysis, and one measurement was dropped with two missing responses. A total of ten measurements were dropped due to having too few responses. 1One additional measurement, "company interview order," was added to the SPSS system tape file for possible later tests to deter- mine if there was drift in the interviewer's interpretation of responses to questions. 2Missing responses were due in only one case to the respondent claiming the infOrmation was confidential or proprietary. In that case, executive salary information was not Obtained. MOst missing responses were due to the inaccessibility of the data to the respondent or the in- applicability of the measurement to the company. 72 Principal Components Analysis One hundred sixty-four measurements remained after all of the eliminations described in the previous section. These were grouped by the twenty-six variables from the basic model shown in Figure 1, page 44. Table 20, pages 293-99, lists the groups of measurements in the two right-hand columns. Each group contains measurements which the re- searcher felt would be associated with a particular variable from the basic model. Such groups were assembled for twenty-five of the twenty- six variables in the basic model. The remaining variable, "materials input diversity," is represented by a single measurement, R11. The focus of interest of this study was on the twenty-six vari- ables developed for the basic model illustrated on page 44. Each of these variables was represented by from one to eighteen measurements. Statistical techniques which are described on pages 89-127 will be used to determine the influence of the twenty-six variables on each other. But a necessary first step is to derive a single measure of each of the variables. The measures of the variables were derived using principal com- ponents analysis perfOrmed with a statistical program package known as SPSS.1 Principal components is a fOrm of factor analysis, and its 1Nie et al., pp. 468-514. An excellent nontechnical description of principal components analysis can be fOund in Green and Tull, pp. 402-22. The idea of using principal components analysis to combine or- ganizational measures came from the Pugh et al. studies; for a specific application, see D. S. Pugh, D. J. Hickson, C. R. Hinings, and C. Turner, "The Context of Organization Structures," Administrative science Quar— terly, XIV (MarCh, 1969), p. 100. 73 application to this study is described on the following pages. The com- bined measures of the variables (to be called components) are listed in Table 4 along with some information about them which will also be ex- plained in the following pages. For nineteen of the variables, a single component was derived, as indicated by a one in the column of Table 4 headed "Number Components." For six of the variables, two components were derived, as indicated by a two in the column. Example of Derivation and Inter- pretation of'First Cbmponent As an example of combining measurements, consider the three measurements that are grouped under staff support:1 Com- C48 R49 C50 pany Staff Staff Clerical No. Ratio Functions Ratio 1 -0.0202985 0.6352100 -0.4951830 2 0.1707500 0.0337000 0.7957610 3 -l.0770100 -0.5682990 -0.47S9200 4 1.0364200 0.0334600 2.3757200 5 -1.9068700 0.6352100 0.4874800 6 -0.7426870 -0.5682990 -0.8227360 7 0.6483580 -0.5682990 -0.8034680 8 0.3976120 1.8387300 0.0828516 9 1.9200000 0.0337000 -0.1868980 10 0.4274630 0.0337000 -0.8612720 11 -0.2411940 0.0337000 1.1811200 12 0.3259700 0.0337000 -0.3795760 13 -0.9038810 -l.l700600 0.4104050 14 0.7916420 -l.l700600 -0.7263970 15 1.3468700 -1.l700600 -0.4951830 16 -0.l695520 -l.1700600 -0.9768790 17 -1 5844800 0.6352100 -0.996l460 l8 -0.4203000 2-4404900 1.8554900 1Note that these scores are standardized to mean zero and stan- dard deviation one. C48, R49, and C50 indicate measurement numbers (see Table 19, page 274. 74 Table 4 Component Description Percent No. of No. of . . Component Name Measure- Compo- Elgen- Varlance mentsa nents value ET" plained Process Sophistication .................. 7 1 3.5 50.5 Output diversity ................ 2 l 1.3 65.2 Overall structural complexity Size ............................ 6 1 5.6 93.9 Job structure complexity ........ 3 l 2.1 71.4 Geographical dispersion ......... 4 1 2.8 69.4 Divisional differentiation ...... 6 1 3.8 63.6 Divisional specialization ....... 4 l 2.4 60.3 Mechanization-general 8 mechanization-computers ....... l4 2 7.9 56.8 Cbntrol system Direct supervision .............. 3 l 2.0 67.2 Staff support ................... 3 l 1.5 48.6 Authority levels ................ 2 l 1.7 86.2 Personnel quality-high level 8 personnel quality-low level ... 15 2 7.0 47.0 Centralization-investment 8 centralization-purchasing ..... 18 2 8.9 49.5 Standardization-jobs & standardization-general ....... 20 2 10.5 52.5 Accounting system Size-infOrmation output & size-resource input ........... 7 2 3.7 53.4 Job structure complexity ........ 4 1 2.0 50.8 Geographical dispersion ......... 6 l 3.4 56.3 Unit differentiation-vertical ... 2 l 2.0 97.8 Unit differentiation-horizontal . 3 l 1.6 54.4 Authority levels ................ 4 l 3.0 73.9 Report differentiation .......... 7 l 4.3 61.5 Decentralization of accounts .... 3 l 1.7 56.3 Sophistication of techniques .... 6 1 3.3 55.2 Mechanization ................... 3 l 2.4 79.7 Personnel quality-education G personnel quality-general ..... 8 2 3.3 41.8 Average ............... . ....... .... 3.0b 62.5C 3See Table 20, pages 293-99, fOr listings of the measurements combined into each component. bFor the 31 components. cFor the 25 groups. 75 One method of combining the measurements in a group into a single measurement would be to add the three measurements fOr each company. This method would not take into account the differing importance of the meas- urements to the variable.1 .A simple and general method of combining measurements is known as a linear combination. In this method, the combined measurement for a given company is calculated by multiplying each measurement by a coef- ficient and summing the products. For example, the coefficients for staff ratio, staff functions, and clerical ratio might be 2, 0.5, and 8, respectively. The combined score for staff support for company number five would be: Staff Staff Staff Clerical . ratio x iii: functions x func- + ratio Clreailizal = coeffi- score coeffi- tions coeffi- score cient cient score cient (2 x -1.90687) + (0.5 X 0.63521) + (8 x 0.48748) -= Combined score, company five = 0.40371. The scores for the other seventeen companies can be calculated in the same way. Note that the coefficients determine which measurements have the greatest influence on the combined score. The larger the coefficient, the greater the influence of the corresponding measurement.2 1It also does not take into account the different scales of the measurements. Measurements with a larger variance would influence the combined measurement more than other measurements. 2Green and Tull, p. 405. 76 What are the appropriate coefficients? The grouping of the three measurements together presumed that each measurement had something to do with the variable, but the exact way that the measurement related to the variable was not known. It seems logical that the variable rep- resents something that is common to all three measurements. The first component extracted by principal components analysis is a linear combi- nation of the measurements which accounts for more of what is common to the three measurements than any other possible linear combination. Principal components procedures compute coefficients which can be used to calculate component scores for each company in the sample. The component coefficients for the three measurements are, respectively, -0.l8578, 0.59188, and 0.54846.1 These can be used to calculate the com- ponent score for company five, as follows: Staff Staff Staff Staff Clerical Clerical ratlo x ratio + functlons )< func- + ratlo x ratio = coeffi- score coeffi- tions coeffi- score cient cient score cient (-0.18578 X ~1.90687) -+ (0.59188 X 0.63521) -+ (0.54846 X 0.48748) Component score, staff support, company five = 0.99759. The component scores for the other seventeen companies are calculated in the same way. 1The coefficients are for the first (unrotated) component. The derivation of subsequent components is discussed below, on page 81. 77 Com- Staff Support pany Component No. Score 1 0.104944 2 0.428329 3 —0.397638 4 1.133827 5 0.9975631 6 -0.650715 7 -0.896963 8 1.063812 9 -0.435986 10 -0.534300 11 0.711681 12 -0.247952 13 -0.297102 14 -l.235087 15 -l.210088 16 -1.193071 17 0.119369 18 2.539277 The staff support component scores were used as a new measure- ment of the variable staff support, and the three old measurements were discarded. ,A unitary measure of staff support has been Obtained, but some of the infOrmation in the original three measurements has been lost since one measurement can seldom convey as much information as three measurements. For assessing the adequacy of the component scores, the relative amount of the infOrmation contained in the three discarded meas- urements that is retained in the component must be determined. The infOrmation value of a score can be thought of as the degree it varies from the mean of the scores on that measurement. For example, 1The component scores were calculated by an SPSS procedure and are more accurate than hand-calculated figures since more significant digits are carried through the calculations. This accounts for the difference between company five's component score as calculated in this text and the component score as calculated by the program. 78 the score for company nine on "staff ratio" is 1.92.1 This indicates that company nine had an above-average staff ratio since the mean of the scores is zero. But how much above average? The average deviation of these scores from the mean, known as the standard deviation, is one. Consequently, it can be said that company nine had a staff ratio which was significantly above average; i.e., ahmost twice the average deviation. If each of the three measurements has a standard deviation of l,2 the total variation of the three measurements is three. "Eigenvalue" is defined as the variance of the component scores3 but can best be inter- preted as the equivalent number of measurements which the component explains. For example, the eigenvalue of the staff support component is 1.458. The staff support component is thus about 50 percent better than any one of the original measurements in explaining the total vari- ation of the three components. In general, components with eigenvalues near one are poor explainers since they explain little more than an original measurement. The maximum.possible eigenvalue is the number of measurements times one (the average variation of each measurement). This would occur only if all the variables were perfectly correlated. Eigenvalue can be used to calculate a second measure of the ade- quacy of the component "percent variance explained": eigenvalue \ total variation original measurements} Percent variance explained = 100( 1See above, page 73. 2Standard scores must always have a standard deviation of one. 3"Variance" is the square of the standard deviation. 79 For staff support, the calculation would be: 100(l.458/3) = 48.6 . o\° Staff support explained 48.6 percent of the variation of the original three measurements. Though eigenvalue and "percent variance explained" indicate how much of the total variance of the three measurements is explained by the component, they do not indicate how much of the variance of each indi- vidual measurement is explained.by the component. InfOrmation about the explanation of individual measurements comes from component loadings. These are the simple correlations of the component scores with the origi- nal measurements. For staff support, the component scores on page 77 are correlated with the original measurement scores on page 73 to Obtain the following component loadings: C48, staff ratio -0.27 R49, staff functions 0.86 C50, clerical ratio 0.80 The range of a simple correlation is from -1 to l. The closer a loading is to 1 or -1, the greater the degree of association between the component and the original measurement. The closer a loading is to zero, the less the degree of association between the component and the origi- nal measurement. Measurements are said to load highly on a component when their loadings are near 1 or -1. Thus "staff functions" loads highly on the component while "staff ratio" loads only weakly on the component. A positive loading indicates that when the component score fOr a company is large the original measurement score for that company 80 tends to be large. A negative loading indicates that when the component score for a company is large the original measurement score for the com- pany tends to be small. The lack of any association between a component and an original measurement is indicated by a zero loading. The three original measurements were selected because they were believed to measure aspects of the variable "staff support." Why then did "staff ratio" load only weakly on the "staff support" component? .Also why did it have a negative loading on the "staff support" compo- nent? While the answers to these questions can only be surmised, it is important to note that "staff functions" and "clerical ratio" are highly correlated with each other, while "staff ratio" is only weakly corre- lated with either "staff functions" or "clerical ratio." In order to explain the most possible variance, the component explained the common aspect of "staff functions" and "clerical ratio" but it was not able to explain.muCh of "staff ratio" because "staff ratio" shared little with the other two measurements. The meaning of a component is determined by the measurements which load highly on it. Thus when a measurement which was originally assigned to measure a variable loads weakly on the component derived to measure that variable or has a loading with the wrong sign, a somewhat different meaning may be applied to the component than was intended for the variable. For example, see below, pages 242-46, for the interpreta- tion of the four overall structural complexity components: "jOb struc- ture complexity," "geographical dispersion," "divisional differentiation," and "divisional specialization." In the case of "staff ratio," the rea- son for the weak and inconsistent loading was assumed to be due to a 81 poor measurement (see below, page 249). Thus the meaning of the com— ponent was not changed from that intended for the variable "staff sup- port." A rough indication of the strong positive association of the "staff functions" scores (above, page 73) and the "staff support" com- ponent scores (above, page 77) can be obtained by comparing these scores. Note that, for fifteen of the eighteen companies, the component score has the same sign as the "staff functions" score. The weak negative association of the "staff ratio" scores and the "staff support" compo- nent scores is indicated by the fact that, for eleven of the eighteen companies, the component score has the opposite sign of the "staff ratio" score. Derivation of’second Cbmpo- nent for Some variables ,A single component was extracted for nineteen of the twenty-six variables in the basic model (see Table 4, page 74). These components were analyzed and interpreted in the manner described on the preceding pages. Though some infOrmation was lost by discarding the original measurements for these nineteen variables, this disadvantage was out- ‘weighed.by the advantage of reducing the number of measurements. For eighteen variables, the percent variance explained by the component was greater than 50 percent, and fOr "staff support" it was very close to 50 percent. Thus it was felt that the components fOr these nineteen variables must measure at least an important aspect of the respective variables. 82 For the other six variables, the level of explanation of the original measurements in terms of percent variance explained by the com- ponent was not considered adequate: Pct. variance Component Name EggigtnggfiTY ponent Overall structural complexity Mechanization 31.2 Cbntrol system Personnel quality 25.3 Centralization 33.6 Standardization 37.1 Accounting system Size 33.3 Personnel quality 23.4 The reason for these poor levels of explanation is that the general degree of correlation among the original measurements fOr the respective variables is low. As was implied for the "staff support" component on page 80, correlation among the original measurements facilitates the explanation of those measurements by a component. For any group of measurements, several components (no more than the number of measurements) can be calculated with principal components analysis. Each component is a linear combination of the original meas- urements similar to that illustrated on page 76, but with different coef- ficients. The first of these components, which has been illustrated, accounts fOr more of the variance of the original measurements than any other possible linear combination. The second component accounts fOr more of the variance that has not been accounted for by the first com- ponent than any other possible linear combination. However, the second 83 component must be uncorrelated with the first component. In other words, the component scores on the second component must have zero correlation with the component scores on the first component. Subsequent components fellow the same pattern: greatest explanation of variance yet unexplained and uncorrelated with all prior components. In order to achieve an adequate level of explanation of the original measurements of the six variables for which the first component was inadequate, components had to be added. However, fOr the sake of minimizing the number of components whose interrelationships must later be analyzed, it was decided to limit the number of components for any group of measurements to two. The addition of the second component increased the percent variance of the original measurements explained as follows: Percent Variance Explained by: First Second Component Name Com- Com- Total1 ponent ponent Overall structural complexity Mechanization 31.2 25.6 56.8 Cbntrol system Personnel quality 25.3 21.7 47.0 Centralization 33.6 15.9 49.5 Standardization 37.1 15.4 52.5 Accounting system Size 33.3 20.1 53.4 Personnel quality 23.4 18.4 41.8 The level of explanation of the original measurements by two components was considered adequate for all but the two personnel quality variables, 1From Table 4, page 74. HF 84 control system and accounting system. Essentially 50 percent of the variance of the measurements of the other four variables was explained by respectively two components1 The inadequate level of explanation of the two personnel quality variables was due to two factors. First, both variables had a lot of measurements——fifteen for control system and eight for accounting sys- tem. Second, the measurements were of a very diverse nature: various aspects of the education level, seniority, and salary levels of employ- ees. Rotation of'First and Second Cbmponents The two components isolated for each of the six variables were rotated befOre being used in the analysis of interrelationships among variables. In order to describe the nature of rotation, it is valuable to examine one variable in detail, "accounting system size," which in- cluded seven measurements. To review what has occurred up to the point of rotation, the principal components procedure has calculated coeffi- cients for a first component which explains more of the variance of the seven.measurements than any other possible linear combination. Using these coefficients in a linear combination of the original measurements such as that illustrated above on page 76, it calculated component scores for the eighteen companies (such as those on page 77 above). Then the procedure calculated the coefficients fOr a second component which ex- plains more of the variance of the seven measurements left unexplained 1"Centralization" was shy by only 0-5 percent. 85 by the first component than any other possible linear combination. The second component scores are calculated using the coefficients in a lin- ear combination of the original measurements. Then the procedure cal- culated the correlations of the two component scores with each of the seven original measurement scores.1 These are the component loadings and are as follows: Accounting Size Compo- Compo- nent l nent 2 Proportion controllership employees -0.77 0.41 Proportion employees receiving reports -0.61 -0.37 Number data centers -0.47 -0.61 Data center elaboration -0.44 -0.36 Average report frequency -0.21 0.33 Proportion controllership expenses -0.65 0.63 Controllership expense emphasis 0.72 0.17 Two components representing accounting system size have been derived.which account in total for 53.4 percent of the variance of the original measurements (see above, page 83). At a later stage in the study, the relationships of these components to other components might be investigated. HOW'Will these relationships be interpreted? Which component is "accounting system size"? What would be the interpretation if only one of the two components related to another component? Observe the loadings. Note that many of the measurements load fairly highly on both component 1 and component 2. Thus there is little opportunity to distinguish the meaning of component 1 from component 2. 1The description and ordering of mathematical steps in this text are convenient for exposition. The computer program mathematics are quite different. 86 All that can be said is that these two components represent accounting system size, are uncorrelated with one another, and cannot be distin- guished in meaning. A solution to this problem is produced by rotation of the two components. Rotation produces two rotated components which are linear combinations of the old components: Coefficient Coefficient Rotated old Old old Old component 1 = component 1 component 1 + component 2 component 2 score for rotated score fOr rotated score component 1 component 1 Coefficient Coefficient Rotated old Old old Old component 2 = component 1 component 1 + component 2 component 2 score for rotated score fOr rotated score component 2 component 2 Note that eaCh rotated component has a portion of both old component 1 and old component 2 in it. Since any two variables can always be fully explained as linear combinations of any two other variables, there is no loss of information due to the rotation.1 But why do it? The purpose of rotation is to find rotated components such that the loadings on the original measurements are as close to 1, O, or -1 as possible. In other words, the original measurements should load highly on components or not at all. This facilitates the differentiation of the meaning of the two components, because measurements will tend to be loaded highly on one component and weakly on the other. Thus the 1In other words, the percent variance explained by the two rotated components is still 53.4 percent. 87 original measurements that load highly on one component can be inspected to see what distinguishes them from the measurements which load highly on the other component. The loadings of the original measurements on the varimax rotated components1 are as follows: Accounting Size Rotated Rotated Compo- Compo- nent l nent 2 Proportion controllership employees 0.26 0.84 Proportion employees receiving reports 0.72 0.09 Number data centers 0.75 -0.13 Data center elaboration 0.58 0.01 Average report frequency -0.10 0.42 Proportion controllership expenses 0.08 0.88 Controllership expense emphasis -0.64 -0.38 Note that all of the original measurements load strongly on one rotated component and almost not at all on the other.2 Thus the mean- ing of the two rotated components can be readily distinguished.’ The meaning of component 1 is determined by the high-loading measurements "proportion employees receiving reports," "number data centers," and "controllership expense emphasis." The high loading on "controllership 1See below, page 258, for the specific analysis of the loadings of accounting size components. varimax rotation was selected from among several available rotation methods because it is the most commonly used method. Varimax rotation produces rotated components which have zero correlation with each other. See Green and Tull, pp. 418-21. 2Controllership expense emphasis is a possible exception, since it loads modestly on rotated component two. ’The two accounting components are interpreted in much more detail below, on pages 258-60. This paragraph is intended merely to suggest how the meaning of components can be differentiated. 88 expense emphasis” was assumed to be due to a measurement error. The other two high-loading measurements have to do primarily with the infor- mation output of the accounting system, and so rotated component 1 was interpreted as size of the infOrmation output of the accounting system. The high-loading measurements on rotated component 2 are "proportion controllership employees" and "proportion controllership expenses." Employees and expenses are company resources allocated to the account- ing system, so rotated component 2 was interpreted as size of the re- source input of the accounting system. The two components extracted for each of the other five variables listed on page 83 above were varimax-rotated in the same manner as those for accounting size. The meaning of these components was interpreted, along with the single-component variables, on pages 239-72 below. These interpretations were used to assign different component names to the two components for each of the six variables. These different names are listed in Table 4, page 74. Cbmponent Scores It is important to review the output of the data-refinement stage of this dissertation. Thirty-one new measurements, called compo- nents, have been derived by means of principal components analysis. These are listed in Table 4, page 74. For each component, there are eighteen component scores, one for eaCh company. All of the original measurements have been discarded except fer one, "materials input diver- sity," which will be retained along with the thirty-one components to make a total of thirty-two measurements. These thirty-two measurements 89 correspond to the twenty-six variables in Figure 1, page 44. Since two components are included in six of the variables, these variables are split to form two variables. There are now thirty-two measurements measuring thirty-two variables, as indicated in Figure l. ANALYZING THE DATA The output of the prior section on refining the data is a set of accounting system variables and a set of organizational variables of three types: process, overall structural complexity, and control system. The basic statistical technique used.in this section, multiple regression, is applied to the prOblem of relating the accounting variables to the organizational variables. One of the multiple regressions used in the study is illustrated in this section as a means of communicating the essential aspects of multiple regression in a nontechnical manner. Then techniques of combining the multiple regression results to answer the questions addressed by this dissertation are described. These questions are classified as explanatory power, explainability, and consistency with the hypotheses. The "Explanatory Power and Explainability" subseCtion develops measures of the strength of influence of the organizational variables and the levels of organizational variables on the accounting system. It also develops measures of the influenceability of accounting variables by organizational variables. The "Consistency with the Hy- potheses" subsection assesses the direction of influence of the organiza- tional variables on the accounting variables as compared with the direc- tion predicted in the hypotheses. The "Analyzing the Data" section 90 concludes with a discussion of the limitations of the research design of the dissertation. Stepwise Multiple Regression As has been implied at various points in Chapters 2 and 3, a key objective of the dissertation is to explain accounting system variables with organizational variables. Thirteen of the thirty-two variables de- veloped in the data-refinement section (above, pages 68-88) are account- ing system variables. The other nineteen variables are organizational variables of three types——process (three variables), overall structural complexity (seven variables), and control system (nine variables)-—which are to be used to explain the accounting variables. An SPSS stepwise linear multiple regression procedure was used to analyze the relation- ships between the organizational and the accounting system variables.l On the following pages, one of the regression procedures calculated in connection.with this dissertation is used to illustrate the nature of stepwise multiple regression and to develop some of the terminology that ‘will be used in subsequent analysis. Linear multiple regression is a very common technique of explain- ing a single variable with a set of other variables.2 Essentially, lin- ear multiple regression attempts to find a linear combination of a set 1Nie et al., pp. 320-67. 2An excellent intuitive description of linear regression can be fatmd in Green and Tull, pp. 343-64. For a rigorous, extensive, more mathematical treatment, see N. R. Draper and H. Smith, Applied Regres- sion Analysis (New York: John Wiley 8 Sons, Inc., 1966), pp. 1-35, 163-77. 91 of explainer variables which is most highly associated with the single variable (called criterion variable in this dissertation).1 As applied to this study, the single variable is any one of the thirteen accounting system components, and the set of explainer variables is any set of the nineteen organizational components. Example of'MuZtipZe Regression As an example of linear combination of explainer components to explain a criterion component, take the regression of the criterion component "accounting size-information output" on the four explainer components: "process-sophistication," "job structure complexity," "direct supervision," and "personnel quality-low level."2 The component scores on these five variables are as follows: Criterion Component Explainer Components Com- Acigu::ing Process JOb Direct Personnel pany Information Sophis- Structure Super- Quality- No. Ou tication Complexity vision Low Level tput 1 -0.602657 -0.296721 0.010537 -O.606413 0.009474 2 1.124027 -0.398268 1.602391 0.458668 0.472895 3 1.340012 1.390160 1.114058 -2.317780 1.995165 4 -0.395408 -0.l38480 1.552997 0.452153 0.882082 5 -l.126679 -0.203618 0.126531 -0.382799 -2.454819 6 0.777250 -0.416430 0.332067 0.335229 1.142752 7 -0.43Z310 -0.200983 1.519192 0.169600 -l.013620 1Criterion variables often are referred to and explainer variables as independent variables. as dependent variables, 2These particular explainer components were selected by the stepwise regression procedure that is described below on pages 98-100. The discussion at this point is facilitated by treating the example as straight multiple regression. 92 Criterion Component Explainer Components Com- ‘Aci%:::}ng Process Job Direct Personnel pany Information Sophis- Structure Super- Quality- No. Output tication Complex1ty Vision Low Level 8 2.324751 2.473550 -l.661982 1.661989 0.379315 9 -0.824774 -0.919824 1.276731 1.138866 0.104121 10 -0.l76136 -0.412887 0.176635 -0.884163 0.448130 11 -0.127615 -0.596342 -0.624720 1.107347 0.058178 12 -0.745879 -0.325225 -0.178166 -0.599827 -0.628232 13 0.098104 -0.529910 -1.646463 -0.328273 -1.378594 14 -1.062548 -0.266749 0.174330 -0.796742 0.281690 15 -0.429995 -0.625398 0.491589 1.343424 0.212429 16 -0.875054 -0.598388 0.366133 -1.195680 -0.059280 17 1.490802 2.395775 0.576389 0.056460 -0.086073 18 -0.355890 -0.330261 0.224648 0.387938 -0.365613 The first step of the multiple regression analysis is to find a set of fOur coefficients which, when multiplied by the four explainer component scores and summed, will produce an estimate of the "accounting size-information output" score, as follows: Estimated accounting size-infor- mation out- put score Process Process = sophis- sophis- tication tication coefficient score Direct Direct + supervision X supervision coefficient score + Job structure complexity coefficient Personnel quality- low level coefficient JOb structure complexity SCOTC Personnel quality- 1ow level score The coefficients determined by the SPSS procedure are, respectively, 0.59857 fOr "process sophistication," -0.373l7 fOr "job structure com- plexity," 0.23679 for "direct supervision," and 0.32078 fer "personnel quality-low level." ponents would be combined as fellows: For company number five, the scores on the fOur com- 93 (0.59857 X -0.203618) + (-0.37317 X 0.126531) + (0.23679 X -0.382799) + (0.32078 X -2.4S4819) Accounting size-infOrmation output estimated score for company five ‘1'0472° The estimated scores for the other seventeen companies are calculated in the same manner. These are listed here along with the actual scores: Accounting Size-Information Output Com- pany Estimated Actual No. Score Score 1 -0.322090 -0.602657 2 0.619880 1.124027 3 1.339020 1.340012 4 -0.272410 -0.395408 5 -1.047200 -1.126679 6 0.072770 0.777250 7 -0.972210 -0.432310 8 2.616010 2.324751 9 -0.723950 -0.824774 10 -0.378670 -0.176136 11 0.157050 -0.127615 12 -0.471730 -0.745879 13 -0.222730 0.098104 14 -0.323030 -1.062548 15 -0.l71550 -0.429995 16 -0.796960 -0.875054 17 1.204710 1.490802 18 -0.306930 -0.355890 Note that the estimated scores are reasonably close to the actual scores, indicating the correlation between the two sets is fairly high. The sign of the estimated score is correct for sixteen of the eighteen companies.l Not only are the estimates good, they are the best that 1The other two companies have scores near the mean of zero, which considerably increases the chance of opposite signs. 94 could be obtained with these four explainers in a linear combination. In other words, the correlation of the estimates with the actual scores is higher for this set of coefficients than for any other possible set of coefficients applied to the four explainer components. This linear combination of explainer components is analogous to the linear combination of measurements which produced the components.1 However, the objective of the linear combination is different. The coef- ficients of the linear combination of measurements (principal components analysis) were chosen so that the correlations of the component scores with the measurements would be as high as possible. The component thus represents the common aspect of the measurements. The coefficients of the linear combination of explainer components (linear multiple regres- sion) are chosen so that the combination scores (or predicted scores) are most highly correlated with the criterion component, not with the ex- plainer components themselves. Thus the combination scores incorporate the portions of the components which are most associated with the cri- terion, not the common aspect of the explainer components themselves. The regression estimates are good, but how good are they? R square is a measure of the explaining power of a regression. An expla- nation of its meaning follows. For a particular company, the quantity that the four explainer components should try to predict can be thought of as the deviation of the "accounting size-infOrmation output" score from the mean of the accounting size scores (mean is zero). For example, company five had an actual score of -1.126679, which is also its deviation 1See above, page 76. 95 from the mean of the "accounting size-information output" scores, zero. The total variation for the eighteen companies that the four explainer components should try to predict is the total of the deviations from the mean for the eighteen companies. However, since some deviations are positive and some are negative, the total of the deviations is always zero. A mathematical solution to the problem of totaling the deviations from the mean is to square each one, which makes it positive, and then sum the squares. Thus the total variation for the eighteen companies that the four explainer components must try to explain is the sum of the squared deviations from the mean. Since the actual scores for "account- ing size-infOrmation outpu " are already deviations from the mean, they can be squared as is: 2 2 2 (-0.602657) + (1.124027) + . . . . + (-0.355890) Sum of squared deviations of = 16 7658 actual scores from.the mean ° ' For a particular company, the quantity that the four explainer components actually did predict is the deviation of the estimated score from the mean.1 For the eighteen companies, the total variation actu- ally predicted is equal to the sum of the squared deviations of the estimated scores from the mean, as follows: (4.32209)2 + (0.61988)2 + . . . . + (-0.30693)2 Sum of squared deviations of = 14 5246 estimated scores from the mean ' ' 1Like the actual scores, the estimated scores are deviations from the mean of zero. 96 R square is a.measure of the ability of a set of explainer com- ponents to predict the values of a criterion component. R square is computed as follows: Sum of squared deviations of estimated scores from the mean Sum of squared deviations of actual scores from the mean R square = Per the regression of "accounting size-information output" on the feur explainer components, R square is as fellows: R square = %%i;%%§- = 0.866. In other words, about 87 percent of what was available to be explained was explained by the four explainer components. The regression coefficients are used as measures of the relative importance of the four explainer components to the prediction of the cri- terion component. Ordinarily the maximum value of a regression coefficient is 1, and the minimum value is -1. Regression coefficients near 1 or -1 are vital to the prediction, while regression coefficients near 0 have little importance to the prediction.1 .As listed above on page 92, the feur regression coefficients are: 1Since all the scores on the components are normalized to mean 0 and standard deviation 1, the coefficients using them are standard- ized regression coefficients. Standardized coefficients can be com- pared with one another since both the criterion component and the explainer components are on the same scale. One measurement, "mate- rials input diversity," which was not normalized since it is not a component, was used as an explainer in some regressions. Wherever a regression coefficient for materials input diversity is presented in this dissertation, it is the standardized regression coefficient. This coefficient is the same one that would be Obtained if materials input diversity had been normalized. 97 Process sophistication 0.59857 Job structure complexity -0.373171 Direct supervision 0.23679 Personnel quality-low level 0.32078 "Process sophistication" is the most important variable to the predic- tion, while "direct supervision" is the least important. The regression coefficients also indicate the direction, positive or negative, of the relationship between the criterion component and a given explainer component when the effect of the other explainer compo- nents is held constant. For example, the fact that the coefficient of ”process sophistication" is positive indicates that, the more sophisti- cated is a company's process, the more its accounting system size in- creases in terms of infOrmation output when the effect of "jOb structure complexity," "direct supervision," and "personnel quality-low level" are held constant. A negative relationship would be indicated by a negative coefficient and can be interpreted as follows: The greater the amount of the explainer variable, the less tends to be the amount of the criterion variable when the effects of other explainer variables are held constant. 1This coefficient is negative because the jOb structure complex- ity component has negative direction, as interpreted below, on page 243. This means that, as the component score increases, the amount of jOb structure complexity decreases. The relationship between the "variable" jOb structure complexity and accounting size-information output is thus positive. In subsequent presentations of regression coefficients, the coefficient signs are adjusted for the direction of the criterion and explainer components. When the criterion component has negative direc- tion, all coefficient signs for that regression are reversed. When any explainer component has negative direction, its coefficient sign is re- versed. See below, page 239, for a more extensive discussion of the direction of components. 98 The Stepwise Procedure For two reasons, the number of the nineteen explainer (organiza- tional) components allowed to explain each criterion (accounting system) component in a multiple regression was restricted. It is generally rec- ommended.in statistical procedures that the sample size be at least twice the number of variables. Since the sample size is eighteen, the number of variables in any multiple regression should be no more than nine and the number of explainers should be no more than eight. Even eight ex- plainers of each accounting component was considered too many since it would be difficult to conceptualize the relationship of that many ex- plainers to a criterion variable. The number of explainers was re- stricted generally to about five. Which of the nineteen potential explainers should be used to explain each accounting component? It seems logical that only those explainers most associated with the different accounting criterion com- ponents should be included in the respective regressions. .A systematic procedure is needed to select those ”most associated" explainers. That procedure is stepwise regression. The SPSS stepwise regression proce- dure used in this study sequentially adds explainer variables that con- tribute most to improving the prediction scores of the criterion compo- nent until it determines that further variables do not contribute significantly to improving the prediction scores. The four explainer components that were used to estimate "ac- counting size-information outpu " were selected by the stepwise regres- sion procedure. In the first step, the procedure examined all the simple correlations of the nineteen potential explainer components with 99 "accounting size-infOrmation output” and selected "process sophistica- tion" as having the highest correlation. It then perfbrmed an F test1 of the significance of "process sophistication" as a predictor of "ac- counting size-information output" and found that "process sophistica- tion" was a significant predictor. The procedure calculated the regres- sion of the criterion "accounting size-information output" on the single explainer "process sophistication." An R square of 62 percent was cal- culated for this regression. Next the procedure examined the partial correlations of "account- ing size-information output" with each of the eighteen explainers not in the regression (all but "process sophistication"). The partial corre- lation is a measure of the association of "accounting size-information output" with one of the eighteen explainers after the effect of "process sophistication" has been removed from each. The procedure selects the 1The statistical nature of the F test is beyond the scope of this dissertation. See Draper and Smith, pp. 24-26, 67-69, 169-71. A.mechanical and intuitive discussion of the use of the F test in this dissertation follows. As each potential explainer is considered for ad- dition to the regression, an F statistic is calculated. The calculated F is compared with a critical F value which is set by the researcher. When the calculated F is greater than the critical F, the explainer is admitted to the regression. When the calculated F is less than the critical F, the explainer is not admitted. The critical F value is derived from an F table in.which three items must be stipulated.by the researcher: confidence level, numerator degrees of freedom, and denomi- nator degrees of freedom. The confidence level was set at 90 percent. The numerator degrees of freedom for the addition of a single variable is always set at one. -The denominator degrees of freedom is calculated as the sample size (eighteen) minus the number of variables in the re- gression after the admission of the explainer (estimated at five) minus one. The denominator degrees of freedom was thus eighteen minus five minus one, equals twelve. The critical F value fer the stepwise regres- sions was determined to be 3.0 from an F table at the 90 percent confi- dence level, with numerator degrees of freedom of one and denominator degrees of freedom of twelve. 100 explainer with the highest partial correlation which is "job structure complexity" and performs an F test of the significance of "job struc- ture complexity" to the regression. Almost any second variable will increase the accuracy of the predicted scores, and thus the R square, since two explainers are better than one. The F test determines if the increase in R square is more than would be expected from the addition of a worthless variable. "Job structure complexity" was determined by the F test to be significant, and a new regression was calculated of the criterion "accounting size-information outpu " on the explainers "proc- ess sophistication" and "jOb structure complexity." This regression had an R square of 73 percent, an increase of 11 percent over the regression with just "process sophistication." The final two explainer components, "personnel quality-low level" and "direct supervision," were added to the regression in two subsequent steps similar to those already de- scribed. Twelve other stepwise regression procedures similar to the "accounting size-infOrmation outpu " regression were calculated with the other twelve accounting system components, respectively, as criterion components, and the nineteen organizational components as potential ex- plainers. Table 5 lists the coefficients of the explainer components that were admitted, respectively, to the thirteen regressions. Explanatory Power and Explainability .A key purpose of this dissertation is to determine whether organizational variables influence the develOpment of the accounting system.and what types of organizational variables have the greatest 101 Table 5 Regression Coefficients for the Stepwise Regressions of the Thirteen Accounting System Components on Nineteen Potential Organizational Explainer Componentsa Accounting System Componentb Organizational Explainer Component A B C D E F G Process Sophistication .............. 0.6 - - 0.2 -0.4 - - Output diversity ............ - 0.5 . . . . . Materials input diversity ... - - - . 0.3 - - Overall Structural Cbmplexity Size ........................ . - - -0.8 0.3 - - JOb structure complexity .... 0.4 - - - . - - Geographical dispersion ..... - - - - - o o Divisional differentiation .. - - - 0.6 - - - Divisional specialization ... - - - - - -0.4 - Mechanization-general ....... ' - - - . - - IMechanization-computers ..... . ~ 0.5 - - - . Cbntrol system Direct supervision .......... 0.2 - . 1.0 . - 0.6 Staff support ............... . . - . - . . Authority levels ............ . - - 0.4 - - - Personnel quality-high ...... - - - 0.5 - ~ - Personnel quality-low ....... 0.3 - -0.4 - - - -0.4 Cent. of authority-invest ... - - - . 0.3 - - Centr. of authority-purchase. - - . - - - - Standardization-jObs ........ - -0.5 - - -0.8 . - Standardization-general ..... - . . -0.8 - - - aCoefficient signs are adjusted to the direction of components. The thirteen columns of accounting system components represent different regressions; each column includes the coefficients for the explainer components which were added to the regressions of the accounting system component for the column identified. bAceounting system components: (A) Size-Information Output; (B) Size-Resource Input; (C) JOb Structure Complexity; (D) Geographical Dispersion; (E) Uhit Differentiation-vertical; (F) Unit Differentiation- Horizontal; (G) Authority Levels. 102 Table 5 (Cont'd.) Accounting System ComponentC Organizational Explainer Component H I J K L 1M Process Sophistication .................... 0.7 - - - . - Output diversity ...... .. .......... . . . . . . .Materials input diversity ......... - - -0.5 . - . Overall Structural Cbmplexity Size .............................. - - -0.3 - . - Job structure complexity .......... - 0.4 . - - . Geographical dispersion ........... . - - . . . Divisional differentiation ........ . . - - - . Divisional specialization ......... - - . . . - Mechanization-general ............. - - . . . . Mechanization-computers ........... - - . 0 . 5 . - control System Direct supervision ................ . 0.4 . - . . Staff support ..................... . - - - -0.4 - Authority levels .................. . - . . - . Personnel quality-high ............ - -0.2 - - . . Personnel quality-low ............. - 0.3 0.4 - - . Centr. of authority-invest ........ 0.6 -0.5 - . . - ICentr. of authority-purchase ...... - . . . . . Standardization-jObs .............. - - - . . . Standardization-general ........... -0.4 0.3 - - - - CAccounting System Components: (H) Report Differentiation; (I) Decentralization of.Accounts; (J) Sophistication of Techniques; (K).Mechanization; (L) Personnel Quality-Education; (M) Personnel Quality-General. 103 influence on the accounting system. .A secondary purpose is to deter- mine what aspects of the accounting system.are most influenced by or- ganizational variables and.what aspects are most insulated from the influence of organizational variables. In order to address these purposes, the terms ”explanatory power" and "explainability" must be defined. "Explanatory power" is the ability of a variable or set of variables to contribute to the ex- planation or prediction of another variable or set of variables. .As discussed above on page 97, "process sophistication" is an important variable to the prediction of "accounting size-infOrmation output." It thus has considerable explanatory power. Groups of variables can be determined to have much explanatory power. For example, the feur ex- plainers of "accounting size-infOrmation output," taken together, were determined to have an R square of 87 percent and thus explain 87 percent of what was available to be explained of "accounting size-information output." An R square of 87 percent is considerably higher than the av- erage (about 52 percent)1 fer the thirteen regressions of accounting com- ponents on organizational components, and thus the four explainers have a great deal of explanatory power with respect to "accounting size- infbrmation outpu ." "Explainability" is the ability of a variable or set of variables to be predicted or explained by another variable or set of variables. It thus refers to criterion variables rather than explainer variables. For example, the above-average R square of 87 percent for the regression of 1See below, page 112. 104 "accounting size-information output" indicates that it is very explain- able by organizational variables. A.basic interest of the dissertation is the relative explanatory power of the three levels of explainer variables ("process," "overall structural complexity," and "control system") with respect to the ac- counting system variables. Though the hypotheses on page 57 above did not predict which levels would have greatest explanatory power, the fecus of the dissertation is on ”overall structural complexity" as an explainer of the accounting system. The expectation is that "overall structural complexity" will have a lot of explanatory power. ”Control system" should also have much explanatory power because of the hypothesized interrelationships within the control system. "Process" was included in the basic model mostly as a control, and the expectation is that its explanatory power will be minimal. A more specific interest of the dissertation is the relative explanatory power of the organizational variables within the different levels of explainers. Which of the variables within the respective levels are important representatives of the level in explaining the ac- counting system, and which of the variables are unimportant? The answer to this question can draw attention to important explainers that should be studied more intensively or to poor explainers that might be removed from the basic model. Another specific interest of the dissertation is the relative explainability of the accounting system variables by the organizational variables. Which of the accounting system variables are most related to organizational variables, and which are least related? 105 Description of Frequencies and R Square Increase Methods Explanatory power and explainability are both assessed in this dissertation by two methods: frequencies of coefficients and R square increase. Table 6 accumulates the infOrmation for the frequencies method, and Table 7 accumulates the infOrmation for the R square increase method. Table 6 is an extension of Table 5 (page 101), in that the number of co- efficients is summed and cross summed. Table 7 substitutes for the coef- ficients the R square increase as each component was added to a regres- sion. The R square increases are summed and cross summed. The frequencies method assesses the explanatory power of a given explainer component by the number of the thirteen regressions fer which the explainer component was incorporated in the regression. For example, "process sophistication" was used as an explainer in four of the thirteen regressions. Consequently there is a 4 in the,f column for "process so- phistication" in Table 6. The frequencies method assesses the explanatory power of a level by the number of times explainers within the level were used in the thirteen regressions. For example, "process level" compo- nents were used seven times as explainers in the thirteen regressions, and there is a 7 in the f'column for ”total process." Explainability of an accounting system component is assessed by means of the number of ex- plainer components that were incorporated in its respective regression. For example, "accounting size-information output" was explained by four explainer components. There is a 4 in the "total frequency" row for "accounting size-infonmation output." 106 Table 6 Frequencies of Regression Coefficients of Accounting System Components on Organizational Explainer Componentsa Accounting System Componentb Organizational Explainer Component A B C D E F G H Process Sophistication ........ 0.6 - - 0.2 -0.4 - . 0.7 Output diversity ...... - 0.5 - - - - - - Materials inp. div. ... - ° - 0 0.3 - - - f ..................... l l 0 l 2 0 0 l Explainability f/EfC .. 1.9 1.9 0 1.9 3.7 0 0 1.9 Overall Structural complexity Size .................. - - o -0.8 0.3 o - - Job struct. compl. .... 0.4 - - . - - - - Geog. dispersion ...... - - - . . - - . Divisional diff. ...... - - - 0.6 - - - - Divisional special. ... - - - - - -0.4 - - Mechan.-general ....... - - - - - - - . Mechan.-computers ..... - - 0.5 - - - - - f ..................... l 0 1 2 l 1 0 0 Explainability f/Ef ... 1.4 0 1.4 2.9 1.4 1.4 0 0 control system Direct supervision .... 0.2 - - 1.0 - - 0.6 . Staff support ......... - . - . . - . . Authority levels ...... - - - 0.4 - - - - Personnel-high ........ o o - 0.5 - - - ° Personnel-low ......... 0.3 - -0.4 - - - -0.4 - Cen. auth.-invest ..... - - - - 0.3 - - 0.6 Cen. auth.-purchase ... - - - - - - - - Standard.-jobs ........ - -0.5 - (d) -0.8 - o 0 Standard.-genera1 ..... - - - -0.8 - - - -0.4 f ..................... 2 1 1 4 2 0 2 2 Explainability f/Ef ... 1.2 0.6 0.6 2.5 1.2 0 1.2 1.2 Total f ..................... 4 2 2 7 5 l 2 3 Explainability f/Ef ... 1.4 0.7 0.7 2.5 1.8 0.4 0.7 1.1 aCoefficient signs are adjusted for the direction of components. This table is an extension of Table 5, page 101. bAccounting system components: (A) Size-Infermation Output; (B) Size-Resource Input; (C) Job Structure Complexity; (D) Geographical Dispersion; (E) Unit Differentiation-Vertical; (F) Unit Differentiation- Horizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decen- tralization of Accounts; (J) Sophistication of Techniques; (K) Mechaniz- ation; (L) Personnel Quality-Education; (M) Personnel Quality-General. cEmplainability: Expected frequencies (Ef): process components, 7/3 = 2.33333; total process, 37 X 3/19 = 5.84211; structural complexity 107 Table 6 (Cont'd.) Accounting System Componentb Organizational f f/Efc Explainer Component I J K L ‘M Process Sophistication .............. - - - . . 4 1.7 Output diversity ............ - - - - ~ 1 0.4 ‘Materials inp. div. ......... - -0.5 - . ° 2 0.9 f ........................... 0 l 0 0 0 7 - Explainability f/Ef ......... 0 1.9 0 0 0 - 1.2 Overall Structural complexity Size ........................ - -0.4 - . - 3 2.3 JOb struct. compl. .......... 0.4 . . - - 2 1.6 Geog. dispersion ............ - - - - - 0 0 Divisional diff. ............ - - . . - 1 0.8 Divisional special. ......... - . - - - 1 0.8 Mechan . -genera1 ............. - - - - - 0 0 Mechan . -computers ........... - - 0 . 5 - - 2 1 . 6 f ........................... 1 1 1 0 0 9 - Explainability f/Ef ......... 1.4 1.4 1.4 0 0 - 0 7 Cbntrol system Direct supervision .......... 0.4 - - . - 4 1.7 Staff support ............... - - -0.4 - l 0.4 Authority levels ............ . - - . . l 0.4 Personnel-high .............. -0.2 - - - - 2 0.9 Personnel-low ....... . ........ 0.3 0.4 - - - 5 2.1 Cen. auth.-invest ........... -0.5 - - - - 3 1.3 Cen. auth. -purchase ......... - - . - - 0 0 Standard.-jdbs .............. . - . - . 2 0.9 Standard.-general ........... 0.3 - - - - 3 1.3 f ........................... 5 l 0 l 0 21 . Explainability f/Ef ......... 3.1 0.6 0 0.6 0 - 1.2 Total f ........................... 6 3 l 1 0 37 - Explainability f/Ef ......... 2.1 1.1 0.4 0.4 0 - - components, 9/7 = 1.28571; total structural complexity, 37 X 7/19 = 13.63160; control components, 21/9 = 2.33333; total control, 37 X 9/19 = 17.52630; accounting components, 37/13 = 2.84615; accounting process, 7/13 = 0.53846; accounting structural complexity, 9/13 = 0.69231; ac- counting control, 21/13 = 1.61538. dThough "standard-jobs" was the first component to be added to the regression of accounting geographical dispersion, it was no longer a significant explainer after the other explainers were added (i.e., its calculated F to remove was under the critical F of 3.0). The regression was recalculated without "standard-jobs." The coefficients for that re- calculated regression are shown here. 108 Table 7 R Square Increase of Organizational Components.Admitted to Regressiona ' b Organizational .Accounting System Component Explainer Component A B C D E F G H Process Sophistication ........ 0.62 - - 0.01 0.18 - - 0.30 Output diversity ...... - 0.22 - - - . . - ‘Materials inp. div. ... - ° - - 0.06 - ° - Z ..................... 0.62 0.22 0 0.01 0.24 0 0 0.30 Explainability Z/EZC .. 4.8 1.7 0 0.1 1.9 0 0 2.3 Overall Structural complexity Size .................. - - - 0.05 0.19 - - . JOb struct. compl. .... 0.11 - - - . . - . Geog. dispersion ...... - - . . . . . . Divisional diff. ...... - ~ - 0.06 - - ° - Divisional special. ... - - - - . 0.17 . - Mechan.-general ....... ' - - - - - - . MeChan.-computers ..... - . 0.23 - - . - . Z ..................... 0.11 0 0.23 0.11 0.19 0.17 0 0 Explainability Z/EZ ... 1.1 0 2.3 1.1 1.9 1.7 0 0 - control system Direct supervision .... 0.06 - - 0.20 o - 0.35 - Staff support ......... - . - - - - - . Authority levels ...... - - ~ 0.05 - - ° - Personnel-high ........ - . - 0.04 - - - - Personnel-low ..... .... 0.08 - 0.16 - - - 0.16 - Cen. auth.—invest ..... - - - - 0.05 - . 0.20 Cen. auth. -purchase . . . - - - - d - - - - Standard.-jdbs ........ - 0.26 - 0.39 0.34 - . - Standard.-general ..... - . - 0.16 - - - 0.14 2 ..................... 0.14 0.26 0.16 0.84 0.39 0 0.51 0.34 Explainability Z/EZ ... 0.5 0.9 0.6 2.9 1.3 0 1.8 1.2 Total 2 (R square) .... 0.87 0.48 0.39 0.96 0.82 0.17 0.51 0.64 Total Expl. Z/EZ ...... 1.7 0.9 0.8 1.8 1.6 0.3 1.0 1.2 aThe thirteen columns headed by accounting system components rep- resent the thirteen regressions whose coefficients are listed in Table 5, page 101. The R square increases listed in this table occurred at the step that the given explainer component was added to the regression of the criterion component. Absence of entry indicates that the explainer component was not added to the regression in that column. bAccounting system components: (A) Size-Infermation Output; (B) Size-Resource Input; (C) Job Structure Complexity; (D) Geographical Dispersion; (E) unit Differentiation-vertical; (F) Unit Differentiation- Horizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decen- tralization of Accounts; (J) Sophistication of'Techniques; (K) Mechaniz- ation; (L) Personnel Quality-Education; UM) Personnel Quality-General. 109 Table 7 (Cont'd.) . . b Organizational .Accounting SYStem Component . z Z/EZC Explainer Component I J K L 1M Process Sophistication .............. - . - . . 1.11 2.0 Output diversity ............ - - - - - 0.22 0.4 .Materials inp. div. ......... . 0.27 - - - 0.33 0.6 X ........................... 0 0.27 0 0 0 1.66 - Explainability Z/EZ ......... 0 2.1 0 0 0 - 1.6 Overall Structural Complexity Size ........................ - 0.11 - - o 0.35 1.9 Job struct. compl. .......... 0.10 - - - - 0.21 1.1 Geog. dispersion ............ - . . - - 0 0 Divisional diff. ............ - - - - - 0.06 0.3 Divisional special. ......... - - - - o 0.17 0.9 Mechan. -general ............. - - - - . 0 0 Mechan. -computers ........... - - 0 . 30 - - 0 . 53 2 . 8 ........................... 0.10 0.11 0.30 0 0 1.32 - Explainability Z/EZ ......... 1.0 1.1 3.0 0 0 - 0. control system Direct supervision .......... 0.09 . - ° - 0.70 1.7 Staff support ............... ~ - - 0.18 - 0.18 0.4 Authority levels ............ . - - - . 0.05 0.1 Personnel-high .............. 0.05 - - - - 0.09 0.2 Personnel-low ............... 0.09 0.15 - ° - 0.64 1.5 Cen. auth.-invest ........... 0.16 - - - - 0.41 1.0 Cen . auth . -purchase ......... . - . . - 0 0 Standard.-j0bs .............. - - ° - - 0.99 2.4 Standard.-general ........... 0.41 . - . - 0.71 1.7 2 ........................... 0.80 0.15 0 0.18 0 3.77 - Explainability Z/EZ ......... 2.8 0.5 0 0.6 0 - 1.2 Total 2 (R square) .......... 0.90 0.53 0.30 0.18 0 6.75 ° Total Expl. Z/EZ ............ 1.7 1.0 0.6 0.3 0 - - CEmplainability. Expected sums (E2): process components, 1.66/3 =0 55333, total process, 6. 75 X 3/l9= 1. 06579; structural complexity components, 1.32/7= 0.18857; total structural complexity, 6. 75 X 7/l9= 2. 48684; control components, 3. 77/9= 0.41889; total control, 6. 75 X 9/19 = 3.19737; accounting components, 6.75/13 = 0.51923; accounting process, 1.66/13 = 0.12769; accounting structural complexity, 1.32/l3 = 0.10154; accounting control, 3.77/l3 = 0.29000. 2 = column or row sum. dThough "standard-jobs" was the first component added to the regression of accounting geographical dispersion, it was no longer a significant explainer after the other explainers were added (i.e., its calculated F to remove was under the critical F of 3.0). An R square increase is shown here because "standard-jobs" increased R square at the step entered. However, the regression was recalculated without "stan- dard-jObs." Coefficients for that regression are shown in Table 6. 110 In order that a given frequency may be evaluated as to whether it indicates high or low explanatory power or explainability, it must be compared with an expected frequency. For an explainer component within a level, the expected frequency is the average frequency of coefficients for the variables within the level. For example, since process components were used seven times as explainers and there are three process components, the expected frequency for each process com- ponent is 7 divided.by 3 equals 2.3. For a level, the expected frequency is the proportion of the total number of coefficients that would be ex- pected based on the number of explainer components in the level compared to the total number of explainer components. For example, the explain- ers are used a total of thirty-seven times in the thirteen regressions. Since explainer components might be expected to be used the same number of times, the proportion of the thirty-seven components that is expected fer "process" is based on the proportionate number of "process" explain- ers. Three of the nineteen explainer components are "process" components. Consequently the expected frequency of components used as explainers for "process" is: 3_2process" components 19 total explainer components 37 total frequency expected frequency total "process" = 5.8 For an accounting component, the expected frequency is the average num- ber of explainer components per regression. For example, explainer components were used a total of'thirty-seven times in the thirteen 111 regressions. The average number of explainer components per regression is 37 divided by 13 equals 2.8. In the right column and the bottom row of Table 6, the actual frequencies are divided by the expected frequencies to obtain measures respectively of explanatory power and explainability. For example, the actual frequency for "process sophistication,” 4, is divided by the ex- pected frequency, 2.3, to obtain a measure of the explanatory power of "process sophistication," 1.7. "Process sophistication" was used as an explainer 70 percent more than might be expected. The total frequency for "process," 7, is divided by the expected frequency for "process," 5.8, to Obtain a measure of the explanatory power of "process," 1.2. "Process" components, taken together, are used as explainers 20 percent more than.might be expected. The total frequency for "accounting size- infOrmation output," 4, is divided by the expected frequency, 2.8, to Obtain a measure of the explainability of "accounting size-information output," 1.4. "Accounting size-information output" is 40 percent more explainable than might be expected. The R square increase method assesses the explanatory power of a given explainer component by the total of the R square increases fer each of the thirteen regressions into which the component was incor- porated. For example, the 2 (meaning summation) column fer "process sophistication," 1.11, is just the total of the R square increases for the four regressions into which "process sophistication" was incorporated as an explainer. The R square increase method assesses the explanatory power of a level by the sum of all the R square increases of all the explainer components in the level that were incorporated in any of the 112 thirteen regressions. For example, the 2 column for total "process,” 1.66, is the total of the R square increases for all the "process" com- ponents that were incorporated into any of the thirteen regressions. Explainability of an accounting component is assessed by means of the total R square for the regression of that component on the organizational components. For example, the X row for "accounting size-information output" (whose explainability was discussed above on page 103) is 0.87, which is the total of the R square increases as each explainer was added to the regression. By definition, the total of the R square increases is the total R square for the regression. The total R squares for the thirteen regressions can be compared with their average, about 52 per- cent. In order that a given summation (2) may be evaluated as to whether it indicates high or low explanatory power or explainability, it must be compared with an expected summation (E2). The expected sum- mations are modeled after the expected frequencies for the frequencies method. For an explainer component within a level, the expected summa- tion is equal to the total sunmation for the level divided by the number of components in the level. For individual "process" components, the expected summation is equal to the total summation for "process," 1.66, divided by the number of components, 3, equals 0.55. For a level, the expected summation is the proportion of the total summation for all re- gressions that is expected based on the number of explainer components in the level compared to the total number of explainer components. For example, the expected summation for the "process" level is: 113 3 "process" components 19 total explainer components 6’75 total summation expected summation total "process" = 1.07 For an accounting component, the expected summation is equal to the av- erage R square for the thirteen regressions, which is 6.75 divided by 13 equals 0.52. Like the frequencies method, the actual summations are divided by the expected summations to Obtain measures of explanatory power and explainability. For example, the actual summation for "process sophisti- cation," 1.11, is divided by the expected summation, 0.55, to Obtain a measure of the explanatory power of "process sophistication," 2.0. "Process sophistication" has twice as much explanatory power in terms of R square increase than might be expected. The total summation for the "process" level, 1.66, is divided by the expected summation, 1.07, to obtain a measure of the explanatory power of the "process" level, 1.6. The "process" level has 60 percent more explanatory power than might be expected in terms of R square increase. The total R square for "accounting size-information output," 0.87, is divided by the expected total R square for accounting components, 0.52, to Obtain a measure of the explainability of "accounting size-infOrmation output," 1.7. "Ac- counting size-information outpu " is 70 percent more explainable than might be expected. 114 Comparison of Frequencies and R Square Increase Methods .A key advantage of both the frequencies and the R square increase methods is that they permit the calculation of the explanatory power of individual explainer components and levels of explainer components in terms of proportions of the total aspects of the accounting system.that are explained. The frequencies method does this by means of proportions of the total number of times explainer components were used for the thir- teen regressions. For example, "process sophistication" was used four times and thus explains 4 divided.by 37 equals 11 percent of the ex- plained aspect of the accounting system. The "process" level explains 7 divided by 37 equals 19 percent of the explained aspect of the ac- counting system. The R square increase method permits the calculation of the proportional explanation of the total of the R squares for the thirteen regressions by individual explainer components and levels of explainer components. In Table 7, the R square increases of individual regres- sions (the columns under the accounting system components) must, by definition, add up to the total R square for the regression. For ex- ample, consider the regression of "accounting size-infOrmation output": 0.62 + 0.11 + 0.06 + 0.08 = 0.87 R square increases total R square The total R squares and the R square changes for individual explainers and levels of explainers can be cross added. In Table 7, the total R 115 squares cross add to 6.75. The R square increases for total "process" cross add to 1.66. The "process" level accounts for 1.66 divided by 6.75 equals 25 percent of the influence of explainer components on the predictions of accounting components. "Process sophistication" accounts fer 1.11 divided by 6.75 equals 16 percent of the influence of explainer components on the predictions of accounting components. Then the pro- portional influence of an individual explainer or a level of explainers is just the summation of the R square changes divided by the total of the R squares. The chief difference between the two approaches is that the frequencies approach ignores the strength of a particular association. .A 1.0 coefficient counts the same as a 0.1 coefficient. The R square increase approach considers the strength of the associations but is biased toward explainers which are added in the early steps of the step- ‘wise regressions. In other words, early-added explainers are credited with too little. The reason for this bias is that the explainers in a particular regression are almost always correlated with one another to a certain extent. The first explainer added is credited with the full effect it has on the prediction in terms of R square increase. The second explainer added (assumed to be correlated with the first) is credited only with its unique contribution to the prediction in terms of R square increase, apart from the contribution of the first explainer. The part of the second explainer that is common to the first explainer has already been credited to the first explainer. The later-added variables are credited with less and less R square increase partly because the early-added variables have already accounted for muCh of 116 the commonality between the explainers. The early-added variables deserve to be credited with more R square increase than later-added variables. Otherwise they would not have been added early. But, as indicated above, they are credited with too much R square increase. Both methods have deficiencies, but their deficiencies are different for the two methods. Consequently, in the analysis of the research findings in Chapter 4, the measurements fer each method are compared with one another in assessing the explanatory power of explainer components and levels and the explainability of accounting system components. consistency with the Hypotheses In addition to determining the strength of influence of organiza- tional variables on the development of the accounting system (explanatory power as discussed in the previous section), a key purpose of the disser- tation is determining the direction of influence, positive or negative, of organizational variables on the development of the accounting system. WhiCh organizational variables and groups of organizational variables lead to the further development of the accounting system.(positive in- fluence) and which tend to suppress the development of the accounting system (negative influence)? Frequencies Method In order to address this purpose, consistency with the hypothe- ses must be defined. Defining consistency will be easier if Table 8 is presented and partially interpreted at this point. Table 8 contains all 117 Table 8 Frequencies of Coefficients Consistent with Hypothesesa Accounting System Componentb Organizational Explainer Component A B C D E F G H Process Sophistication ........ 0.6 - - 0.2 -0.4* . . 0.7 Output diversity ...... - 0.5 - . - . . . Materials inp. div. ... . . . . 0,3 . . . Number consistent ..... l l - 1 1 - ° 1 Proportion consistent . l l - l 0.50 . - 1 Overall Structural complexity Si 28 ...... o ........... ° ° ° - 0 . 8* O . 3 0 0 o JOb struct. compl. .... 0.4 - . . . . . . Geog . dispersion ...... ~ . . . . - . . Divisional diff. ...... . - . 0.6 . . . . Divisional special. ... - - - . . -0.4* . . Mechan.-general ....... - . . . . . . . Mechan.-computers ..... - - 0.5 . . . . . Number consistent ..... l - 1 1 1 0 . - Proportion consistent . l - l 0 50 l 0 - ° control system Direct supervision .... 0.2* . . 1.0* . - 0.6* - Staff support ......... - - - . . . . . Authority levels ...... - - . 0.4* . . . . Personnel-high ........ . . - 0.5* . . . . Personnel-low ......... 0.3* . -0.4 - . - -0.4 . Cen. auth.-invest ..... - - ° - 0.3* - o 0.6* Cen. auth . -purchase . . . - - . . . . . . Standard.-jobs ........ - -0.5 . . -0.8 . . . Standard.-general ..... - - - -0.8 - - - -0.4 Number consistent ..... 0 1 l 1 1 - 1 1 Proportion consistent . 0 l 1 0.25 0.50 ° 0.50 0.50 Total Number consistent ..... 2 2 2 3 3 0 1 2 Proportion consistent . 0.50 l l 0.43 0.60 0 0.50 0.67 aCoefficient signs are adjusted for the direction of components. This table is an extension of Table 5, page 101. Asterisk (*) = coeffi- cient has a sign which is inconsistent with one of the hypotheses. 118 Table 8 (Cont'd.) . b - Organizational .Accountlng System Component £22; :3: Expla1ner Component 1 J K L M sist . Con . Process Sophistication .............. - - - - - 3 0.75 Output diversity ............ - - - ° - l 1 Materials inp. div. ......... - 0.5* - - ° 1 0.50 Number consistent ........... ° 0 ° - ° 5 - Proportion consistent ....... - 0 ~ - - - 0.71 Overall Structural Complexity Size ........................ - -0.4* - - ~ 1 0.33 JOb struct. compl. .......... 0.4 - - - - 2 1 Geog. dispersion ............ ° - - - ° ° ° Divisional diff. ............ - - ~ - - 1 1 Divisional special. ......... - - . - . 0 0 Mechan.-general ............. - - - - . - - Mechan.-computers ........... - - 0.5 - - 2 1 Number consistent ........... l 0 1 ° - 6 - Proportion consistent ...... . 1 0 1 - - . 0.67 control system Direct supervision .......... 0.4* - - - - 0 0 Staff support ............... - - - -0.4 . 1 1 .Authority levels ............ - - - - - 0 0 Personnel-high .............. -0.2 - . - - l 0.50 Personnel-low ............... 0.3* 0.4* - - - 2 0.40 Cen. auth.-invest ........... -0.5 - - . - 1 0.33 Cen. auth.-purchase ......... - ° - - - - - Standard.-jobs .............. - - - - - 2 1 Standard.-genera1 ........... 0.3* - - - - 2 0.67 Number consistent ........... 2 0 - 1- ° 9 - Proportion consistent ....... 0.40 0 . l - . 0.43 Total Number consistent ........... 3 0 l 1 - 20 . Proportion consistent ....... 0.50 0 l l . - 0.54 bAccounting system components: (A) Size-Infermation Output; (B) Size-Resource Input; (C) Job Structure Complexity; (D) Geographical Dispersion; (E) Unit Differentiation-vertical; (F) Unit Differentiation- Horizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decen- tralization of Accounts; (J) Sophistication of Techniques; (K) Mechaniz- ation; (L) Personnel Quality-Education; (M) Personnel Quality-General. 119 of the regression coefficients for the regressions of the accounting system components on the organizational components which were presented in Table 5 (page 101). Some additional information is interspersed through the table. Note that some of the coefficients are singled out by an asterisk (*) which indicates that the sign of the coefficient is inconsistent with one of the hypotheses. The numbers of consistent coefficients (ones without an asterisk) are summed and cross summed similar to the way the total number of coefficients was summed and cross summed in Table 6 (page 106). Then the numbers of consistent coefficients in the respective groupings are divided by the total num- bers of coefficients to get the proportion consistent. The three hypotheses listed above on page 57 postulate some relationships between levels of organizational variables and the ac- counting system. For instance, hypothesis one suggests that structur- ally complex organizations tend to have more fully developed accounting systems. If a single measure of structural complexity of the eighteen companies and a single measure of accounting system development were available, testing this hypothesis would.be simple: a positive rela- tionship would be looked for as an indication of the validity of the hypothesis. In other words, the companies with high scores on "struc- tural complexity” should tend to have high scores on "accounting system development" and vice versa. UnfOrtunately, single measures of either structural complexity or accounting system development have not been developed, and it is necessary to work with multiple measures. In this dissertation, there are seven structural complexity and thirteen accounting system variables. 120 There are thus 7 times 13 equals 91 potential relationships between structural complexity variables and accounting system variables. These ninety-one potential relationships are represented by the ninety-one spaces in Table 8 in the thirteen "Accounting System'Variable" columns of the "Overall Structural Complexity" section. Not all of these rela- tionships were found to exist. Those spaces with coefficients indicate relationships that were found. Spaces containing single dots indicate relationships that were not found. How will the relationships that are found be evaluated as to the extent they confirm.hypothesis one? They are evaluated by their direc- tion, positive or negative. Positive direction means that when one var- iable has a high value the other variable tends to have a high value. Negative direction.means that when one variable has a high value the other variable tends to have a low value. Positive-direction relation- ships tend to confirm hypothesis one, and negative-direction relation- ships tend to disconfirm hypothesis one, given the following conditions are true: 1. The structural complexity variables are valid aspects of structural complexity, and high scores on them constitute greater structural complexity of the company, other things being equal. 2. The accounting system variables are valid aspects of the development of the accounting system, and high scores on them constitute greater development of the accounting sys- tem. Therefore, in order to confirm hypothesis one, many of the re- lationships found between structural variables and accounting system variables should be positive and few should be negative. In general, confirmation of one of the three hypotheses depends upon the extent that 121 the relationships found between the variables of the two levels have the direction, positive or negative, which is indicated by the respective hypothesis. A measure of the degree of confirmation of a given hypoth- esis is the proportion of the relationships found.which have the direc- tion, positive or negative, that is suggested by the hypothesis. .As discussed above on page 120, relationships are indicated.by regression coefficients. Thus the measure of confirmation is the proportion of the regression coefficients of accounting system variables with the variables in one of the explainer levels which have the sign that is suggested.by the respective hypothesis. For hypothesis one, the measure of confirmation is the propor- tion of the regression coefficients of accounting system components on structural complexity components which are positive. Table 8 indicates that there are six positive coefficients within the section, indicated by the absence of an asterisk, out of nine. Therefore, the proportion of consistent coefficients is: 6 consistent structural complexity coefficients 9 total structural complexity coefficients proportion consistent structural complexity coefficients = 0.67. For hypothesis two, postulating the inverse relationship between the stage of development of the accounting system.and other control systems, the measure of confirmation is the proportion of the regression coef- ficients of accounting system components on control system components 122 which are negative. Table 8 indicates that nine of the twenty-one coefficients are negative, and thus the proportion consistent is 9 divided.by 21 equals 0.43. For hypothesis three, postulating the posi- tive relationship between "process" and accounting system development, the measure of confirmation is the proportion of the regression coeffi- cients of accounting system components on "process" components which are positive. Table 8 indicates that five of the seven coefficients are consistent, and thus the proportion consistent is 0.71. consistency of'Individual components Testing the hypotheses thus involves determining if the relation- ships between the accounting system level and the three organizational levels tend to have prevailing directions, positive or negative. Though testing the hypotheses is the major goal of the dissertation, it is de- sirable to have more specific information on the effects of each of the organizational components on the accounting system and the response of individual accounting system components to the organizational variables. The explanatory power of the organizational variables and the explain- ability of the accounting variables was discussed in the prior section (above, pages 100-116). In this section, techniques are developed for assessing the direction of influence, positive or negative, of'individ- ual explainer variables and the direction of increase of accounting system variables in response to changes in organizational variables. Consistency with the hypotheses must be defined separately for accounting variables and fer organizational variables. For an organiza- tional variable (explainer), consistency with a hypothesis is the extent 123 that any relationships found between the organizational variable and the accounting system variables have the direction, positive or negative, that is suggested by the hypothesis that applies to the level in which the explainer is located. Consistency with a hypothesis is measured by the proportion of coefficients for the regressions into which the explainer was incorporated that have the appropriate sign. For exam- ple, the consistency of ”process sophistication" with hypothesis three (the only hypothesis applicable to "process" variables) is the extent that any relationships between "process sophistication" and accounting system variables are positive. Table 8 indicates that three out of the feur coefficients fer the regressions into whiCh "process sophistica- tion" was incorporated have positive signs. Consequently, the consist- ency of "process sophistication," measured.by proportion of coefficients, is 0.75. For an accounting variable (criterion), consistency with the hypotheses is the extent that any relationships found between the ac- counting variable and organizational variables have the direction, positive or negative, that is suggested by the hypotheses that apply to the levels in which the respective explainers are located. Con- sistency of an accounting component is measured by the proportion of the coefficients of the explainers used to explain the accounting component which have the appropriate sign. For example, the consistency of "ac- counting size-information output" is the extent that the coefficients of any "process" explainers used to explain "accounting size-infOrmation output" have positive signs per hypothesis three, the coefficients of any structural complexity component have positive signs per hypothesis 124 one, and the coefficients of any control system components have negative signs per hypothesis two. Table 8 indicates that accounting size has feur explainers: "process sophistication," whose positive coefficient is consistent with hypothesis three; "jOb structure complexity," whose positive coefficient is consistent with hypothesis one; and "direct supervision" and "personnel quality-low level," both of whose posi- tive coefficients are inconsistent with hypothesis two. Thus the consistency of "accounting size-information output" is 2 consistent coefficients divided by 4 total coefficients equals 0.50. R Square Increase Method Just like the frequencies method for explanatory power, the frequencies method for consistency with the hypotheses ignores the strength of particular associations. An inconsistent 0.1 coefficient counts the same as an inconsistent 1.0 coefficient. This defect is overcome by the R square increase method for consistency with the hy- potheses. Table 9 includes the same R square increases as Table 7 (page 108). The R square increases are tagged with an asterisk when- ever the coefficient for the respective explainer has a sign which is inconsistent with one of the hypotheses, as indicated by an asterisk in Table 8. The R square increases which are consistent are summed and cross summed just as the R square increases were summed and cross summed in Table 7. The consistent R square increase sums (2C) are divided by the total R square increase sums (Z) to Obtain the propor- tion of R square increase consistent with the hypotheses for individual 125 Table 9 Proportion of R Square Consistent with the Hypothesesa Accounting System Componentb Organizational Expla1ner Component A B C D E F G H Process Sophistication ........ 0.62 - - 0.01 0.18* - - 0.30 Output diversity ...... - 0.22 ° - ¢ - - - .Materials inp. div. ... - - - - 0.06 ° - - 2c .................... 0.62 0.22 - 0.01 0.06 . - 0.30 Consistency ZC/Z ...... l 1 ° 1 0.25 ° - 1 Overall Structural complexity Size .................. - - . 0.05* 0.19 - - - JOb struct. compl. .... 0.11 - - ‘- - - - - Geog. dispersion ...... . . . - . . . - Divisional diff. ...... - - - 0.06 - - - - Divisional special. ... - - - - . 0.l7* - . .Mechan.-general ....... - - . - - . - - iMechan. -computers ..... ~ - 0.23 - - - - - 2c .................... 0.11 - 0.23 0.06 0.19 0 . - Consistency Zc/X ...... l - l 0.55 l 0 - - Control system Direct supervision .... 0.06* - ° 0.20* - - 0.35* - Staff support ......... . . - . . - . . Authority levels ...... - - - 0.05* - - - - Personnel-high ........ . - . 0.04* - - - - Personnel-low ......... 0.08* - 0.16 - ° - 0.16 ' Cen. auth.-invest ..... - - . . 0.05* - - 0.20* Cen. auth.-purchase ... ° ° - - - - - - Standard.-jObs ........ . 0.26 - 0.39 0.34 . - . Standard. -genera1 ..... - - - 0.16 - - - 0.14 Zc .................... 0 0.26 0.16 0.55 0.34 - 0.16 0.14 Consistency Ec/E ...... 0 1 1 0.65 0.87 - 0.31 0.41 Total 2c .................... 0.73 0.48 0.39 0.62 0.59 0 0.16 0.44 Consistency ZC/Z ..... . 0.84 1 l 0.65 0.72 0 0.31 0.69 3This table is a modification of Table 7, page 108; the R.square increases are the same, but the summations are different. = the explainer component whose addition to the regression resulted in this R square increase (per Table 7) has a sign which is inconsistent with one of the hypotheses (per Table 8). creases consistent with hypothese S. 2c = sum of R square in- Z = sum of R square increases. Asterisk (*) 126 Table 9 (Cont'd.) Accounting System Componentb Organizational Z /X Expla1ner Component I J K L 1M C' C Process Sophistication .............. - - - - - 0.93 0 84 Output diversity ............ - - - - - 0.22 1 Materials inp. div. ......... - 0.27* - - - 0.06 0 18 2c .......................... - 0 - - - 1.21 - Consistency ZC/Z ............ - 0 - - - - 0.73 Overall Structural complexity Size ........................ - 0.1l* - - - 0.19 0 54 JOb struct. compl. .......... 0.10 - - - - 0.21 1 Geog. dispersion ............ - - - - - - - Divisional diff. ............ - - - - - 0.06 l Divisional special. ......... - - ' - - 0 0 IMechan.-general ............. - - - - . - - IMechan. -computers ........... - o 0.30 - ° 0.53 1 Q .......................... 0.10 0 0.30 - - 0.99 . Consistency ZC/Z ............ l 0 l - - - 0.75 Con tro l Sys tem Direct Supervision .......... 0.09* - - ° - 0 0 Staff support ............... - . - 0.18 - 0.18 1 Authority levels ............ - - . - - 0 0 Personnel-high .............. 0.05 - - - - 0.05 0.56 Personnel-low ............... 0.09* 0.15* - - - 0.32 0.50 Cen. auth.-invest ........... 0.16 - - - - 0.16 0.39 Cen. auth.-purchase ......... - - ' - - - - Standard.-jObs .............. - - - - - 0.99 1 Standard. -general ........... 0.41* . - - 0 0.30 0.42 Q .......................... 0.21 0 - 0.18 - 2.00 - Consistency ZC/Z ............ 0.26 0 - l - - 0.53 Total Q .......................... 0.31 O 0.30 0.18 - 4.20 - Consistency ZC/Z ............ 0.34 0 l 1 ~ - 0.62 bAccounting system components: (A) Size-Information Output; (B) Size-Resource Input; (C) Job Structure Complexity; (D) Geographical Dispersion; (E) Unit Differentiation-vertical; (F) Unit Differentiation- Horizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decen- tralization of Accounts; (J) Sophistication of Techniques; (K) Mechaniz- ation; (L) Personnel Quality-Education; (M) Personnel Quality-General. 127 explainer components, levels of explainer components, and accounting system components. Though the R square increase method for consistency with the hypotheses does consider the strength of particular associations, it has the same bias toward early-added explainers as the R square increase method for explanatory power (see above, page 115). Like the use of the two methods for explanatory power, the results of the two methods for consistency with the hypotheses will be compared with one another in assessing the consistency of explainer components, levels of ex- plainer components, and accounting system.components with the hypoth- eses . Limitations In this section several restrictions on the applicability of the conclusions of the study are discussed. These restrictions are nec- essitated by the nature of the research design. The restrictions in- clude: generalization of the findings, techniques of organizational measurement, the reliability of infOrmation provided.by the respondents, the feasibility of demonstrating causality, the incompleteness of the basic model, the effects of the small sample size on the research de- sign, and multicollinearity. The most basic restriction is the result of the selective sample. .As was discussed on page 67, no statistical generalization of the find- ings to a population of organizations other than the sample is permis- sible. However, it is natural to want to apply the findings to other 128 organizations. When doing so, it must be kept in mind that particular characteristics of this sample may account for the findings. The techniques of organizational measurement are generally in a very early stage of development. Since there is little standardization of the measurements of various organizational concepts, the results Of different studies often are not comparable.1 Where possible, measure- ments that had been used in other organizational studies are used in this dissertation, but many of the concepts of this dissertation are new and new measurements had to be developed.2 The conclusions of the study are thus subject to the validity of the measurements. .A second measurement prOblem is the reliability of the infOrma- tion provided.by the respondents. .MOst respondents were cooperative and seemed knowledgeable, but it is difficult to know if their responses were inaccurate or biased. Inaccuracy of the responses would prObably lead to the inability to find relationships whiCh do exist. Bias of the responses may lead to the discovery of relationships which do not exist. Obviously, bias is the greater problem. The conclusions of the study are subject to the reliability of the information provided by the re- spondents. All of the hypotheses are stated in a causal way since the basic model concerns how control systems develop in organizations. This cross-sectional field study is not capable of demonstrating any causal 1James L. Price, Handbook of Organizational Measurement (Lexing- ton, Mass.: D. C. Heath and Company, 1972), pp. 1-2. 2See below, pages 239-72. 129 relationships, including the causal aspect of the hypotheses, for two reasons. First, all measurements were taken at a single point in time so there is no time period for changes to occur. Second, even if se- quential measurements had been taken, there are multitudinous variables not incorporated in the analysis which may cause spurious relationships to appear among the variables in the model or Obscure relationships such that they do not appear. Only associations among the variables can be demonstrated. These associations will be examined in this dissertation as to whether they appear to be causal. It is hoped that future longi- tudinal studies may shed.more light on the causal nature of the associ- ations found in this dissertation. The basic model is very likely incomplete in the sense that im- portant explainers of the accounting system.and important measures of the accounting system are left out. The basic model could be incomplete in two ways. First, variables could be missing within the levels. For example, the process level may not include all the important aspects of "process" or the accounting system level may not include all the impor- tant aspects of the accounting system, The second type of incompleteness is that the model may not include other levels of components which might be significant determinants of the accounting system. For example, a level of environmental variables might have been included. .As was men- tioned on page 61 above, an attempt was made to control for environmental factors by sample selection. But environmental factors could influence simultaneously the feur levels of variables in the study, resulting in spurious associations among the variables in those levels. It is hoped that the control over levels of variables not included in the basic model 130 is sufficient and that there are enough significant variables within the levels so that tentative conclusions can be drawn about the relationships among the levels. Though justified by the great volume of measurements collected for each company, the restricted sample size of eighteen had pervasive effects on the way the data were analyzed. First, since the number of variables analyzed, thirty-two, was much greater than the sample size of eighteen, no single statistical procedure was possible to test simulta- neously all the relationships among the variables. This led to the separate statistical procedures (multiple regressions) fer each of the thirteen accounting system variables. It also led to the restriction of the number of the nineteen explainers allowed into the regressions by means of the stepwise multiple regression procedure. The separate regressions for the thirteen accounting system vari- ables means that the interrelations among the accounting system variables are ignored by the analysis.1 However, sums and averages are calculated across separate regressions in order to Obtain measures of explanatory power and consistency with the hypotheses. In interpreting these meas- ures, it is important to keep in mind that they ignore the interrelations of the accounting system components. 1Ideally, such interrelations of criterion accounting system variables should have been taken into account by means of a.procedure such as canonical correlation, which finds the highest correlation between a linear combination of explainers and a linear combination of criterions. In fact, multiple canonical correlation would.have been desirable since there are three levels of explainers. Canonical corre- lation was impossible for the small sample size of eighteen and the large number of variables, thirty-two. 131 The restricted sample size also prevented the validation of the multiple regression equations on a holdout sample of companies. In interpreting the findings, the lack of validation must be considered. .A frequently encountered problem in studies of organizations is the generally strong association among explainer variables, known as multicollinearity. This restricts the ability of the researcher to de- termine the explanatory power of individual explainer variables. Fortu- nately, multicollinearity was not a great problem in this dissertation. The average intercorrelation of the organizational components admitted to the thirteen regressions was only 10 percent.1 The highest fer any regression was only 28 percent. These intercorrelations are indicated in Table 10. 1The low degree of intercorrelation of organizational explainer variables in this dissertation is partly due to stepwise regression. Stepwise regression tends to minimize the intercorrelation of the ex- plainer variables admitted to the regressions. .At each step after the first explainer is admitted, explainers are admitted that explain most of the remaining variance of the criterion variable. Consequently, explainers that are minimally correlated with the explainers already admitted will tend to be selected. These are more likely to have "new infOrmation" about the criterion. 132 Table 10 Intercorrelations of Organizational Explainers Regression Of 0;:2higi- In:::::§:e- Acco$atr1inagblseystem tional Ex- lation of pla1ners Explainers Size-information output .................. 4 0.13 Size-resource input ...................... 2 0.06 Job structure complexity ................. 2 0.01 Geographical dispersion .................. 7 0.26 Unit differentiation-vertical ............ 5 0.16 Unit differentiation-horizontal .......... 1 - Authority levels ......................... 2 0.10 Report differentiation ................... 3 0.28 Decentralization of accounts ............. 6 0.15 Sophistication of techniques ............. 3 0.21 Mechanization ............................ 1 - Personnel quality-education .............. 1 - Personnel quality-general ................ 0 - Average .... ...... . ........ . .............. 0.10 Chapter 4 INTERPRETATION OF THE RESEARCH FINDINGS The emphasis in this chapter is on analysis of the research find- ings. The preceding chapter develOped the research data that are analyzed here, though some refinement and reassembly of the data are done in this chapter. The fellowing chapter integrates the research findings of this chapter. The purpose of this chapter is to examine the relationships between an organization and its accounting system.and to examine the internal relationships among characteristics of its accounting system. The internal relationships are examined in the first section, while the relationships of the accounting system to the overall organization are examined in the second section. The dissertation is concerned with some general questions about these relationships: 1. How important is the overall organization to the determination of Characteristics of its accounting system? 2. Do the hypotheses predict correctly the relationships of the three levels of organizational variables ("process," "overall structural complexity," and "control system") to the account- ing system? 3. Which of the three levels of organizational variables is most important to the determination of characteristics of the ac- counting system and which is least important? 4. Which organizational variables are most important to the de- termination of characteristics of the accounting system and which are least important? 133 134 5. Do the organizational variables used in this dissertation influence the accounting system in the way expected if the hypotheses are true? 6. Which accounting variables are significantly influenced by organizational variables and which are not? 7. How are accounting variables related to one another? The means of addressing these questions in this chapter are to analyze two sets of relationships: the relationships among the thirteen accounting system variables and the relationships between the thirteen accounting system variables and nineteen organizational variables. The first section of the chapter covers the relationships among the thirteen accounting system variables. The second section covers the relationships between the accounting system variables and the organizational variables. In both sections, the thirteen accounting system variables are treated as criterion variables (variables to be explained). In the first sec- tion, these accounting system criterion variables are explained by other accounting system variables. In the second section, the accounting sys- tem criterion variables are explained by organizational variables. The statistical technique that was used to isolate these rela— tionships is stepwise multiple regression, which was discussed in Chap- ter 3 on pages 90-102. This technique finds a subset of explainer variables which best explains each accounting variable. The two ex- plainer sets are the twelve accounting variables other than the one that is being explained (the criterion) and the nineteen organizational vari- ables. Different subsets of accounting and organizational explainer variables were selected.by the multiple regression procedure for differ- ent accounting criterion variables. Thus two sets of regressions are 135 analyzed in this chapter: the set of thirteen regressions of each ac- counting system variable on the other twelve accounting system variables, covered in the first section, and the set of thirteen regressions of each accounting system variable on the nineteen organizational variables, cov- ered in the second section. The multiple regression technique also produces measures of the strength of relationship between each account- ing variable and each of the subset of explainer variables used to explain it. For each of the two sets of regressions, the measures of strength of relationship between individual accounting variables and each explainer variable are summarized by means of the techniques explanatory power and explainability, discussed above on pages 100-116. These techniques pro- duce measures of the explanatory power of each explainer variable (or- ganizational or accounitng) with regard to the respective sets of regressions. Furthermore, measures of the explainability of account- ing system.variables by respectively the set of other accounting system variables and the set of organizational variables are produced. Measures of the explanatory power of the levels of organizational explainer vari- ables ("process," "overall structural complexity," and ”control system") are also produced. The extent that the set of regressions of accounting components on organizational components tend to confirm the three hypotheses stated on page 57 above are assessed by means of the technique "consistency with the hypotheses,” develOped on pages 116-26. Like explanatory power and explainability, this technique summarizes an aspect of the relationships between individual accounting and organizational variables. It produces 136 measures of conformance of the research findings with the hypotheses fer levels of organizational variables ("process," "overall structural complexity," and "control system"), for individual organizational vari- ables, and for individual accounting variables. RELATIONSHIPS WITHIN THE ACCOUNTING SYSTEM The basic model proposed no explicit hypotheses as to rela- tionships among the characteristics of the accounting system. Never- theless, before attempting to explain accounting system characteristics ‘with organization structure characteristics, it is necessary to exam- ine the interrelationships within the accounting system. The same statistical technique, stepwise multiple regression, is used to analyze the relationships among the accounting system com- ponents as is used to analyze the relationships between accounting system components and organizational components. Stepwise multiple regression is explained in detail fer the organizational regressions on pages 90-100. The application of stepwise multiple regression to the accounting system regressions is described here only in a cursory fashion. Each of the thirteen accounting system components was explained in terms of a linear combination of a subset of the other twelve ac- counting components. Thus there are thirteen regressions with the thirteen accounting components, respectively, as criterions. For a given criterion, the subset of explainer components used to explain the 137 criterion was selected by the forward-stepping stepwise multiple re- gression procedure described above on pages 98-100.1 Table 11 shows the coefficients of the stepwise regressions of each accounting system component on all other accounting system com- ponents. The thirteen accounting system components, treated as cri- terion variables, are represented across the top. The potential accounting system component explainers are listed along the left margin. The explainers which were entered into the respective regres- sions have their standardized regression coefficients in the columns fer those regressions. For the purpose of comparison, Table 21 (below, page 300) lists the zero-order Pearson correlations of the accounting components with each other. Classification as Input and Output Components The thirteen accounting system components can be divided into two groups, input and output components. Input components deal with the organizational and physical arrangements necessary to perfbrm the func- tions of the accounting system, production and dissemination of infOrma- tion. Output components deal with the nature of that infOrmation and the places where it is sent. For example, "size-resource input" is an input component having to do with the quantity of human and financial resources applied to the accounting function. "Report differentiation" is an out- put component having to do with the elaboration of different types of 1The critical F value for these regressions was determined to be 3.0, the same critical F value as that for the organizational re- gressions. See above, page 99, fOotnote 1. 138 Table 11 Regression Coefficients for the Stepwise Regressions of Each of the Thirteen Accounting Components on the Other Twelve Accounting Componentsa Ex- Criterionsb plain- ersb A B c D E F G H I J K L M A . . . . . . . 0 9 . . . . B . . . . . . . . . . . . C . 0 5 . . . 0.6 . . . . -0 5 . D . . . 0 6 . . . . . . . . E . . . 0 8 . . -0.4 . -0 6 . . . F . . . . . . . . . . . . G . . . . . . . . . . . . H 0 5 . . . . . . -0.5 . . -0 4 . I 0 3 . . . . . 0 5 . . . . . J . . . . -0.4 0.4 . . . . . . K . . . . . . . . . 0 3 . . L . . -0.6 . . . . -0 6 . . . . M . . . -0 4 . . . . . . . . aCoefficient signs are adjusted for the direction of cemponents. The thirteen columns headed by criterion components represent different regressions. Each column includes the coefficients for the explainer components which were added to the regression of the criterion component that heads the column. bemponents: (A) Size-Infermation Output; (B) Size-Resource Input; (C) Job Structure Complexity; (D) Geographical Dispersion; (E) unit Differentiation-vertical; (F) Unit Differentiation-Horizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decentralization of Accounts; (J) Sophistication of Techniques; (K) Mechanization; (L) Personnel Quality-Education; (M) Personnel Quality-General. 139 accounting reports and the degree to which accounting reports are pre- pared fer different parts of the company. An important aspect of the interrelationships of components within the accounting system is the nature of the associations of the input and output components. These are described.below on pages 152-55. Nine of the thirteen accounting system components are input components, and four are output components. The input components are: l. Size-resource input 2. Job structure complexity ) 3. Geographical dispersion Accounting . . . . . structural 4. un1t d1fferent1at10n-vert1cal i complexity components 5. Unit differentiation-horizontal 6. Authority levels v 7 . Mechanization 8. Personnel-education 9. Personnel-general Five of the nine accounting input components can be identified as ac- counting structural complexity components. These are "job structure complexity," "geographical dispersion," "unit differentiation-vertical," "unit differentiation-horizontal," and "authority levels." These com- ponents have to do with the breakup of the accounting system along various dimensions. The output components are: 1. Size-infOrmation output 2. Report differentiation 140 3. Decentralization of accounts 4. Sophistication of techniques Explanatory Power and Explainability of'Accounting Components Two characteristics may be relevant for determining the relative importance of accounting system components: 1. Explanatory poweru the ability of an accounting component to explain other accounting components 2. Explainability: the ability of an accounting component to be explained by other accounting components Both explanatory power and explainability are measured by two methods——- frequencies of coefficients and R square increase—awhich,were discussed on pages 100-116. Table 12 calculates the total frequencies of coeffi- cients (f) for the accounting components treated as explainers on the right margin and the total frequencies of coefficients (f) fer account- ing components treated as criterions on the bottom margin. Table 13 calculates the total R square increase (2) for accounting components treated as explainers on the right margin and the total R square (2) for accounting components treated as criterions on the bottom margin. .As was explained on pages 110-13, the total frequencies and R square increases are divided by expected frequencies and R square increases to obtain measures of explanatory power and explainability. Thus there are two measures of explanatory power of accounting components and two measures of explainability of accounting components. .All these measures are assembled in Table 14 by the concepts explanatory power and explainability. Keep in mind that average explanatory power or explainability is indicated by a score of 1.0. Scores greater than one 141 Table 12 Frequencies of Regression Coefficients of Accounting Components on Each Othera Ex- Criterionsb plain- f ersb A B c D E F G H I J K L M Phlox A . . . . . . . 0 9 . . . . 1 0 7 B . . . . . . . . . . . . 0 0 0 c - 0.5 - - - 0.6 . . - - -.s . 3 2.1 D . . . 0.6 . . . . . . . . 1 0.7 E - - - o 8 - - - 4 - -.6 - - - 3 2 1 F . . . . . . . . . . . . 0 0 0 G . . . . . . . . . . . . 0 0.0 H 0 5 - - ~ - - - - 5 - - -.4 - 3 2 1 I 0.8 - - - - - 0.5 - - - - - 2 1.4 J . . . . -.4 0,4 . . . . . . z 1 4 K . . . . . . . . . 0.3 . . 1 0 7 L . . -.6 . . . . - 6 . . . . 2 1 4 M . . . - 4 . . . . . . . . 1 0 7 f 2 1 1 2 2 1 2 2 2 2 0 2 o 19 - f/Efc 1.4 0.7 0.7 1.4 1.4 0.7 1.4 1.4 1.4 1.4 o 1.4 o - - 3This table is an extension of Table 10 (page 132). Coefficient signs are adjusted for the direction of components. components: (A) Size-Infermation Output; (B) Size-Resource Input; (C) JOb Structure Complexity; (D) Geographical Dispersion; (E) unit Differentiation-Vertical; (F) unit Differentiation-Herizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decentralization (of.Accounts; (J) Sophistication of Techniques; (K) Mechanization; (L) Personnel Quality-Education; (M) Personnel Quality-General. CExpected frequency (Ef) = 19/13 = 1.46154. 142 Table 13 R Square Increase of Accounting Components Admitted to Regressions on EaCh Othera Ex- Criterionsb 2c plain- Z ___ ersbABCDEFGHIJKLM A o o o o o o o .48 o o o o 0.48 1.2 B o o o o o o o o o o o o 0.00 0.0 C - .22 ° ° ° .19 ° ° ° ° .33 ° 0.74 1.9 D o o o .51 o o o . . o o o 0.51 1.3 E ° ° ° .51 ° ° .15 ° .35 ° ° ° 1.01 2.6 F o o o o o o o o o o o o 0.00 0.0 G o o . o . . . o o o o o 0.00 0.0 H .25 ° ° ° ° ° ° .23 ° ° .13 ' 0.61 1.5 I .48 ° - ° ° ° .20 ° ' ° ° ° 0.68 1.7 J - - - - .17 .16 - ° ° ° ° - 0.33 0.8 K o o o o o o o o o .11 o o 0.11 0.3 L ° ° .33 ° ° ’ ° .20 ° ° ° ° 0.53 1.3 M o o o .12 o o . . . . o o 0.12 0.3 E .73 .22 .33 .63 .68 .16 .39 .35 .71 .46 .00 .46 .00 5.12 ° Z/EZC 1.9 0.6 0.8 1.6 1.7 0.4 1.0 0.9 1.8 1.2 0.0 1.2 0.0 ° ° aThe thirteen columns headed by criterion components represent the thirteen regressions whose coefficients are listed in Table 10 (page 132). The R square increases listed in this table (other than the last two columns and rows) occurred at the step that the given explainer com- ponent was added to the regression of the criterion component. Single dots indicate that the explainer component was not added to the regres- sion in the respective column. bemponents: (A) Size-Infermation Output; (B) Size-Resource Input; (C) Job Structure Complexity; (D) Geographical Dispersion; (E) unit Differentiation-Vertical; (F) unit Differentiation-Herizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decentralization of Accounts; (J) Sophistication of Techniques; (K) Mechanization; (L) Personnel Quality-Education; (M) Personnel Quality-General. CExpected sum (132) = 5.12/13 = 0.39385. 143 Table 14 Explanatory Power and Explainability of Accounting Components with Respect to Other Accounting Components Expfigaziory Explainability Accounting Component Fre- R Fre- R quencya Squareb quencya Squareb Size-information output ......... 0.7 1.2 1.4 1.9 Size-resource input ............. 0.0 0.0 0.7 0.6 Job structure complexity ........ 2.1 1.9 0.7 0.8 Geographical dispersion ......... 0.7 1.3 1.4 1.6 Unit differentiation-vertical ... 2.1 2.6 1.4 1.7 Unit differentiation-horizontal . 0.0 0.0 0.7 0.4 Authority levels ................ 0.0 0.0 1.4 1.0 Report differentiation .......... 2.1 1.5 1.4 0.9 Decentralization of accounts .... 1.4 1.7 1.4 1.8 Sophistication of techniques .... 1.4 0.8 1.4 1.2 Mechanization ................... 0.7 0.3 0.0 0.0 Personnel quality-education ..... 1.4 1.3 1.4 1.2 Personnel quality-general ....... 0.7 0.3 0.0 0.0 I— b aTable 12. Table 13. 144 indicate high explanatory power or explainability. Scores less than one indicate low explanatory power or explainability. Zero scores indicate no explanatory power or explainability. An important initial question that must be asked is, "To what extent are the accounting components, as a group, interrelated?" A rough indication of this is provided by the average R square for the thirteen regressions which is 5.12 divided by 13 equals 0.39.1 The average number of explainers per regression.was 19 divided by 13 equals 1.5.2 Thus an average of 39 percent of the variation in eaCh accounting component was accounted for by a subset of one or two other accounting components. The top three explainers were selected from the explanatory power column in Table 14 by finding the accounting components which have high scores on both frequency and R square. These are: 1. Unit differentiation-vertical 2. JOb structure complexity 3. Report differentiation Only one of these, "unit differentiation-vertical," has good explainabil- ity. The other two, "jOb structure complexity" and "report differentia- tion," have poor to fair explainability. Three components have zero explanatory power. These are: 1. Size-resource input 2. unit differentiation-horizontal 3. Authority levels Ime Table 13. 2From Table 12. 145 These zero-explanatory-power components are very poorly explained by other accounting components (though none has zero explainability). The most explained four accounting components were selected from the explainability column in Table 14 by finding the components which have high scores on both R square and frequency. These are: l. Size-information output 2. Geographical dispersion 3. Unit differentiation-vertical 4. Decentralization of accounts Only "unit differentiation-vertical" and possibly "decentralization of accounts" also have good explanatory power. "Size-infonmation output" and "geographical dispersion" have only fair explanatory power. Two components are not explained by any other accounting component. These are: l. Mechanization 2. Personnel quality-general These zero-explainability components were also poor as explainers of other accounting components. Two components have not been mentioned as having exceptional explanatory power or explainability. These are: l. Sophistication of accounting techniques 2. Personnel quality-education Both have fairly average explanatory power and explainability. Some generalizations can be drawn from these Observations. .Ac- Counting components with low and medium explanatory power have the cor- ‘responding degrees of explainability and vice versa. But accounting 14‘- 146 components with high explanatory power do not necessarily have high explainability, and accounting components with high explainability do not necessarily have high explanatory power. The single exception is "unit differentiation-vertical," which has high explanatory power and high explainability. The chief difference between explainability and explanatory power is that the interrelations of the components are taken into ac- count for explainability since a single multiple regression is involved, while the interrelations of the components are not taken into account for explanatory power since the numbers of coefficients and the R square increases are summarized across regressions. Thus high explainability means a large association with the common variance of the accounting components. On the other hand, high explanatory power means a large amount of unique characteristics since it results from the component being used as an explainer in numerous regressions. Path Analysis Path analysis was used to map out the relationships indicated by the stepwise regressions. The components which were not used to explain any other components were put at one end of the paths and the components which were not explained by any other components were put at the other end of the paths. Then the intermediate components were filled in sequentially. Path analysis places the components which are closely related locationally close to one another and reveals clusters or groups of closely related components which could be treated as dimensions of the accounting system broader than any one accounting system variable. 147 The path analysis mapping developed in this dissertation is shown in Figure 2. The three accounting system components not used to explain any other accounting system components (zero explanatory power) are "size- resource inpu ," "authority levels," and "unit differentiation-horizon- tal.” These three components were placed at the top of Figure 2. The two accounting system components not explained by any other accounting system component (zero explainability) are ”personnel-general" and "mechanization." These two components were placed at the bottom of Figure 2. The chains of relationships indicated by regression coeffi- cients were filled in between the components which do not explain any others and the components which are not explained by any others. First the components which explain the nonexplaining components are plotted. For example, "size-resource inpu " is explained by "job structure com- plexity," as is indicated by the regression coefficient of 0.5 in Table 11. The rest of the relationships are plotted in the same manner. Note that the top component in all linkages is explained by the bottom component, as is indicated by ascending arrows. However, in some cases the regressions indicate both directions of relationships; i.e., two components both explain and are explained by each other. This is indicated.by two-ended arrows on the same level. For example, "unit differentiation-vertical" and "sophistication of techniques" both ex- plain and are explained by each other. As was mentioned above on page 145, components whiCh had poor explanatory power also had.poor explainability, and vice versa. It fellows that the nonexplaining and unexplained components in Figure 2 l [Ii 3 ‘ll 1". . III I i l 148 NONEXPLAINING Size Authority Unit Differ- COMPONENTS: Resource Levels entiation- Input Horizontal \ 1 0. S , x. ____________ - 0.5 [Decentral- ~90. 9-*- Size- } [ization of Information \\Accounts Output : \\ \ / l \ -0.5 0.5 |\o.4 \ \ / \\ Job <—-0.6—v-Personnel--—-0.4 \ Report OUTPUT \ Structure Education I Differen- COMPONENTS \ Complexity I tia ion \\ \ —— ——— \ _\\\ \\ ——————————————— +——~ \\ \ I Geographical«~——-0. 6-—a-Unit Differ~+f— 0.6 —-L-.Sophistica-l I Dispersion entiation- I I tion of | L Vertical l I Techniques l ~\ I l \ l \___A___/ I DIVISIONALIZATION } _________ .2 . 0.3 UNEXPLAINED Personnel- Mechanization COMPONENTS: General Explainer Explained component component Figure 2 Path Analysis of Accounting System Interrelationships 149 have poor explanatory power and explainability. High-explanatory-power components and high-explainability components are located in the center of Figure 2. Output Component Cluster The output components are all in the middle of the structure in Figure 2. In other words, they both explain and are explained by the input components. The input components are thus separated into those which are explained, directly or indirectly, by output components (at the top of Figure 2) and those which explain, directly or indirectly, output components (at the bottom of Figure 2). Three of the four output components form a very tight group: "decentralization of accounts," "size-infOrmation output," and "report differentiation." The fourth output component, "sophistication of tech- niques," is relatively close to the other three, with only "unit differ- entiation-vertical" separating them. This would suggest that the char- acteristics of the output of accounting systems might blend into a single dimension of accounting output. The very strong positive association of "size-information outpu " and "decentralization of accounts" suggests logically that accounting systems tend to grow in size by presenting more information to lower company levels. The positive association of "size-information output" and "report differentiation" suggests that accounting systems also grow by presenting more different types of reports. The negative association of "decentralization of accounts" and "report differentiation" is more difficult to explain. Apparently, companies' accounting systems can 150 develop in.two alternative directions: the preparation of basic cost reports for lower levels of the company and the elaboration of more sophisticated types of reports for the higher levels. Consequently, decentralization of accounts and report differentiation are negatively related to each other since accounting systems which grow in terms of decentralization of accounts have little report differentiation and vice versa . Relationships among Input Components The relationships of the components at the top of Figure 2 go from the nonexplaining components——“size-resource input," "authority levels," and "unit differentiation-horizontal"——to the output component cluster. There are two groups of these components. "Accounting size- resource input" and "authority levels" are closely related by means of "job structure complexity," while "unit differentiation-horizontal" is independently related to the output component cluster. It appears from Figure 2 that companies with larger accounting systems (in terms of resource input) tend to have more complex account- ing jOb structures. Companies with greater numbers of authority levels ‘within the accounting system also tend to have more complex accounting jOb structures. However, the companies with more complex jOb structures tend to have less-educated accounting personnel. This may indicate that, the higher the education level within the accounting system, the more of the total accounting task each employee can handle, and thus the less complex the jOb structure needs to be. This would suggest that educa- tion level and jOb structure complexity are substitutes for one another 151 in accomplishing the accounting task. However, the negative association of "personnel-education" with the output component "report differentia- tion" would suggest that "personnel-education" does not contribute positively to the elaboration of different types of reports. The relationships at the bottom of Figure 2 go from the output components to the components which are not explained by any other com- ponents. There are two groups of these: the accounting divisionalization group and the "mechanization" component. The accounting divisionalization group has to do with the dis- persal of accounting activities down and out to lower-level units (usu- ally divisions) of the company. .As a company grows, the accounting function, originally concentrated in a single unit, may have to be geographically and organizationally dispersed to the dispersed divisions of the company in order to make accounting information more accessible to and controllable by divisional personnel. This dispersal is reflected in the positively related two-component grouping of "unit differentiation- vertical," the development of lower-level (prObably divisional) account- ing departments, and "geographical dispersion," the development of ac- counting departments at different locations (prObably divisions). Such dispersal may entail a lowering of the overall quality of accounting personnel, as is indicated by the negative regression coefficient of "geographical dispersion" on "personnel-general." Smaller-scale divi- sional accounting departments might have to concentrate more on routine data processing, whereas larger, centralized accounting departments, having economies of scale, might be able to retain.more highly qualified specialists. The association of divisionalization of the accounting 152 function with decline in personnel quality is consistent with the neg- ative association of divisionalization with the output components "re- port differentiation" and "sophistication of accounting techniques." In general, the more dispersed is the accounting fUnction, the less complex is the output of the system, at least partly due to a decline in the general quality of personnel. Relationships between Input and Output Components There are six associations between input and output components indicated in Figure 2——three positive and three negative. The positive associations are: Authority 1evels——Decentralization of accounts Unit differentiation-horizontal——Sophistication of techniques Mechanization——Sophistication of techniques The negative associations are: Personnel-educationr—Report differentiation Unit differentiation-vertical——Report differentiation Unit differentiation-vertical——Sophistication of techniques Though no explicit hypotheses were developed as to the interrela- tions of accounting system components, it seems logical that, as the quantity and complexity of inputs to the accounting system.becomes greater, the quantity and complexity of the outputs from the system should also become greater. Just as overall structural complexity of the company provides for the production of products, the structural complexity (as measured by input components) of the accounting system provides for the production of reports (as measured.by output components). This surmised relationship suggests that the relationships between input 153 and output components should be positive if the respective components measure increasing elaboration of the accounting system. The three positive associations seem to confirm the surmised relationship of input and output components. The positive association of authority levels and decentralization of accounts suggests logically that the development of the account structure down to the lower levels of the company is accompanied by the downward development of the account- ing organization structure. The positive association of "unit differentiation-horizontal" and "sophistication of techniques" also supports the surmised relation- ship. Note that "unit differentiation-horizontal" has to do with the breakup of the central accounting department, as opposed to divisional accounting departments. Thus it probably is associated with unit specialization within the accounting function since multiple units would.probably be assigned different functions. Some of the techniques included in "sophistication of techniques" might even require separate units within the central accounting department. For example, a standard cost system might require a standard-setting department. Consequently, the association between "unit differentiation-horizontal" and "sophisti- cation of techniques" seems highly reasonable. The positive association of "mechanization" and "sophistication of techniques" also supports the surmised relationship. Apparently, mechanization is instituted to free accounting personnel of some of the burdens of routine data processing. This leaves them time to concentrate on implementing more sophisticated accounting techniques. 154 The three negative associations seem to disconfirm the surmised relationship. .AS‘was indicated on page 151, the negative association of "personnel-education" and "report differentiation" is very difficult to explain. Since "personnel-education" is negatively related to the three primary arrangement components-—”size-resource input," "jOb structure complexity," and "authority levels"——and they in turn are related posi- tively to the output component cluster by means of "authority levels" and "decentralization of accounts," perhaps "personnel-education" reflects only the absence of "jOb structure complexity" which, along with the other two primary arrangement components, is instrumental to sophisticated ac- counting output. Thus "personnel-education" would have no unique influ- ence on output components. An alternative explanation is that "person- nel-education" does contribute positively to some aspect of accounting system output not measured in this study, perhaps flexibility. Thus it is possible that accounting system structural complexity, measured.by "jOb structure complexity" and "authority levels," can routinely produce complex accounting reports but educated personnel are required to cope with nonroutine situations; i.e., provision of'problem-solving informa- tion. It is recommended that fature studies measure flexibility as an aspect of accounting system output. As was maintained on page 152, the more dispersed is the ac- counting function in terms of divisionalization, the less complex is the output of the system in terms of report differentiation and sophistica- tion of techniques. These two negative associations would suggest that divisionalization has a negative effect on accounting output, at least in terms of the two output components "report differentiation" and 155 "sophistication of techniques." The question might be asked: 'Mmy disperse the accounting function if it damages output?" The answer is that divisional accounting departments may be more responsive to the needs of divisional operating personnel. Responsiveness was not meas- ured in this study, but it is recommended that it be included in future studies. Alternative Types of’Accounting systems The inconsistent associations between input and output components suggest that all the aspects of accounting output may not develop in the same direction. In other words, one accounting system may emphasize one set of output characteristics while another may emphasize another set. One way of looking at the results of the accounting system re- gressions involves the following assumption. Accounting systems tend to fall on a dimension with the following polar extremes: Polar Types of Accounting Systems I 2 Sophisticated accounting Elementary accounting techniques techniques High-level reports High and low level reports Centralized Divisionalized Departmentalized Uhdepartmentalized Mechanized data Manual data processing processing High-quality personnel Lower-quality personnel Rigid and unresponsive Flexible and responsive to infOrmation needs of to infOrmation needs of lower-level managers lower-level managers 156 .Accounting systems can occur anywhere between the extremes. However, when an accounting system has one of the characteristics of one of the extremes, it tends to have the other characteristics. Centralized accounting systems have a single accounting office for the entire company. This office may have many specialized depart- ments. Divisionalized accounting systems have accounting offices at division headquarters. Centralized accounting systems tend to produce reports only for high-level management, utilize sophisticated account- ing techniques, have high-quality personnel, and have mechanized data processing systems. Divisionalized accounting systems tend to produce reports for both high- and low-level management, utilize elementary accounting techniques, have lower-quality personnel, and utilize manual data processing systems. The final characteristics which distinguish the two types, flexibility and responsiveness, were not measured in this dissertation. It is the opinion of the author that some characteristics suCh as these must work to induce companies to divisionalize (more gen- erally to decentralize) their accounting systems. Otherwise the advan- tages of centralization would prevent decentralization altogether. Support for the association of characteristics illustrated on page 155 comes from a number of associations in Figure 2 (page 148). The negative association of "decentralization of accounts" and "report differentiation" discussed on page 149 suggests that accounting systems either concentrate on producing more reports for managers at lower company levels (as measured by "decentralization of accounts") or they concentrate on producing more sophisticated reports fer the higher levels (as measured by "report differentiation"). 157 What kind of accounting organizational arrangements (input com- ponents) are associated with these two alternative types? Accounting structural complexity (as measured by "authority levels" and "jOb structure complexity") is most associated with production of reports fer lower company levels (as measured by "decentralization of accounts"). As might be expected, the provision of reports to lower company levels requires the development of'more levels in the accounting organization structure. But divisionalization is negatively related to two measures of sophistication.of'high-level output: "report differentiation" and "sophistication of techniques." Why then do companies divisionalize their accounting systems? .A possible reason for this strange finding is discussed in the fellowing paragraph. In this sample of small companies, it is likely that the major influence on the number of accounting authority levels is divisionaliza- tion of the accounting system. In other words, companies with account- ing offices at division headquarters will almost invariably have more, authority levels in their accounting systems than those with only a single accounting office. It may be speculated that divisionalization, along with authority levels, contributes to the production of more re- ports for lower company levels. The imperfections of measurement and research design may have Obscured the (positive) relationship between divisionalization and decentralization of accounts.1 1The two negative associations——-"decentralization of accounts" ‘with "report differentiation," and "unit differentiation-vertical" with "report differentiation"——may be combined to form.a positive association between "decentralization of accounts" and "unit differentiation-vertical." 158 Two other associations in Figure 2 support the association of the characteristics of centralized accounting systems. The positive asso- ciation of "mechanization" and "sophistication of techniques" suggests indirectly that centralized accounting systems which utilize sophisti- cated accounting techniques also tend to have more mechanized accounting data processing systems. The positive association of "unit differenti- ation-horizontal," a measure of the departmentalization of the central accounting office, and "sophistication of techniques" also suggests indirectly that centralized accounting systems whiCh utilize sophisti- cated techniques tend to be departmentalized. RELATIONSHIPS OF THE ACCOUNTING SYSTEM TO THE OVERALL ORGANIZATION The central concern of the dissertation is the relationships between characteristics of an organization and characteristics of its accounting system. The multiple regression technique described in Chapter 3 (pages 90-100) was used to analyze the relationships between thirteen accounting system variables and nineteen organizational vari- ables. There are thirteen regressions, one for each accounting variable. For each regression, a subset of the nineteen organizational variables was used to explain the accounting variable. The regression coefficients fer these subsets are listed in Table 5 (page 101). Two basic purposes of analyzing the relationships between account- ing system variables and organizational variables are to determine the strength and direction, positive or negative, of influence of the organi- zational variables on the accounting system variables. .A corollary pur- pose is to determine the influenceability of accounting system variables 159 by organizational variables. Both strength of influence and influence- ability are measured by two techniques, frequency and R square.1 Both techniques involve, generally speaking, the number of regression coeffi- cients for various categories of relationships between accounting and or- ganizational variables. Strength of influence is evaluated with respect to the entire set of organizational variables, the three levels of or- ganizational variables, and the individual organizational variables. The direction of influence is assessed with respect to three hypotheses (stated above on page 57) which predict the direction, posi- tive or negative, of the relationships of, respectively, the three levels of organizational variables to the accounting system variables. The measures of confermance of the research findings with the hypotheses are called "consistency with the hypotheses" and are of two types, fre- quency and R square.2 Both types involve, generally speaking, the extent that various groupings of regression coefficients have the signs, positive or negative, that are expected if the hypotheses are true. Consistency ‘with the hypotheses is evaluated with respect to the levels of organiza- tional variables and the individual organizational variables. The Extent of’Influence of'an Organization on Its Accounting System The most basic question that can be asked in this study is, "How important are characteristics of an organization to the determination of 1See above, pages 100-116. 2See above, pages 116-27. 160 the characteristics of its accounting system?" With respect to the components and the multiple regression approach used in this disserta- tion, the question becomes, "How much of the variation in the account- ing components scores can be accounted fer by linear combinations of the organizational components?" .As was explained on pages 95-96, R square is a measure of the proportion of the variation in a criterion accounted fer by a linear combination of explainers. The average R square for the thirteen regressions of accounting components on organizational compo- nents is thus a measure of the average ability of the organizational com- ponents to explain accounting components. The average R square is 52 percent.1 It should be noted that the average number of organizational components used as explainers in the thirteen regressions was 2.8. In other words, only three of the nine- teen potential explainers were necessary to explain 52 percent of the variation in the accounting components. The average R square of 52 per- cent might be compared with the average R square for the regressions of the accounting components on other accounting components which was 39 percent.2 It follows that accounting components were much more easily explained by organizational components than they were by eaCh other. In general, it can be said that characteristics of an organization are vital to the determination of characteristics of its accounting system. 1This is calculated as the average of the R squares in the sec- ond to the bottom row of Table 7, page 108. 2See above, page 144. 161 The Influence of’the Explainer Levels on the Accounting system There are three levels of explainer variables in this disserta- tion: process, overall structural complexity, and control system. The process level includes three variables, the overall structural complex- ity level includes seven variables, and the control system level includes nine variables. The concern of this section is the strength and direc- tion of influence of each of these explainer levels on the accounting system level which includes thirteen variables. The strength of influ- ence is measured by explanatory power, and the direction of influence is measured by consistency with the hypotheses. In order to keep perspec- tive, it may be useful while reading this section to refer back to the basic model on page 44 and the regression coefficients for the thirteen regressions of accounting components on organizational components in Table 5 (page 101). In this section, the analysis concentrates on three groupings of coefficients shown as separate sections of Table 5. Table 15 assembles measures of the explanatory power and con- sistency with the three hypotheses of the three levels of explainers. A brief discussion follows of the key aspects of the measures necessary to use them. LMuch of this discussion applies also to subsequent sec- tions. For extensive treatments of the derivation of the measures, see pages 100-116 for explanatory power and pages 116-27 for consistency with the hypotheses. Explanatory power of a level is the ability of the variables within that level to explain accounting variables. More specifically, it is the extent that variables in the level were chosen by the thirteen 162 Table 15 Explanatory Power and Consistency with the Hypotheses of Explainer Levels Explainer Level Item Overall Process Structural Iggfiigi Complex1ty Explanatory power compared to expected3 Frequenciesb .................. 1. 20 o . 7o 1 .20 R square increaseC ............ 1.60 0.50 1.20 Proportion of total explainedd Frequenciesb .................. 0.19 0.24 0.57 R square increaseC ............ 0.25 0.19 0.56 Consistency with hypothesese Frequenciesf .................. 0.71 0.67 0.43 R square increaseg ............ 0.73 0.75 0.53 Number components ................. 3 7 9 aAverage explanatory power, compared to expected, is indicated by a score of 1.0. High explanatory power, compared to expected, is indicated.by scores in excess of 1, while scores less than 1 indicate low explanatory power. Zero explanatory power would be indicated by a score of 0. bTable 6, page 106. CTable 7, page 108. dThe maximum proportion of total explained is 1.0, and.the minimum is 0. ePerfect consistency with the hypotheses is indicated by a score of 1.00. Perfect inconsistency is indicated by a score of 0. fTable 8, page 117. gTable 9, page 125. 163 stepwise regression procedures as explainers of the accounting variables. Explanatory power, compared to expected, includes measures of the extent that the individual explainer variables in the level, on the average, are used more or less as explainers than variables in other levels. Explanatory power, proportion of total explained, includes measures of the proportional influence of the variables in the level on the explana- tion of accounting variables, regardless of the number of variables in the level. Naturally, levels with.more variables will explain.more in terms of proportion of total explained. For example, the three indi- vidual process variables were chosen more often as explainers than vari- ables in the overall structural complexity level, as indicated by the "compared to expected" measures. The nine control system variables, as a group, explained.more of the variability of the accounting variables than the three process variables, as indicated by the "proportion of total explained" measures, because control system.had.more variables. TWO measures of each form of explanatory power, frequencies and R square increase, are presented because no faultless measure could be developed in Chapter 3. The frequencies measures ignore the size of associations indicated by regression coefficients, but the R square increase measures are biased toward explainers which were added early to regressions. The focus of the dissertation has been on the three hypotheses incorporated in the basic model. These predict that the relationships between variables in the three explainer levels and accounting variables ‘will have certain directions, positive or negative. For example, con- trol system variables are predicted to have negative relationships with 164 accounting variables, per hypothesis two (page 57). For the multiple regression research design of this dissertation, these relationships are regression coefficients, and the regression coefficients of account- ing components on control system components are predicted to have nega- tive signs. The two numbers under consistency with the hypotheses for control system are measures of the extent that any regression coefficients of accounting components on control system components are negative. Like those for explanatory power, the frequencies measure of consistency with the hypotheses ignores the size of particular associations, while the R.square increase measure is biased toward early-added explainers. The measurements of consistency with the hypotheses fer process and overall structural complexity strongly support, respectively, hypoth- eses one and three (see page 57).1 The frequencies measures indicate that 71 percent of the coefficients of accounting system components on process components were positive, the direction predicted by hypothesis three, and 67 percent of the coefficients of accounting system components on overall structural complexity components were positive, the direction predicted by hypothesis one. The even larger R square increase measures for process and overall structural complexity indicate that the large 1Since it is not possible to generalize statistically from this selective sample of companies to a larger population of companies, about all that can be said about the confirmation of the hypotheses is that, fer this sample, the preponderance of relationships is in the direction expected. It is also important to note that the rough nature of the measurements of accounting and organizational variables makes high pro- portions, near 1.0, very unlikely. It is the judgment of the author that these proportions are as high as could have been expected given the rough nature of the measurement process. 165 associations were even.more consistent than the small associations. In light of these measurements, the following statements can be made: 1. When companies have high scores on their process variables, they tend to have high scores on their accounting system variables. 2. When companies have high scores on their overall structural complexity variables, they tend to have high scores on their accounting system variables. The consistency with hypothesis two measures fer the control system, 0.43 and 0.53, do not confirm.or disconfirm hypothesis two. Only 43 percent of the coefficients of accounting system components on control system components were negative, as predicted by hypothesis two. But the larger associations, as measured by R square increases, were more consistent (negative) than the smaller associations. This caused the R square increase measure to be greater than 50 percent.1 As might be expected, control system components, as a group, explained a larger proportion of the variation in accounting system components, almost 60 percent, than any other level since there were more control system components. Also, the average control system 1It is interesting to note that, for all three levels, consist- ency with the hypotheses is greater in terms of R square increase than in terms of frequencies. That is to say, explainers that were added early to the stepwise regressions tended to have coefficients with signs that were more consistent with the hypotheses. If the regressions had been more restrictive in admitting explainers (i.e., a critical F value greater than 3.0 had been set), the signs of coefficients would.have been Inore consistent with the hypotheses. Apparently, the first explainers ladmitted to the regressions indicate significant relationships (i.e., high explanatory power and consistency with the hypotheses). Later explainers admitted may indicate more indirect and possibly spurious trelationships between the explainers and one of the accounting vari- ables. Such indirect relationships could have been caused by the simul- ‘taneous effect on the accounting variable and the organizational variable (of a third variable. 166 component explained 20 percent more than the average of all organiza- tional components. The control system level is an important determi- nant of the accounting system. The importance of the control system level is surprising in light of the fact that hypothesis two could not be confirmed. In general, it can be said that the control system level contains variables which have many strong positive and negative associ- ations with accounting system variables. The average component in the overall structural complexity level explained 30 to 50 percent less than the average of all organizational components. Even though there were more than twice as many components, the overall structural complexity level explained.no more of the varia- tion in accounting components than the process level, 20 to 25 percent. The overall structural complexity level, which was originally predicted to be an.important explainer of the accounting system (see page 104), had unexpectedly poor explanatory power. This was surprising in light of the fact that hypothesis one was strongly confirmed. In general, it can be said that the overall structural complexity level has variables which have consistently positive but relatively weak associations with accounting system variables. The process level was included in the basic model primarily as a control and was not expected to have much explanatory power. Yet the average process component explained from 20 to 60 percent more than the average of all organizational components. They were thus the strongest of the explainer components. Though the process level, with only a third of the components, could not explain as much as the control system level, it did explain as much as the overall structural complexity level. 167 Process was the only explainer level which had good explanatory power and was very consistent with its hypothesis (three). In general, it can be said that the process level has variables which have very strong and consistently positive associations with accounting system.variables. The Influence of the Emplainer variables on the Accounting system The purpose of this section is to analyze the strength and direc- tion of the relationships of each organizational variable to the entire set of accounting variables. It addresses two questions: 1. How much does the explainer component contribute to the linear combination estimates of the accounting component scores (explanatory power)? 2. Are the directions of the coefficients of the explainer component consistent with the hypothesis which applies to the explainer level of which the explainer component is a member (consistency with the hypothesis)? In this section, the rows of Table 5 (page 101) are analyzed. For example, the first row in the table includes four coefficients. These are the coefficients of four accounting components on "process sophistication" for the four of the thirteen regressions into which "process sophistication" was admitted as an explainer. These four coefficients (and the associated R square increases) are assembled into measures which indicate whether "process sophistication" is an important explainer of the thirteen accounting system components and whether its coefficients tend to confinm hypothesis three.1 The reader 1See above, pages 100-116 and 116-27, respectively, for dis- cussions of the techniques, explanatory power, and consistency with the hypotheses. «in—w 168 will find it useful to keep the text open to Table 5 as he reads this section. Table 16 accumulates measures of the explanatory power and consistency with the hypotheses of each of the nineteen explainer com- ponents. Note that the explanatory power measures are centered on a score of l, which indicates the component has average explanatory power fer the explainer components within its level. .A score of 0 indicates the component was not added to any of the thirteen regressions and thus has no explanatory power. Scores less than 1 indicate poor explanatory power, while scores in excess of 1 indicate good explanatory power. The consistency measures are proportions. Perfect consistency is indi- cated by a score of 1, while perfect inconsistency is indicated by a score of 0. Once again, the two methods, frequencies and R square increase, are used to form.separate measures of both explanatory power and con- sistency with the hypotheses. The frequencies method ignores the strength of associations indicated by regression coefficients, while the R square increase method is biased toward early-added explainers. In this section, the individual explainer components are ana- lyzed by levels of explainers. The main Objective of the analysis is to separate variables which are good representatives of their explainer levels from variables which are poor representatives of their explainer levels. Good representatives have more explanatory power than other components within the explainer level and have coefficients whose signs are consistent with the hypothesis which applies to the respective ex- plainer level. Poor representatives have poor explanatory power and/or 169 Table 16 Explanatory Power and Consistency with the Hypotheses of Explainer Components Explanatory Consistency with Powera Hypothesesb Explainer Component Fre- R Fre- R quencyC Squared quencye Squaref Process Sophistication ............... 1.7 2.0 0.75 0.84 Output diversity ............. 0.4 0.4 1.00 1.00 .Materials input diversity .... 0.9 0.6 0.50 0.18 Overall structural complexity Size ......................... 2.3 1.9 0.33 0.54 Job structure complexity ..... 1.6 1.1 1.00 1.00 Geographical dispersion ...... 0.0 0.0 ° ° Divisional differentiation ... 0.8 0.3 1.00 1.00 Divisional specialization .... 0.8 0.9 0.00 0.00 Mechanization—general ........ 0.0 0.0 - - Mbchanization-computers ...... 1.6 2.8 1.00 1.00 Cbntrol system Direct supervision ........... 1.7 1.7 0.00 0.00 Staff support ................ 0.4 0.4 1.00 1.00 Authority levels ............. 0.4 0.1 0.00 0.00 Personnel quality-high ....... 0.9 0.2 0.50 0.56 Personnel quality-low ........ 2.1 1.5 0.40 0.50 Centr of authority-invest .... 1.3 1.0 0.33 0.39 Centr of authority-purchase .. 0.0 0.0 - - Standardization—jabs ......... 0.9 2.4 1.00 1.00 Standardization—general ...... 1.3 1.7 0.67 0.42 aAverage explanatory power is indicated by a score of 1.0. High explanatory power is indicated.by scores in excess of 1.0, while low ex- planatory power is indicated by scores less than 1.0. bPerfect consistency with the hypotheses is indicated by a score of 1.00. Perfect inconsistency is indicated by a score of 0.00. Spaces containing a single dot indicate there were no coefficients for this ex- plainer with which consistency with the hypotheses could be calculated. cTable 6, page 106. dTable 7, page 108. eTable 8, page 117. fTable 9, page 125. llll‘lul I ll 1|||l III [I III-II Ill. I III! .II l'llli 1 170 consistency. The direction of their influence on the accounting sys- tem may not be consistent with the level of which they are a member, or they may have little influence on accounting system variables. "Process sophistication" is clearly the most representative of the process level components. It has exceptional explanatory power and is fairly consistent with hypothesis three. "Materials input di- versity" has somewhat lower than average explanatory power and is somewhat inconsistent with hypothesis three. "Output diversity" was admitted to only one of the thirteen regressions, and thus has low explanatory power, though the single coefficient is consistently positive. The lack of significant influence on the accounting sys- tem by the two variables "output diversity" and "materials input di- versity," which were derived from the Pugh et al. studies, is notable. The only variable in this level with significant influence, "process sophistication," was the result of miscellaneous measurements assembled by this author.1 Two of the overall structural complexity components, "jOb struc- ture complexity" and."mechanization-computers," are representative of overall structural complexity in that they have better than average explanatory power and are (perfectly) consistent with hypothesis one. The only other better than average explanatory power component is com- pany size, but strangely it is somewhat inconsistent with hypothesis one. Two out of three of its coefficients are negative, but the coefficient represented by the largest R square increase ("unit differentiation- 1See above, page 53. 171 1 making the R square measure of vertical") is consistently positive, consistency greater than 0.50. The relationships of company size to divisionalization of the accounting system, as represented by accounting "geographical disper- sion" and "unit differentiation-vertical," is particularly interesting. Company size leads to the development of accounting departments at divi- sion headquarters and below. But company size is negatively related to geographical dispersion of the accounting system. It should.be noted that most of the measurements which were incorporated into ac- counting "geographical dispersion" had the effect of overall company geographical dispersion and/or size removed.2 Thus the interpretation of accounting "geographical dispersion" is the additional geographical dispersion of the accounting system.beyond what is expected based on the geographical dispersion and.Size of the company. In contrast, "unit differentiation-vertical" did not have the overall company effect re- moved.3 Therefore, an interpretation of these results is that company size leads to the development of accounting units away from the home office but at a rate less than the rate of increase in company size. Two of the overall structural complexity components, "geo- graphical dispersion" and "mechanization-general," were not used as explainers in any of the thirteen regressions and thus have no explana- tory power. The two divisional components, differentiation and spe- cialization, have poor explanatory power, each having been incorporated 1See Table 9, page 125. 2See below, page 261. ’See below, page 267. 172 in only one of the thirteen regressions. The positive coefficient of accounting "geographical dispersion" on "divisional differentiation" is consistent with hypothesis one though it is explainable by the direct relationship between the two components. Divisions of companies are often at different locations. Only companies with multiple locations provide the opportunity for the accounting fonction to be geographically dispersed. Similarly, the negative coefficient of accounting "unit differentiation-horizontal" on "divisional specialization," though inconsistent with hypothesis one, is also explainable by a direct re- lationship between the two components. In small companies, the con— trollership departments may perform many nonaccounting staff functions, such as personnel or finance. To the extent that these staff functions are split off from controllership and put into separate divisions, the company has more different types of divisions (divisional specializa- tion), but the controllership division does not have to be broken up into subunits (accounting "unit differentiation-horizontal"). The poor consistency of the control system level as awhole1 limited the number of representative control system variables; i.e., those with good explanatory power and consistency. Most of the compo- nents with better than average explanatory power were inconsistent with hypothesis two. The only two representative components were the two standardization components. Both were better than average explainers. "Standardization-jObs" is perfectly consistent with hypothesis two, while "standardization-general" is moderately consistent. 1See above, page 165. 173 "Direct supervision" was an important control system explainer, but its positive coefficients were totally inconsistent with hypothesis two. Thus the relationship of "direct supervision" to the accounting system is not in accord with hypothesis two, which predicts that control systems have negative relationships to one another.1 .A possible expla- nation of this finding is that direct supervision and the accounting system.do act as control systems but they are complementary. To the extent that a company emphasizes direct supervision as a control system, it must provide accounting information to the supervisors so they can effectively control operations. "Personnel quality-low level" was a very important explainer of the accounting system, but "personnel quality-high level" was not. Neither of the personnel quality components was particularly consistent or inconsistent with hypothesis two. A.possible explanation of the lack of good consistency is that personnel quality, though it alleviates the need for other control systems,2 also necessitates the provision of ac- counting infOrmation to make the personnel more effective. Like direct supervision, personnel quality may have a complementary relationship to the accounting system. Thus personnel quality may have a dual role: alleviating control prOblems and requiring the provision of more account- ing information. The greater influence of the quality of lower-level personnel suggests that the degree of lower-level personnel quality creates an organizational environment which has a great influence on 1See above, page 57. 2See above, page 57. 174 the development of the accounting system though the direction of influ- ence on parts of the accounting system is mixed. The "centralization of authority-purchasing" component was not used as an explainer in any of the thirteen regressions and thus had no explanatory power. The "centralization of authority-investment" component was only an average explainer. But more important, it was inconsistent with hypothesis two since two out of three of its coeffi- cients were positive. The lack of much importance of the centralization variables as explainers was surprising as well as the inconsistent re- lationship to the accounting system. The "authority levels" component was a poor explainer, having been added to only one of the thirteen regressions. The coefficient for this regression was positive, inconsistent with hypothesis two. This suggests that the split-off of authority levels from other forms of structural complexity as a control-prOblem-alleviating instead of control-problemrproducing characteristic, described on page 47, may not have been appropriate. Apparently the number of authority levels con- tributes to the control problems created by structural complexity and thereby necessitates the development of the accounting system. The "staff support" component was a poor explainer, having been added to only one of the thirteen regressions. However, its coefficient 'was consistent with hypothesis two. To a limited extent, then, the greater is the development of nonaccounting staff fUnctions, the less needs to be the development of the accounting system. 175 The Influenceability of'Accounting Variables by Organizational Variables The purpose of this section is to analyze the strength and di- rections of influence of the organizational variables on each account- ing variable. It addresses two questions: 1. To what extent is the accounting component explained by a linear combination of organizational components (explain- ability)? 2..Are the directions of the coefficients of the organizational components in the linear combination consistent with the hy- potheses which apply to the level in which the respective organizational components are located (consistency with the hypotheses)? In this section, the columns of Table 5 (page 101) are analyzed. For example, the first column, headed by accounting system "size-informa- tion output," includes feur coefficients. 'These are the coefficients of the four organizational components which were used to explain accounting "size-information output." These four coefficients are assembled into measures which indicate to what extent accounting "size-information out- put" is explained by organizational components and.whether the signs of the coefficients are consistent with the respective hypotheses.1 'The reader will find it useful to keep the text open to Table 5 as he reads this section. Table 17 accumulates measures of the explainability and consis- tency with the hypotheses of each of the thirteen accounting components. The reader will find it easier to understand the table if he concentrates 1See above, pages 100-116 and 116-27, respectively, for dis- cussions of the techniques, explainability, and consistency with the hypotheses. Explainability and Consistency with the Hypotheses 176 Table 17 of.Accounting System Componentsa Structural Control Accounting System Process Complexity System Total Component Freq R Sq Freq R Sq Freq R Sq Freq R Sq Explainability Size-infor output ..... 1.90 4.80 1.40 1.10 1.20 0.50 1.40 1.70 Size-resource input ... 1.90 1.70 0.00 0.00 0.60 0.90 0.70 0.90 Job struct compl ...... 0.00 0.00 1.40 2.30 0.60 0.60 0.70 0.80 Geog dispersion ....... 1.90 0.10 2.90 1.10 2.50 2.90 2.50 1.80 Unit diff-vertical .... 3.70 1.90 1.40 1.90 1.20 1.30 1.80 1.60 Unit diff-horizontal .. 0.00 0.00 1.40 1.70 0.00 0.00 0.40 0.30 Authority levels ...... 0.00 0.00 0.00 0.00 1.20 1.80 0.70 1.00 Report differen ... ..... 1.90 2.30 0.00 0.00 1.20 1.20 1.10 1.20 Decen of accounts ..... 0.00 0.00 1.40 1.00 3.10 2.80 2.10 1.70 Sophis of techniques .. 1.90 2.10 1.40 1.10 0.60 0.50 1.10 1.00 IMeChanization ......... 0.00 0.00 1.40 3.00 0.00 0.00 0.40 0.60 Personnel-education ... 0.00 0.00 0.00 0.00 0.60 0.60 0.40 0.30 Personnel-general ..... 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 consistency with HypothesesC Size-infer output ..... 1.00 1.00 1.00 1.00 0.00 0.00 0.50 0.84 Size-resource input ... 1.00 1.00 ° ° 1.00 1.00 1.00 1.00 Jdb struct compl ...... - - 1.00 1.00 1.00 1.00 1.00 1.00 Geog dispersion ....... 1.00 1.00 0.50 0.55 0.25 0.65 0.43 0.65 Unit diff-vertical .... 0.50 0.25 1.00 1.00 0.50 0.87 0.60 0.72 Unit diff-horizontal .. - - 0.00 0.00 - - 0.00 0.00 Authority levels ...... . - - - 0.50 0.31 0.50 0.31 Report differen ....... 1.00 1.00 - - 0.50 0.41 0.67 0.69 Decen of accounts ..... - - 1.00 1.00 0.40 0.26 0.50 0.34 Sophis of techniques .. 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 ZMechanization ......... - - 1.00 1.00 - - 1.00 1.00 Personnel—education ... - - - - 1.00 1.00 1.00 1.00 Personnel-general ..... - - - - - - - ' 3Frequency and R square data are taken from Tables 6 (page 106) and 8 (page 117), and Tables 7 (page 108) and 9 (page 125), respectively. bAverage explainability is indicated by a score of 1.00, high explainability is indicated by scores in excess of 1.00, and low ex- plainability is indicated by scores less than 1.00. CPerfect consistency with the hypotheses is indicated by a score of 1.00; perfect inconsistency is indicated by a score of 0.00. Spaces containing single dots indicate there were no coefficients for this cat- egory with which consistency with the hypotheses could be calculated. 177 at first on the columns denoting total explainability and total consis- tency with the hypotheses. Note that the explainability measures are centered on a score of 1.00, which indicates the accounting component is explained about as much as the average accounting component. A score of 0.00 indicates that no organizational components were used to explain the accounting component. In other words, given the tests for entry of components to regressions, no linear combination of any of the nineteen organizational components could be found to explain the accounting com- ponent. This was true of accounting "personnel quality-general." Scores less than 1.00 indicate poor explainability, while scores in excess of 1.00 indicate good explainability. The consistency measures are pro- portions. Perfect consistency is indicated by a score of 1.00, while perfect inconsistency is indicated by a score of 0.00. Once again the two methods, frequencies and R square increase, are used to form separate measures of both explainability and consis- tency with the hypotheses. The frequencies method ignores the strength of associations indicated by regression coefficients, while the R square increase method is biased toward early-added explainers. The explainability measures are broken into fOur groups: total, control system, structural complexity, and process. Total explainability is the extent that the accounting component is explained by a linear com- bination of the organizational components, as compared to the average explainability of the thirteen components. For example, the component "sophistication of accounting techniques" has a total explainability about average for the thirteen components. Explainability by a level is the extent that the component was explained by explainer components 178 in that level, as compared to the average explainability of accounting components by explainer components in that level. For example, "sophis- tication of accounting techniques” is explained.much more than average by process components, a little more than average by structural com- plexity components, and less than average by control system components. The consistency measures are also broken into four groups: ‘total, control system, structural complexity, and.process. Total consistency is the extent that the coefficients of organizational components in a linear combination used to explain an accounting component have the signs that are expected if the hypotheses are true which apply to the levels in which they are located. For example, the total consistency of "size- infbrmation outpu " is 50 percent or better (0.50 or 0.84). An inspec— tion of the first column in Table 5 shows that the positive coefficient on "process sophistication" is consistent with hypothesis three, the positive coefficient on "job structure complexity" is consistent with hypothesis one, and the two positive coefficients on "direct supervi- sion" and "personnel quality-low level" are inconsistent with hypothesis two. Thus two out of four coefficients are consistent, and the frequency measure is 50 percent.1 The control system measure of zero indicates the control system coefficients are all inconsistent, while the struc- tural complexity and process measures of one indicate their coefficients are all consistent. The two most explained accounting components are "geographical dispersion" and "decentralization of accounts." Neither of these K 1The R square increases are used to weight the coefficients for the R square measure. See above, pages 124-27. 179 components is very consistent or inconsistent. Both are explained mostly by the control system level of variables and both derive most of their inconsistency from that level. "Geographical dispersion" thus is the leading accounting input component, and "decentralization of accounts" is the leading accounting output component. "Geographical dispersion" and "unit differentiation-vertical" were treated together with one another as "divisionalization of the accounting system" in the analysis of relationships within the account- ing system.1 It is notable that both are excellently explained by or- ganizational components. Divisionalization of the accounting system is an important accounting system concept which is strongly related to or- ganizational variables. In addition to "decentralization of accounts," the rest of the accounting output components——"size-infOrmation outpu ," "report differ- entiation," and "sophistication of techniques"——are well explained by organizational components. In the analysis of relationships within the accounting system, "sophistication of accounting techniques" was fOund to be the only accounting output component not directly related to the other output components.2 It preserved its idiosyncratic nature here by being the only output component to have all of its coefficients in- consistent with the hypotheses. None of the other accounting components besides the divisional- ization components and the output components were significantly explained 1See above, page 151. 2See Figure 2 (page 148) and the discussion on page 149. 180 by organizational components. In general, it can be said that the over- all organization exerts its primary influence on accounting systems (in this sample of companies) by determining whether the accounting system is divisionalized and by necessitating that certain types of accounting output be produced. Apparently other organizational arrangements (ac- counting input components) such as quantity of resources applied to the accounting system ("accounting size-resource input"), aspects of organ- zation structure of the accounting system besides divisionalization ("job structure complexity," "authority levels," and "unit differentiation- horizontal"), personnel quality ("education" and "general"), and mech- anization are more unique to particular accounting systems.1 1It may be speculated that there are tradeoffs among these unique aspects of accounting systems. For example, greater job struc- ture complexity and.mechanization may be associated with lower person- nel quality, and vice versa. The multiple regression research design ignored the interrelationships among accounting components that would have incorporated such tradeoffs. See above, page 130, for a discussion of the nature of this limitation. ——-~ ‘~__.. Chapter 5 INTEGRATION OF THE RESEARCH FINDINGS In Chapter 4, the individual researCh findings were discussed without very much consideration of how they relate to each other. The purpose of this chapter is to attempt to pull together the research findings into a relatively simple and understandable block of inter- nally consistent research findings. The first step in integrating the research findings is summariz- ing the major generalizations from Chapter 4. Then the basic model, originally developed in Chapter 2, is revised to account for the re— search findings, particularly those which were found to be inconsistent with the Chapter 2 basic model. The reasons fer the Changes in the revised basic model are discussed in two sections——one on the organi- zational levels and variables, and one on the accounting system vari- ables. Finally, some revisions of the three hypotheses developed in Chapter 2 are necessitated by the basic model revisions. GENERALIZATIONS FROM THE RESEARCH FINDINGS There are many expected.and unexpected findings in the data that were analyzed in Chapter 4. Before attempting to integrate them, it is useful to assemble the generalizations that were developed in Chapter 4. 181 10. 182 As a group, characteristics of the overall organization are important to the determination of characteristics of the accounting system (page 160). . Characteristics of the accounting system are not very iMr portant to the explanation of other Characteristics of the accounting system (page 144). . Divisionalization-centralization is an important dimension of the accounting system which is strongly related to other accounting variables and to organizational variables (pages 146, 179). . The characteristics of accounting output are important variables of the accounting system which are strongly re- lated to other accounting variables and to organizational variables (pages 144, 179). . The output of accounting systems apparently develops in two alternative directions: the provision of more sophisti- cated information only for high-level management or the provision of more unsophisticated information for lower levels of management (page 150). . Centralized accounting systems tend to emphasize the pro- vision of sophisticated infOrmation for high-level manage- ment while divisionalized accounting systems tend to empha- size the provision of unsophisticated information to high and lower-level management (page 156). . Increases in the development of the overall structural complexity level are associated with increases in the development of the accounting system (this confirms hy- pothesis one), though the overall structural complexity level was not as strongly associated with the accounting system as anticipated (page 166). . Increases in the sophistication of the production process are strongly associated with increases in the development of the accounting system (this confirms hypothesis three; see page 167). . The control system level is strongly associated with the accounting system.but increases in the development of dif- ferent characteristics of the control system are associated ‘with both increases and decreases in the development of the accounting system (this fails to confirm hypothesis two; see page 166). Process sophistication is the only important process vari- able used in this dissertation (page 170). 183 ll. Increases in company size do not lead to increases in the development of the accounting system, though they signif- icantly influence some accounting system characteristics (page 170). 12. All of the pure structural complexity variables (those involving the breakup of the organization into parts on various dimensions: "job structure complexity," "divi- sional differentiation," and "authority levels") were positively associated with the development of the account- ing system (pages 170, 172, and 174). 13. All of the standardization variables were negatively asso- ciated with the development of the accounting system (the relationship between the accounting system and standardiza- tion is thus consistent with hypothesis two; see page 172). 14. The control systems "direct supervision" and "personnel quality" have positive relationships to the development of the accounting system. Apparently supervisors and higher- quality personnel need more accounting information to be effective. This complementary relationship apparently supersedes the negative relationship predicted by hypoth- esis two (page 173). 15. Decentralized companies do not have more fully developed accounting systems, as predicted by hypothesis two (page 174). 16. Accounting variables, other than divisionalization-central- ization and the accounting output variables, are not sig- nificantly influenced by the overall organization (pages 179-80). REVISION OF THE BASIC MODEL Though two of the three hypotheses incorporated in the basic model are confirmed, there are so many inconsistencies and loose ends in the research findings that some revision of the basic model is ad- visable. Figure 3 outlines a revised basic model which is designed to incorporate many of the research findings while retaining as much sim- plicity as possible. The roles of some of the explainer variables are reinterpreted in light of the research findings. Some variables which 184 Process Sophistication assumed } INSTRUMENTAL FACTORS Structural Complexity ——- Jab struct compl, Divisional diff} Authority levels Information Concentration of Mechanization Managerial Resources Mech-computers Gent of'auth-inv, Personnel-high assumed { CONTROL SYSTEM Demand Emphasis General Standardization Other Staff Demand for for on Super- Personnel of Procedures Functions De ci 5 i on- Control vision Quality Standard-jobs, Staff support, . . . . . . .Maklng In- Infbr- Direct: .Rersonnel- Standardiz- DiviSional formation mation super low general specialization comple- l l imentarity ACCOUNTING SYSTEM Organization- d—Size-infor output,—v- Out ut g ally mandated Output Report differen, —"30 h 1’; ti- character- diffusion sophis of'tech,-——+- p . . . cat10n istics +—Decen of accounts A _*__ . . . . . _ Geog dispersion, Dlglliszleogl «— Unit diff-vert, Centralized -+—Authority levels Uhdepartmen- . . . Departmental— — talized Unit diff-horiz—y- ized Manual data Mechanization-—- Mbchanized processing data processing Lower-quality .+_ High-quality personnel Personnel-genl personnel Flexible & Rigid G responsive to unresponsive to infOrmation information needs of lower- needs of lower- 1evel managers level managers Size- 1 resource input Autonomous Character- Job struct compl, . . Personnel-educ lstlcs Figure 3 The Revised Basic Model (This figure is a revision of the basic model presented in Figure 1, page 44) 185 appear to act together are merged into overall concepts. The roles of the explainer levels are reinterpreted as necessary. The accounting system variables are rearranged in accord with their influenceability by organizational variables and by each other. These changes are de- scribed in the following two sections. Table 18 includes the regression coefficients for the stepwise regressions of the accounting variables on the potential organizational variables. These coefficients are the same as those in Table 5 (page 101), except that the rows and columns have been rearranged in accord with the revised basic model. The accounting system variables (columns) which are presumed to act together are placed together. The reader will find it useful to refer to Table 18, as well as Figure 3, while reading the discussion in the following sections. RECONSIDERATI ON OF THE ORGANIZATIONAL LEVELS AND VARIABLES The purpose of this section is to analyze and justify the changes that were made in the organizational levels and variables of the revised basic model, Figure 3. The major changes that were made in the levels were as follows: 1. The process level was narrowed to the single variable "proc- ess sophistication," since the other two process variables were not 1mportant. 2. The overall structural complexity level was renamed "instru- mental factors" to indicate that it includes more than just structural complexity. 3. Complementarity between the accounting system and other control systems was recognized as possibly accounting for positive relationships. 186 Table 18 Regression Coefficients for the Stepwise Regressions of the Thirteen Accounting Components on Nineteen Potential Organizational Explainer Components——Rows and Columns Rearranged per the Revised Basic Modela Organizationally dated Characteristics Organizational . . Centralized- . Output Diffu51on- . . . Explainer Component Sophistication DHHHE;;?- .A H J I D E G Process Sophistication ................... 0.6 0.7 - - 0.2 -0.4 . Output diversity ................. - - - - - ~ - Materials input diversity ........ - - -0.5 . . 0.3 . Instrumental factors Size ............................. - - -0.3 ° -0.8 0.3 - Structural complexity Job structure complexity ....... 0.4 - - 0.4 - - . Divisional differentiation ..... - - - - 0.6 - ° Authority levels ............... . . . - 0 4 - . Information,mechanization Mechanization-computers ........ - - . . - . - Concentration of managerial resources Centralization of auth-invest .. - 0.6 - -0.5 - 0.3 - Personnel quality-high ......... - . - -0.2 0.5 - - Control system Complementary Direct supervision ............. 0.2 - - 0.4 1.0 - 0.6 Personnel quality-low .......... 0.3 - 0.4 0.3 . - -0.4 Standardization of procedures Standardization-jObs ........... - - - - - -0.8 - Standardization-general ........ - -0.4 - 0.3 -0.8 - - Other staff functions Staff support .................. . - . - - . - Divisional specialization ...... . - - - - - - 8These are the same coefficients as those in Table 5 (page 101). The thirteen columns of accounting system components represent different regressions. Each column includes the coefficients for the explainer components which were added to the regressions of the accounting system component which heads the column. The rows and columns are rearranged 187 Table 18 (Cont 'd.) Autonomous Characteristicsb Coef. Organizational Si Explainer Component F K M B C L E315 Process Sophistication .................. - . . . . . Output diversity ................ - - - 0.5 - - + Materials input diversity ....... . . . . . . Instrumental factors Si 23 ............................ ' ' ° 0 O 0 Structural complexity Job structure complexity ...... - . - . . . Divisional differentiation .... . - . - . . Authority levels .............. - . - . . . + Information mechanization Mechanization-computers ....... . 0.5 - - 0.5 - Concentration of managerial resources Centralization of auth-invest . - . . . . . Personnel quality-high ........ ~ . . . . . Control system Complementary Direct supervision ............ - . - . . . + Personnel quality-low ......... - . - - -0.4 - Standardization of procedures Standardization-jobs .......... - - . -0.5 - - Standardization-general ....... . . . . . . Other staff functions - Staff support ................. - - - - - -0.4 Divisional specialization ..... -0.4 - . - - - as follows: The organizational variables (rows) which are presumed to act together are placed together. The accounting system variables (col- umns) which are presumed to act together are placed together. bAccounting system Components: (A) Size-Information Output; (B) Size-Resource Input; (C) Job Structure Complexity; (D) Geographical Dispersion; (E) Unit Differentiation-vertical; (F) Unit Differentiation- Horizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decen- tralization of Accounts; (J) Sophistication of Techniques; (K) Mechaniz— ation; (L) Personnel Quality-Education; (M) Personnel Quality-General. 188 Some of the organizational variables are merged into overall concepts or redefined. The three variables which measure structural complexity are merged. The new variable, concentration of managerial resources, is proposed to account for the inconsistent coefficients of two organizational variables. The meanings of other variables are re- fined. An especially important finding is the strong and consistently positive relationship of the process level to the accounting system. There are several implications of this finding. First the accounting system does not develop only in response to the control needs of a com- pany. As was explained on page 39, one of the main reasons for includ- ing the process level (the only aspect of context measured in this dis- sertation) was that the stage of development of the accounting system may be related to demands for decision-making information necessitated by the complexity of the production process. The strength and consist- ency with hypothesis three of the explanation of the accounting system by the process level supports this explanation. The structural complexity level in the basic model includes some variables, such as company size and mechanization, which are not, strictly speaking, structural complexity. As was discussed in the model development stage on page 50, the key requirement for inclusion in the overall structural complexity level was that a variable was perceived to increase control and coordination prOblems. An assumption of the model, though it was not tested in this dissertation, is that context influences the variables that are included in the overall structural complexity level. The overall structural complexity level is renamed 189 "instrumental factors” to clarify the fact that it includes more than pure structural complexity. These instrumental factors are necessary to the accomplishment of the production process, but they may have ef- fects on the development of the accounting system either because of additional control needs or because of additional infOrmation needs. .All three of the "pure" structural complexity1 explainer vari- ables used in this dissertation——”j0b structure complexity," "divisional differentiation," and "authority levels"——were found to lead to the de- velopment of the accounting system. It had been proposed in the model development stage on page 46 that authority levels, in contrast to the other two structural complexity variables, might alleviate control prob- lems and thus (as a control system) be negatively related to the ac- counting system. The unifbrmly positive effect of all three structural complexity variables, including "authority levels," suggests the more general proposition that increasing structural complexity (of all types) leads to the development of the accounting system. Thus the three vari- ables are merged into the concept "structural complexity" and.incorpor- ated in the revised basic model as an instrumental factor, and "authority levels" is removed from the control system level. Since hypothesis two, postulating negative relationships among control systems, was not confirmed, a reconsideration of the role of the control system level is necessary. It seems unlikely that the account- ing system is the only part of an organization which contributes to the l"Structure"'was defined above, on page 27, as something com- posed of parts. .All three of these variables have to do with the breakup of the organization into parts along various dimensions. 190 control function. Why then would positive relationships be found between control systems? One explanation is that there may be complementary re— lationships between control systems. One control system may need the other in order to be fully effective. This explanation has been sug- gested to explain the positive relationships between the accounting system and both personnel quality and direct supervision. Supervisors and.higher-quality personnel can be assumed to need.more accounting infOrmation to be fUlly effective in perfOrming their control functions. If the complementary explanation is true, then all-negative relationships between the accounting system and other control systems cannot be ex- pected. This complementarity explanation is incorporated in the revised basic model fer the relationships of the accounting system to direct supervision and personnel quality. In contrast to direct supervision and personnel quality, the control system standardization variables had strong and consistently negative relationships to the accounting system in confbrmance with hypothesis two. Thus it can be said that companies with standardized procedures do not need as large or complicated accounting systems. Standardization of procedures thus remains in the control system level. Negative relationships to the accounting system are expected since no complementarity is evident in the research findings. The only other control system variable that was consistent with hypothesis two was "staff support," which has a single negative coeffi- cient on accounting system."personnel quality-education." Thus "staff support" is not a very important explainer of the accounting systeml However, the overall structural complexity variable ”divisional 191 specialization" was also negatively related to the accounting system. It had been assumed in the model development stage (page 49) that a company with many different types of divisions (as measured by "divi- sional specialization") would have more control prOblems and thus need a more fully developed accounting system. The negative relationship of "divisional specialization” to the accounting system disputes that ex- planation. It is possible that many of the different types of divisions perfbrm.staff functions which are, in fact, control systems. For exam: ple, personnel and data processing divisions exert control over employ- ees within their areas of responsibility. If this explanation is true, then "divisional specialization" and "staff support" can be classified as control system staff functions which have negative relationships to the accounting system. In other words, companies with well-developed staff functions other than accounting do not need as fully developed accounting systems. It is interesting to note that, of the two mechanization vari- ables, only "mechanization-computers" has any effect on the accounting system, ”Mechanization-general"'was not used as an explainer of any of the accounting system components. Thus this dissertation has found no evidence that the general level of mechanization of a company, in- cluding the degree of production mechanization, has anything to do with the stage of development of its accounting system. On the other hand, ”mechanization-computers" was an important explainer whose relationships to the accounting system were consistent with hypothesis one, confirm- ing its role as an instrumental factor (fermerly overall structural complexity). It can be said that more computerized companies have more 192 fully developed accounting systems.1 The explanation of this positive association suggested by hypothesis one is that computerization contrib- utes to the control problems of the company, perhaps because of the variety of employees required in a computer operation, and thereby necessitates the development of the accounting system to alleviate those problems. The centralization variables were somewhat of an enigma. It was assumed in the model development stage (pages 34 and 46) that cen- tralization acts as a control system by preventing deviation from top management plans. It thus is expected to be negatively related to the accounting system, which is also assumed to act as a control system. It seems logical that decentralized companies would.need additional reports on the perfbrmance of lower-level decision-making management. The research findings dispute this assumption. "Centralization of authority-purchasing" was not used as an explainer of any of the ac- counting system variables, while "centralization of authority-invest- ment" was positively related to the stage of development of the account- ing system, Two explanations can be dismissed immediately. It is difficult to conceive of decentralization as contributing to control or of the existence of complementarity between the accounting system and centralization. Consequently, another explanation must be sought. 1"Mechanization-computers"was positively related to two accounting system variables, "mechanization" and "jOb structure com- plexity." The relationship of "overall company mechanization" to "accounting system mechanization" is to be expected. The relationships of ”mechanization-computers" to "accounting jOb structure complexity" is the major concern here. 193 Why would centralized companies have more fully developed ac- counting systems? A possible explanation is that some companies may employ highly educated and skilled.management at the top management levels instead of dispersing their managerial resources to all levels. High-quality managers have a great need for sophisticated accounting information which ordinary managers might not be able to use. In such companies, most decisionsmaking is confined to top management, since lower-level management is not qualified to make decisions. If this ex- planation is true, then "centralization of authority-investment" actu- ally measures "concentration of managerial resources." SuCh concentra- tion is an instrumental factor since it facilitates the accomplishment of the production process. In light of the re-evaluation of the role of centralization, "personnel quality—high level" was re-examined. It was fOund on page 173 not to be consistent or inconsistent with hypothesis two. It now seems more appropriate to place it under the caption "concentration of managerial resources" developed for "centralization of authority- investment." It seems logical that control is exercised primarily over lower-level employees, and thus "personnel quality-low level," as dis- cussed on page 173, would act as a control system, "Personnel quality- high level" is proposed to measure concentration of'managerial resources at the highest level. In addition to the three explainer variables which were not used to explain any accounting system variables,1 three other variables were lThesewere "geographical dispersion," ”mechanization-general," and "centralization of authority-purchasing." 194 not incorporated in the revised basic model (Figure 3). The most impor- tant was company size, which was an important explainer of the account- ing system but had varied effects on the development of the accounting system. Company size apparently induces the development of some ac- counting system characteristics while restricting the development of others.1 In consequence, it was not possible to postulate its overall relationship to the accounting system, and it was left out of the revised basic model. The other two variables not incorporated in the basic model were "output diversity" and."materials input diversity." Both of these proc- ess variables were weak explainers, and one was inconsistent with hy- pothesis three. "Process sophistication" so dominated the process level of explainers that it was adopted as the overall concept in the revised basic model in Figure 3. RECONSIDERATION OF THE ACCOUNTING SYSTEM VARIABLES The purpose of this section is to incorporate key accounting system concepts and to order the accounting system variables in the re- vised.basic model. The key accounting system concepts found in Chapter 4 were "centralization-divisionalization” and "sophisticated high-level output, unsophisticated lower-level output." Two bases of ordering the accounting system variables were used. First they were ordered by the degree they are influenced by organizational variables. Second they are ordered by the types of accounting systems developed in Chapter 4:2 1See Table 18, page 186. 2See above, pages 155-58. 195 centralized-sophisticated techniques versus divisionalized-unsophisti- cated techniques. ,A very important finding of this disseration was that account- ing system characteristics differ greatly in the extent they are deter- mined by characteristics of the overall organization.‘ Some accounting Characteristics, such as the output variables and the divisionalization variables, are well explained by organizational variables, while others, such as "personnel-education" and "unit differentiation-horizontal," are poorly explained. This suggests the possibility that some Character— istics of accounting systems may be mandated.by the overall organization. These characteristics in turn may influence other characteristics of the accounting system which are more autonomous from the overall organization. In Figure 3, the accounting system characteristics are ordered vertically by the extent of their influenceability by the overall organization. The accounting system characteristics are ordered horizontally by the association of characteristics of the accounting system (alterna- tive types of accounting systems) discussed on pages 155-58. A.basic accounting system distinction was found there between those which empha- size sophisticated accounting infbrmation for top—level management and those which emphasize elementary accounting information for high and lower levels of management. Emphasis on sophisticated accounting infor- mation is labeled "output sophistication" in Figure 3, while emphasis on elementary accounting infOrmation for lower management levels is labeled "output diffusion." This basic distinction is essentially 1See above, pages 167-74. 196 mandated by the overall organization since all of the output variables are extensively explained by organizational variables. In summary, the overall organization determines whether the accounting system will em- phasize output sophistication or output diffusion. A second basic accounting system distinction is between those which are divisionalized and those which are centralized. Centralized accounting systems have a single accounting office for the entire com- pany, while divisionalized accounting systems have accounting offices at division headquarters. Like the output variables, the divisionaliza- tion variables which measure this distinction are also extensively ex- plained by organizational variables. It was suggested that centralized accounting systems tend to emphasize output sophistication, while divi- sionalized accounting systems tend to emphasize output diffusion. Con- sequently, the output and divisionalization characteristics are ordered in the same direction in Figure 3. The rest of the accounting system variables were explained less 1 Some of the characteristics than average by organizational variables. represented by these variables tend to be associated with one or the other types of accounting systems: centralized—output sophistication and divisionalized-output diffusion. Centralized accounting systems tend to be more departmentalized, more mechanized, and have higher qual- ity personnel than divisionalized accounting systems. Three accounting system variables were not clearly associated with the centralized-output sophistication, divisionalized-output diffusion dimension. These were 1See Table 17, page 169. 197 "size-resource input," "job structure complexity," and "personnel- education." Since they were also poorly explained by organizational variables, they were placed at the "autonomous" end of the accounting system ordering of characteristics. The possible association of "flex- ibility and responsiveness to information needs of lower-level managers" with the "divisionalized-output diffusion" type was suggested on page 156, though flexibility and responsiveness were not measured in this dissertation. REVISION OF THE HYPOTHESES Perhaps the most important changes in the revised basic model concern the relationships between the levels of explainer variables—-— process, instrumental factors, and control system——and the accounting system level. The direction of these relationships was predicted by the hypotheses listed on page 57. The directions actually found, as well as changes in the variables and levels of variables discussed in the preceding sections of this chapter, necessitate some refinements of those hypotheses. Hypothesis three stated: 3. The more sophisticated is the production process of the or- ganization, the more the accounting system.must be developed since it must provide more and.better infOrmation fer manage- ment decisions. The strong and consistently positive relationship of the process level to the accounting system confirms hypothesis three, and no revision is necessary . 198 Hypothesis one stated: 1. Structurally complex organizations tend to have more fully developed accounting systems to contribute to the resolution of greater control and coordination problems, given the proc- ess and stage of development of other control systems is held constant. The consistently positive relationship of the overall structural com- plexity level to the accounting system confirmed hypothesis one. If the definition of structural compexity in the hypothesis is restricted to "pure" structural complexity,1 the hypothesis is very strongly confirmed. However, the generally poor explanatory power of the overall structural complexity level, the reinterpretation of some of the variables within the level, and the renaming of the level "instrumental factors" require some refinement of hypothesis one. Hypothesis one explained that control and coordination prOblems as a result of overall structural complexity necessitated the develop- ment of the accounting system. The control and coordination explanation is probably true for "pure" structural complexity but may not be true for the other instrumental factors, information.mechanization and con- centration of managerial resources. Companies with high concentration of managerial resources need sophisticated accounting infOrmation fer top-level management decision-making, but their control infOrmation needs are prdbably not as great since authority is centralized. Thus the effect of concentration of managerial resources on the accounting system does not go through the control system. 1These are the three variables having to do with the breakup of the organization into parts: ‘"jOb structure complexity," "divisional differentiation," and "authority levels." 199 The effect of information mechanization on the accounting system was discussed on pages 191-92. The control and coordination explanation does not seem reasonable for infOrmation.mechanization. The positive relationship between the accounting system and information mechaniza- tion is probably due to one or both of the following reasons: (1) The same infOrmation needs that necessitate the development of the account- ing system also favor computerization to improve the efficiency and effectiveness of the overall information provision function. (2) In small companies, such as those in the sample, the computer function is located within the accounting function. In light of the above, hypothesis one is restated as fellows: 1. Organizations with highly developed instrumental factors tend to have more fully developed accounting systems either to contribute to the resolution of greater control and coordination prOblems or to satisfy greater needs fer information for decision-making. Hypothesis two stated: 2. The stage of development of the accounting system is inversely related to that of other control systems, when process and structural complexity are held constant, since control sys- tems are partial substitutes for one another. The key change with respect to the control system level is that negative relationships can only be expected when there is no complementarity.‘ Consequently, hypothesis two is restated as follows: 2. The stage of development of other control systems is inversely related to the stage of development of the accounting system since control systems are partial substitutes for one another, given process and structural complexity are held constant and also given there is no complementarity between the accounting system and other control systems. When complementarity does exist, the accounting system may be positively related to the complementary control system. 1See above, page 190. Chapter 6 SUD/MARY AND CONCLUSIONS The purpose of this chapter is to tie the dissertation together, discuss its implications, and suggest avenues for future inquiry. There are four sections of the chapter. The first reviews the steps in the inquiry in a brief manner so that the reader may have a perspective on the entire dissertation. The second section summarizes the important findings of the study. The third section discusses some implications of the findings of studies of this type for accountants and managers in organizations. The fourth section proposes future researCh in the area of the organizational implications for accounting. REVIEW OF THE STEPS IN THE INQUIRY In order that the reader may gain perspective on the entire dissertation, this section summarizes the major steps in the disserta- tion. The first step, in Chapter 2, is to review literature in the fields of sociology (organization research) and accounting which led to the development of the basic model of this dissertation. The basic model consists of some general presumed relationships between about twenty variables of an organization and about ten variables of its ac- counting system. These general relationships are incorporated in three hypotheses which are tested in this dissertation. The hypotheses 200 201 predict the direction, positive or negative, of relationships between three levels of organizational variables and the accounting system variables. The research design is covered in Chapter 3. ,A sample of eigh- teen small manufacturing companies in southern Michigan was selected to test the hypotheses and generally examine relationships between vari- ables of their overall organizations and variables of their accounting systems. Over one hundred questions were addressed to the controllers or chief financial officers of these companies in on-site interviews. The responses to these questions were, in some cases, mathematically manipulated to produce measurements which correspond to the variables in the basic model. For each variable in the basic model, there were from.three to twenty measurements after the mathematical manipulations. These were combined into a single measurement fer each variable using principal components analysis. The relationships between the accounting system variables and the organizational variables were calculated using stepwise multiple regression analysis. This technique finds subsets of the organizational variables whiCh are the best possible explainers of eaCh accounting variable. Using the results of the multiple regression analysis, meas- ures of the explanatory power of each organizational variable are calcu- lated.which are, generally speaking, the number of times it is used in the subsets for the different accounting variables. The explanatory power of levels of organizational variables is calculated in a similar manner. The explainability of each accounting system variable is cal- culated, generally speaking, as the number of organizational variables 202 in the subset used to explain it. Measures of the degree the research findings confirm the three hypotheses are calculated. These are called "consistency with the hypotheses" and are essentially the proportions of the relationships between accounting components and organizational components which have the direction, positive or negative, that is pre- dicted by the hypotheses. Consistency with the hypotheses is calculated for individual organizational variables, levels of organizational vari- ables, and accounting system variables. The relationships among the accounting system variables were also calculated using stepwise multiple regression analysis. ,Measures of explanatory power and explainability of the accounting variables with respect to other accounting variables were calculated. In addi- tion, the relationships among the accounting variables were analyzed using path analysis. The output of the preceding techniques are the researCh findings which are interpreted in detail in Chapter 4. Chapter 5 attempts to integrate the research findings of Chapter 4. .A revised basic model is developed which strives to explain those research findings which are inconsistent with the original basic model. The hypotheses are also revised in accord with the revisions in the basic model. IMPORTANT FINDINGS This section summarizes the most important findings of the study. These findings are incorporated in the revised basic model and revised hypotheses developed in Chapter 5. The purpose of this section is to lay out the important findings in a narrative style. 203 The most important finding of the study is that organizational characteristics are important to the determination of many characteris- tics of the accounting system. The researCh design of this dissertation cannot reveal whether the organizational characteristics cause Changes in accounting system characteristics, but that seems to be a logical assumption. Structural complexity, defined as the breakup of the organiza- tion into parts along various dimensions, leads to the development of the accounting system. .A logical explanation of this finding is that structural complexity creates control and coordination problems for organizations which must be alleviated by the development of control systems, particularly the accounting system. The sophistication of the production process is a very important factor which leads to the development of the accounting system. A logi- cal explanation for this finding is that process sophistication creates a great demand for accounting information for decision-making. The stage of development of the accounting system is negatively related to the stage of development of some control systems. These control systems are "standardization of procedures" and "other staff functions." .A logical explanation of these negative relationships is that each of the control systems helps alleviate control and coordina- tion problems created by structural complexity. The control systems are to some extent substitutable in alleviating these prOblems. To the extent that one is highly developed, the other does not need to be as highly developed. 204 The stage of development of the accounting system is positively related to the stage of development of some other control systems. These control systems are "supervision" and "personnel quality." An explanation of these positive relationships is provided by complemen- tarity, the extent that control systems developed together are more effective than any one developed separately. Supervisors and quality personnel are more effective in alleviating control and coordination problems when they are provided with adequate accounting infOrmation. Some "instrunental factors" other than structural complexity lead to the development of the accounting system. An "instrumental factor" is a characteristic of a company which provides the means fOr it to manufacture product. For example, the organization structure of a company, as measured by structural complexity, provides the environ- ment within which the manufacturing process takes place. TWo other "instrumental factors" were found: "information mechanization" and ”concentration of managerial resources." Concentration of managerial resources is the extent a company puts exceptionally high-quality per- sonnel at the top managerial levels, leaving much poorer personnel at lower managerial levels. It is logical that high-quality top executives may demand sophisticated accounting information for decisions that lower-quality top executives in other companies might not be able to use. The positive relationship between overall company infOrmation mechanization and accounting system mechanization is intuitively logical. The overall organization influences primarily the nature of the output of the accounting system. Accounting system output involves the nature of reports, the places in the organization where they are sent, 205 and the sophistication of accounting techniques. The overall organiza- tion has its greatest effect on whether accounting systems stress output sophistication or output diffusion. Output sophistication is the empha- sis on producing sophisticated accounting reports for top-level manage- ment. Output diffusion is the emphasis on producing unsophisticated information for high and low levels of management. The characteristics of accounting system output seem to require the development of certain patterns of accounting system organization structure. ,A centralized accounting system seems to be associated with output sophistication, and a divisionalized accounting system seems to be associated with output diffusion. Centralized accounting systems have a single accounting office at the company headquarters, while divisionalized accounting systems have accounting offices at division headquarters. Some other characteristics of accounting systems seem to be associated with the two types: centralized-output sophistication and divisionalized-output diffusion. Centralized accounting systems tend to be more departmentalized, more mechanized, and have higher qual- ity personnel than divisionalized accounting systems. IAfl’LICATIONS FOR ACCOUNTANTS AND MANAGERS The purpose of this section is to suggest some practical uses for the findings of this and similar studies of the relationships of characteristics of the overall organization to characteristics of the accounting system. Perhaps the most promising application is to the design of accounting systems. Research of this type can provide designers with 206 an indication of the typical type of accounting system organization structure associated with given accounting output requirements and the typical type of accounting output necessitated by given overall organiza- tional characteristics. Of course the typical accounting output and accounting organization structure characteristics are not necessarily the optimum ones. But the range of values of a characteristic consid- ered by the designer may be narrowed when he knows the typical value. As an example, consider a company with great concentration of managerial resources at the top level. It may be found in studies similar to this one that such companies require very sophisticated accounting information.1 It has been found by this study that sophisti- cated output tends to be produced by centralized accounting systems. Consequently, an accounting system designer fer such a company might lean toward a centralized accounting system. Another very interesting example goes beyond accounting system design to overall organization design. It involves the tradeoff among control systems. This dissertation has fOund that the stage of develop- ment of the accounting system.is negatively related to the degree pro- cedures are standardized and the stage of development of other staff functions, say personnel and data processing. Given a fixed need for control and coordination, mandated by process, instrumental factors, and perhaps other variables, is it cheaper to standardize procedures, develop the accounting system, or develop other staff functions? Also, 1The research findings for this relationship were not clear cut, though it seems intuitively reasonable. 207 which of the three control systems are more effective in alleviating control and coordination problems? Research such as this can begin to answer these questions. Research of this type can make generally known to accountants the organizational and accounting terminology and concepts they need to consider in making decisions not involving accounting systems de- sign. For example, the concepts of output sophistication and output diffusion ought to be known to accountants. An accountant in a company that emphasizes output diffusion should recognize the limitations of the accounting system so far as producing sophisticated accounting in- fOrmation. .Accountants ought to recognize some of the organizational factors which produce demands respectively for control information and decision-making infOrmation. For example, evidence from this disserta- tion suggests that complex organization structure creates a demand for control information. On the other hand, it has been suggested that the sophistication of the production process creates a demand for decision- making infOrmation. Another very important role of the research is the impact it may have on the attitudes of managers and accountants within organiza- tions. Both should appreciate the organizational constraints on the accounting system. IManagers often castigate accountants personally for being inflexible and malicious. If they understood the degree that the nature of the accounting system is governed by organizational consider- ations, they would realize that accounting personnel are not to blame fer the nature of the system. On the other hand, knowledge of organi- zational considerations might induce accountants to be more responsive 208 to the needs of managers, as reflected in their role in the overall organization. PROPOSED FUTURE RESEARCH The purpose of this section is to suggest various types of re- search in the area of the relationship between the overall organization and the accounting system. Two purposes should.be served by such re- search; it should overcome certain shortcomings in the research design of this dissertation; and it should cover some new sUbstantive areas, such as the reaction of accounting users to various types of account- ing systems. The most important weakness of this dissertation has been the inability to demonstrate causality. The measurements in this cross- sectional study were taken at a single point in time and thus there is no way to show that changes in one variable cause changes in another. Yet causality has been the primary interest of the dissertation. All of the hypotheses were stated in a causal manner. It would have been desirable to be able to say that Characteristics of overall structural complexity caused the development of the accounting system. Yet all that could be said was that characteristics of overall structural com- plexity are associated with the stage of development of the accounting system. The only way to establish causality is with a longitudinal study of some type. JMeasurements must be taken at multiple points in time for the same companies. This would require either a very long study in which the interviewer returned to the companies at periodical intervals, 209 or it would require companies which could provide data on organizational and accounting Characteristics at one or more points in the past as well as the present. The second approach is much more feasible but places some restrictions on the study. First, the data collected must be of a nature that typical companies would be able to obtain them readily fer the past point in time. It is very likely that a complex interview instrument such as that used in this dissertation could not be used in such a longitudinal study. Second, companies must be selected which have had changes in the organizational Characteristics measures. For example, sample companies which had not grown in size could demonstrate nothing about the influence of size on the accounting system. In sum- mary, a longitudinal study is recommended with a limited number of carefully chosen measurements collected from a sample of companies that have grown and otherwise changed their organizations. Such companies should of course be able to reconstruct the measurements at a given past point in time. Another major shortcoming of this dissertation has been the pervasive effects of the small sample size, as discussed on page 130. It is recommended that future studies, whether cross-sectional or longi- tudinal, have a sample size of at least thirty, and that a holdout sample in addition to the thirty be used to validate the relationships that are found. In addition to the preceding statistical considerations, future studies should include some additional substantive areas. The strength of process as an explainer of the stage of development of the accounting 210 system suggests that other Characteristics of organization context1 should be examined. The environment of the organization may be an im: portant explainer of the stage of development of its accounting system. Such environmental characteristics as the type of community (urban or rural), the political situation, the existence of consumer and other special-interest groups in the community, the extent of labor unioniza- tion, and the economic growth rate of the community might be considered. Naturally a complex and changing environment can be expected to be asso- ciated with the development of financial accounting (the production of information for outsiders to the organization). Additional reports might be required fer government agencies, community organizations, special-interest groups, etc. But it would be interesting to find if complex environment were associated aISO'With the development of manage- ment accounting (the production of information for employees of the organization). Do managers need more infOrmation to cope with a complex and changing organization environment? The measurement of process sophistication in this dissertation 'was essentially a very crude treatment of technology. It was not ex- pected that it would play such a vital role with respect to the develop- ment of the accounting system. Future studies should.measure technology in more systematic and orthodox ways than has been done in this study. It has been shown in this study that the technology of the organization is important to the stage of development of the accounting system. Research questions for future studies should be directed toward 1See above, page 38, for a discussion of context. 211 determining what aspects of technology are important to the accounting system, and specifically how those aspects relate to the accounting system. APPENDICES APPENDIX A QUESTIONS FOR INTERVIEWS AND MEASUREMENT RULES 212 .ouaufiu ucaaov a nu azowuorw on» u« undoa Handout uzu we uzwvu can u «woman can we unwed «nafiuev can no ucuuu on» On nouuuu ouucu o» oucouuosv «we vcaom .v ...uoa ecu a» aouuaa oau unwed “weaves 05a oyoflv anon» tucaoua Hmawoev oo newcu ruouuuo flan unnpcoc .n .meuvco you unseen uuouu cuav a u; "a“: nu umuou an as; hopes: canvases - on on >wosw~ a“ wueuo>n much .uouuuwuu macaocu on“ we nouuuu «nu oxaua>q .ao«u¢a:v auuoco «maauaaa a new vuuuofloe nu ouuozu use case one! cock .Uuo .~ on mousse vacuum 0:; .H no ouaazu uauuu osu anon. .ouoaooa nuance odnuuuau nos .« .u on uo>aou on u no. A on using. as» n oueum .u mold... mmwmom :2 33.. mmo«uu on» uawuaqzuuau nu auo>ou cyanogen to Iuov«0:09 you and ouaxunwauq no nouunaov gash .aaHUuuuo osu ca couumaoq coca osu masque cu vausuwmaau one annuu«muo ca ausaumuana no nevusamv .chocam am .n .uuo .aouaqaaou .uuwod floccuuaumcawuo .aaann nuanaua vovanucu can new rovsu ecu no uawu> H-uuaau ecu .oaannuon an anosuuuofino as .Ounaau-o ucoecoamuu ecu o>o£ .ounnonaupc an: nu Imuum>n tuucuqou a «a .uo:=a>uu no go:- .o~nu«uc> ouuuuaouaac an no vane: ..uuu .uuwcaasou .nuuao «unequouaeouuo .aouaa uosvoun on» we uou«vfl« «Au mo umuuo>a vuusxuma a mean: ado-«unuuauasu any we use:« hoeaaoo H~a9u>o ad uunuuncou ..ouo .uowuunaoo huauuuunno .auuca uaaeuu unassumuo .oo=«« unseen; uo>o nuouuav coda: ouuouuouuuueno sauaaou aaauopo an ozone mouuavsu coaunosv a can: .N .vovaaoxu on v~=on. xuauason use onuuuao unequzuou can cathodes» on. .ucouuuosc o>uuuoaucu a“ vuv:~ua« on vases. auuoczon ecu casual noqcmaaou onu «an we ucowuauou «an new cockedaao ~a< .uooquooav du- ou uuuauou saw: vu>uoopo on unsozu hungoaon och .%omaaou Havana can 55 cum; on museuucao ones: «0 when no «on .noqaaaaou huawawmpze an: evade new a» eucwuov u« was aqua—ac: uo nan: can» an vacqnuudov nu audacuon xcuaeou use .cu:u:~u:« uauuomesafi season have: a»: nous: acouuqnucamuo «a use no nouuuuuzamuo so on nwaaqoac we and: och .a0«anfiou nozuo an vouuouuccu no vocab use «avenues ucavcoaovcu cue sous: no“ ccaaeco cu vouuuuuaou an swag. 05a .hu lueua:oucoo .nvoaov cu aqua h~o>wuanou anon o>ns cows) quacnnaoo cu vousuoauuu on. clean». douuoou so: autumn-asuu cu u« noun-con on» no oceans; woman can .M 83. 42.3353“ and: Euro; 82. 358E up. ”2258 mmqam Ewe/mag Dz< mzmH>mmEH mom mZOHHmmSO (x VfiHAH7mwArmx‘ gas 182 audau =m<flx A< .n unavoum sun-«us oau no oouuo usuunoo owuuo>< .c «Auoououua «douhoav conuqu non uaosu new: gonna-non noon-bu no acne o con: 03 so: .n «candida-u. «ouoooouu hook uo anon o ooh: o: no: .« «Anon o ools on no! puuoao soda-owoouuo no opus nooouoo oAu one: .u “mummuumm .oouuooom oouuoavoun o mow-hounds .uoovoua any do huuuouwv nouxuoa onwu uuogu uo uoos noon. on: ooouoaoao on. noohodnlu non-H noouwn .5 .uoo«>uoo=o ouooouu no nowuooouoo on no. Aucu.oon.ma «do» Hocouuqouuov cauuooa- ooov anon ooooouu no mouuooouoo sou: ooohoaoao who-«ouooooooo oouau “noon no no vacuuov ow and: augouonuouuooo < .o .uoououoaoo o and: ooohouaao cool no noun» no acouuooaaoo huoo ovanuou ou ooouuvu no. nude: .n .Hooon huuuosuoo voouo- onu oo ow Axuomoo on: how o>quoooxo mowzo on» on ouuooou 0:: unovuoousuoouo o .ouou-xo won .~o>on auuuosuoo unuwu «so no on on woouuov ow ox: o>uuoouno wouso 0:» Iowa noon voyages: own o~o>o~ suquosuo< .v .nouuoou no one. o.H-wouuuo on» ad vovofluou on: noon-«non no huoooo on» on noon.» on: ooohoaolu .Aououuuo one no douuooo «a non- osu nova-Hoodoo ow non-o ouvuonoo no .naoood huuquuoa no woo-lo nouns Honouuuouuon cone. asau> nauuquauuanaom mmuuouu noun unoooouom ouooIou-um flidogflh uuonu couscous-aha acu mo nuanced owned «0 cownoooono can .anooh nouna on vouovOnn unouoo>ou we onaooae owned mo uuooanno can on out onooo wounon Ioouocqa Havoc: Hounoo uneawxonooo no: noon nu cacao: vaoo ovoox no noon on: .voooomuo no wonananuoou nosuosa no oooavnowon .onouh as. no>o unqoaooo ounnoucw uuaou goon no noaoaa unuuonn on» connooov ads» on nobod- on» on enouon .cnuo» osu noon Announouos an.» onooo soon on no vanuuounaoo on: onoou Asonoooon nuance voov unoloonoooo nonvonu woo gonna-on no nooono noouuuwo canan «mmwanncnuon oooaouolo no noaasz onuu uhunuumxoc Ao nounonunv vonooqocou 0A vnoozo ncannn new .nounonunv vonovnocou on unaon. nonnnn gen can ecu eonuunesn on nounuanouucoamon no canon no oaooo can on oucononuuo nonununomno o .o~o>oa Anwnosnoq noonouunv Adoonovuoooo do On. mouvnoa oaao ozn gnu: ooannn non «H .nounouunv onnoc one «abandons can on ooonoaneo no noonnuoaw dunno. on» an oo>o .odnun van. on» an vonoaoo on canon: has» .~o>o~ hnwnoznon anon can haunoanxonaao no on- wcnono: coon can can: «wanna can um .voo: on nuns nooaomcon «Han Inconuoou .no>oao= .oannn now ononaauu o no venoaou on nu noznosa noonenonou nooohoanlo no noaooo nusaunz can on vacuoo can cannn non ago no wonvnoa can .nanoaou oooouoau nuns: oanuu no“ canon. can noon: wound. on .noonnooou noonouuqo anounon anon Iooao on. congenolo o.»nca¢ou ago no neon :0: .Au no unannouoo any nonnona coma noznqn AAA-snow anon oozouowonu-«v hooalou can nuance: so on anaesaao och .ooo» puooaoanlu noon can oovoaoou loans» no coon nuance on» on noooonouuuv noun) oannn non canon. «so nova: wound-none on cognnn non ozn tonne; aooa no: done on. ooouonoao o.noaalou osn no noon so: .0« too» on vases. nonaoool can .hnwonnuonao no one. osn non .oounun son an coo:- unouuus no onco- hn soon-o0 one no one» nxodoso noon loonno non oan no noun-nooono no cannon nloo o3» noon nennnn Aon noononnnv soon so: .ou oan and: sonuoonn on. ooounooov oooaa honor. .onoouonxoono no .onooso gluon .ou one-no Hooch .su colon onouoa nonconno econo- uonon .on ooaooson dunno. Manon .na oouau uo—nou noounoo noun: unoo«n«oduon nounwn non wool. nonnonnoooooo connnn non no noun-noaoau non-n no con-«pan an. 13:82 oo—noonnnououu 215 nooohonolo no noaono antenna onn on vooooo one coconoon guns: onnnn no“ onuono ogn nova: eonnno noono on. nounonnlo a.ponazou osn no nous no: .nu nnooxonnou noon «an oovonoon gonna onnnn non canon. on» news: nonnn-oono on. ooononnao o.n¢-nloo can no noel so: .ou «nonnao noon noon in 3% and: non neon-=2. nal 8: .2 honor. .onovnosxoono no .onoooo nonoh .cn one... «anon .sn con-o ononoa conceal. n-oooo donor .0" soaoooon noose. «anon .nn udnau lannou unanaoo voocnucouv 1 annoy on can: .novno vno: on oooonon 1nnv anoo onn nn one. can oonovnonoo on anaono renown non .nnnnc anon no .oonn Ionmnuoo xoa .cOnnuuon acoaaoo .oona Hon ansoonnoom n on cocononon on ouoononnnv undo oz» nn noononnno vonovnoooo on no: wnoosa nza sonnooon no .auoUOna .nuooona noononnnv a on ouounonon on ouoononnnv snoo can nn oo>o noononnnv vonovnncou on snooze meannn non .noununnnv ounovnneou on vases. nonnnn new o;n one vonnunvon on nonnnnnnnnooaaon no «noon no snoo- o=n on cocononnnv nooonnnouno o .ano>on annnonuoo noononnnv Anoqnovnooou no on. wonvnoz coco on» sun: nonnnn now nn .ncononnnv unnoc «no unbnnnuon on» on cochonnao no noonnuann noonoo can nn coon .onnnu can. on» no voucooo an annex: nozn .no>on annnosnoo anon can anonaanuonaao no one ncnunoa damn osn nun: uannnn can nH .voo: us noon naoaomcon 0H:- Inocnnoou .noooaoz .onnnn new ononaaoo I an vonaaoo on an noznoga noonanonov onnnn non can no monvnoa on“ .nanooom an .uoOnuuoon noononnnv finennoa anon no unannouoo osn nonnozz cnnn noenon Annoanon noon nonunowonunnv noon-on osn nosnoc: no on oncogene use .ooon Iona-o no upon Hosnoo on» an neocononnnv on nonnnn non ocn cannon noon no: none 1:6» on vnooso nonooool osn .nnnbnnUOnao no oxen on» non .oonnnn non on .0030 unonnnu no nono- ha noon-co can no onon Ivonne non ozn no conuonoaono no oonnov on» gun: connobon on. noonnooov ooosh oodau HIGOnnnonnon nannnn non wool. cannonunooooo nonnnn now no nonnonoaonu “OJ-A no flononbna can. nonuooonn IonuIOnnnooonu 2115 .o oo ooooooon «an onooo .oonnouon ooo snoo on onusn oogal .x no con voo on ooonnouoo on oouooanon ooooo oao son «on onooo .nonnooo no oooo o.nonunnno oon on vooonoon ono noon-nooo no snoaoo osn on nnooon co: oooaonoou .nononnno can no nonnooo no oooo oan nonnonoonoo on oonoo snoooeoo unononon no .o~o>oa mnwnoonoo no nonaoo onn on onosn ono Anaconoonooo nnoon voo oooononv «on monnonoonoo on ono>on ononoooo no .HOUHtflag gHUU-avou“ unfldIU.uflu ”3‘s 3°: o0“ luuvfigou 003 “H. .UQQUCHIQ‘ HO .“fiuafiflv .onoononooo no oonnoooo ooonoon noo oo non oonnononooo ooonuon no>nn=u Iono nonsu osn no nonnooo no oooo can on nos: .nu nooonnooon hoooaoo noon-no nool oon hooooo on oonoVnoooo ono ononunnno no»: «.23.. did a; ooon .nonunnno oo. oo nonnnooo oooooo nonnoononnno nnoo ooonnooon o5» oooonoa oonnl on ouoonono osn on nos: .nn on onoononooo no oonnoooo .nonooon on oooonoa oooonono nlmuuuummuuulfl no non-3o nooznn: voouoo on» sun) oonu .Aoooaou ocn an anon on noon IoUOn on» on soounooo ono ooohonono soon so: .cu osn nn oo>o oooonoon on non unooxo ool noonoo o no ounnno ono>nno ooh .nonono .N no A on an oOnumooo . on ooooooon son non cuss _ .u oo non voo nu ooonnoosv on one nooooon oovoo woo oon oon son on noo onoozo mooaonoeo noooeou noo .3258. no 2:13 :23... 2: 5n: 82 8.. 33.. 2.23 2:. are: 2: .3: Iouon one 0» ooounooo ono ooonounlo noon oou .ns onoon noonoonnoou o oo ononooo suns: ounuoowo no moonnooon one .vooomuon on onoono connooon hooaacu nuan: moo no .oOnnonnnnonEoo nonocou .nonoooon .monoo onooo .n on «N oonnoooo on ooooooon oon oan oonz .3:- «any .uonnonuonoooa new ooOnnxuon .Anoonno «noun ooooa no oo connnoo oonunbauoo hoonloo ono o nauuxov zooaooo can no vonnonnoou no CoonUIOOA QOnonotoo Addonuoao mail :08 u< .NN voouo no: Rona oooanoa noon on onus» nn apoo enonoooo vonovnooou ono ooonnooon .nonoUHoon on oonnnnnonoooaoon no onooo no ooouo «on on oooonomnno noocnnnonno o ooono noononnno vonoonocou on snoozo oonnnn won .oaou noon on .ono>on nnnnonnoo noononnno oo nooooo oonnnn oz» «mono: oonnnn non on oonoa anononono nonnano nozno no .nonooo .nonoan «one: osn onoo Inn .onononozh .oonnnn ononoooo oonoono Icon on nn: onooso onnnn aoa osoo ogn no oooonn noononnno .Aooon no connonnoonon anno Hoonnno>v ono>on anonoo no .hnnnon noon .oonono no oonnonnoononnno no oonwoo ozn ooon noouon Anson no connonnoononnnv Honooonnoov onznuonno non onn no oonu uononono no oonwov ozn oonlnonov on on oonnoooo oz» no ooconao non ouonn .onnnn son on: noooo noo oo .oonoooo nooonno oonnnn eon onw,aoo:mwo unnoo on non o.oo ooonnoo Mon nooonunonnoo ooonnouon nooao oooAOHQEn no oonnonnoouoou ooonnooon no 83.38838 oononooona «32853 oonuoonnnooo—u 217 Aonoono>on canon-nanosn onoonnv nonnxnoa can on noonoou ago u: too: noolonovo no ouono unuoIOnoo noon och .A Aoonnuooonnv nannxnoa can on boooloo 0:» so too: noolnnoro oxn no anon ask .- ”aonnnunov anonon nouoo o.oo nonnononou nonzonnon osn no suns: .nn nuonnoouona nognnon non onnoo co:- noouo on nonu>nnoa sun-unnnn on onnoo oonnoooono oaoao 0:» no nonooa nonnos o now nounoo osn no nonnnooono nos: .an .x no on woo Auuoou voo :znv mm noonn nouov gnon unoom .H on ma oOnnouno on neuron you can: «mono: oonnunvono noono on ouuovono nnuon no noon nu>nnov ooona oouon oonn uuovono guns: on unno: o>ono osn no noon no: .nn .nom nonnou=vv non-no ooxon oonnusvonn gonna" 0:» 0n unonon on: oooaonnao can no hon: no: .Nn. - nAonnoo nonnononov coo: onosvono song: on nn n no Ann nonnauown onn on nnooun on: ooononolo ozn no noon 3o: .nn. unooov on: no o>nnouono . n . .-. .. .. nongu on» On unoaon noon; unnuooo .nosno o.oo: none: you no suoo coo-o: ssoonwouo onoum .naonn0n no on oonn uomm noonnooon nouoo oaou .nnnnooon one: non-a: onuavonm nAnnonoo on: new o>nnooouo nonzu us» on uncoon noon: o.oo: Aooonon>nvv onno: on» on oonnnnnAnooon noon no non-ounooo oon non onooo non on noon .on .n on. n ouoanon anon Inoov oo oonnonnnonuooo no ooonwnv ononvolnonon none-unnoaoo oonnuooonn oonnunonnn ha voon>noaoo sounvon nooooo oo coo A oo ooonnonuooo on noooooo oon nonnoonvon nosooo no onoun nooomo—olo oonzn nooon one on unonon on: ooononnlo oon no noon on: .nu ...—a lflu ‘8 IguUOOa Anson—3 onnoo too; guano—o on: nov gnu—638 nonsu . ul 1 . I u u a u I .noonpnoaoo o gun: ooonounao onol no oonnn no oobnnuonnoo Anoo ovouuon on voonnov ono onnob lulu-Inuit non“ on onnoo coo: nhnoaov on: nov onnnouono nonou“.onno= oonoo woo .nonoosunon .oonnuovono «no ooonuon nook .ooooon nuooona oon gonzo ouoonon onnoo ono Annno: nnono on vooonoo now onnos nonuonooo .noonnnoaoo o sun. ooononnoo one! no.oon:u no noonnuonnou undo ovonuon on voonnov ono ounoa .ooownuoon on» non .ouao oon nooou ono ooonnooon oonnuooono noononnno boo: so: .on .ooonnuoon no anoo oonsnoaoo onoon>non=o oonnoamwno oonnunonnn noovn>noon oo:3 noononoaoo o no no onnoo woos Annaooo on: non obnnooono nonzu sun: ooononnlo onol no oonzu no ooonnoou Inou hnoo ovonoon on ooonnov ono onnob Iona. Hooonunonnon oonnonnonon nonou noognnn Hobon nooood 823333..» 3.5 III sonnoonnnooodu 218 .nnowonoo noon no oooon oonn nonnnw oonn nonnonoonooo nnocn noon oooownno on onus» oood too .oonono noonon sons: oonnooonao .n oo ooonoool wonnonnooouooon owoo mums: lmnwou , nooononooo noono non noooq no oonnnoooo ooononooo soon so: .oc ooooono .oonnooonoo ooooono ooonnoo oonnonuunoow noouonono nonoooonn noooow .nc on wowonuon oonnonoonoow woo oonn :oowono on wonnmmm nonnonoonoow oonoon ooooono nnonoo woo onoa nonooo «ouch .nc—n woon nonwonono .nonnoou woonn «ones .wo onnon uncon>nomoa oonononoozu noonna nuamrm uoflhlou nonconon noono nonnooo oooooxo anon Ion woo omoa no vuonsnw oonnonoonoow ooamaooo non nooo Ionooo nonnooo on. nononunonnooo nooonnonooo onn on we: soon .~.ono [loo oon ooow ononnnooohn onnnuono aooo so: .om— .ooonsooo ouooo noun noooonsoo anon - uooouo woo onononaou nno oo oonnonoono ooooono onoownoa nonooooo nooooo .onon .mn now woo Honoon noooonooo nooooo owanoon Renown: .oonnooon nouns-on one nonnnonooo anoo oonn wooooonan ooononoo noon: woo nonoooo oooooooo noono hoo woo .Anna oonnoooooo uonn nonwonoonv ooooouo nonoooou nooooo nonoh .wn on woo: oonnonoo .ooononooo noon-on :- no 8:38 .33.: .a 82.33 ononaooou nooonaono ooooowo anoawnoo nononooo on oonnnwwo on woouonnono oAn oo opo: nooooou one ooow ononoonoo nos: .nm onoouoon n, .AoUU. .oon .nnnunnnoono .nno .nooo. canon-noooo nunooo oooonooo on wononon ooooonuo nooooo «onon .on A.olonoo oooo on noaooo ononnoonnno ononnuw wonncnnooo nonooooo xooowoon no nonnoonwo woo nonnooooasnnoo oonuno noonon gong: oonnooona< . ooonouoo nonwoonnnnoo woo oonnooonoo onunuuonuonn onoon woo ooonaool wonooon a}: 1.3.3.... ...... .... .3...- oooI uno- § 33% o‘ "H en 0‘ NM CU! ononnnoonnn unnnoonu oooonoo nonooaou noun ooanoonnonooooo oooooooo nonooo oonnoonunno none-n 219 oonou umnwou noonononw nonnonooo ouoo no no>on Annnoonoo noooon oon ooooo unononwooon nonon one oo onoooo no non-so nouon osn on non: .so “wooonnoouv otn onnsz nooon nnnnn osn oo.oa no- oonon>nw nonnonuonaooo ozn on onoxno: oonnoowono .onoaowo non .ooozonaao anoo |n>nooooooo on non osoo woo onoon>noooo oonnunonnn on goo non moononoao ooozn no ooom .wonouon ono ooononoao nnoo un>noaooooo on» no noon onocu no>on osn no nonooo on; roan ooon ooo wonoaooo on Bong: Ho>on osn oo oooaonooo «no «wonoon .ono .onnoo oonoo woo .uonuogonao .oonn nonwono nno owoaoon noon .ooooon noaw noononnnw nonnonono oooo no «onus nunnoonoo noaoon on» no onoooo no non-so nonon oon on noon .0: ton; one song: susonzn onnoo ono «onno: nnono 0n woooono oov unno: nonnonono .ooononnflo no>on onnnn ooonunoaon woo sunnn nogno on. gun: onoo ansnoooo nnonu woo nooannooow Honnoou An undone osn on noononooo no>onunoonon oon owonoon .oaaeoxo ongn non .nonon :uono can oo on no! .oonon>nw wonnonoonoooo ozn on woownooo .nooannooow nonnoou ha In—ooo onn on noohonoao no>onnnoooon on» :3: no»: 5.: 2: 8 3 no. 83> Inw monnonoonaooo osn on onoxno: oonnoow Iona .ongaoxo now .wououon cno oooaono loo hnoenauozoaooo noon onus) «ooon osn zones ono>on oo ooozonaco nno woo Auo>on anon no on non onoon>nonoo «soov wonouon ono aooaonoeo znoon>nooooooo any no noo- onoo: no>un eon oo ooononooo «no owonuon naonoonoonooo nnoon woo oooononv onoononoooo oonuoowonn oonnlnonnn uno no nonnooo no oooo ooonooo on» on non: .nc .ummmmuumm .nonnoou no son. o.nonunnno oon on wuwonuon ono noon-nooo no snooow oon on unooon oz: oooaonoou .nononnno oon no nonnoou no ooao on» nonuonoonoo on oonoonwnonoo no .onooon nunnoonoo no no» 1o5o eon nonnonoonoo on ono>on ononoooo wonownoooo goo ono ouoouonooo no oonnon now oooh .nononnno osn oo oonnnoos oooo on» nuance on wonownooou ono onononnno on onoononooo no oonnooow .nonoooo on 33. 1823.8 nonnooo no oooo oonnunonnn flmmuoonnnooodu 2220 ,.£onooaon noxnoo no mmonoosoon nonnonoon -ooo noonno on snow oanno .onna ooooo oomonaeo ooo noonn no ooo aoooeou o5» nn nouaoonooow woo :unooo oon non ouoao osn noose .wonnowonuo cannuowonn on ononHonuouo non ooooo monsoon: osn xuonu ooo on .monooono omoonamoon on huow omanoonno noon: nonoq noonnono unnooonoowono ooo: ooononuoo noon so: .cn oononoao ooo noonn no no: noooaoo can nn Anoo mono toon; non ouooo as» xuosu mono ...onu .onoahon oo zoom .onoomo nonwooaowon no Aaoooooo ozn on ooCnnnooo nozno wnoo oano noon ooonoov ononuonnw no wnoon can no onoauoo an woononnoo nouoooool noononooo oonnonnononowo nonnooo nowno noononooo ononoonoo noonooouuo nonconoonn nonoonnoo oooonnonw noononoooo moo annonooon wowonoxa nnnonno «omen onnonno nonooooonnoou nonooonm tllllu woxooso ooonnonon unnnom nonooouono onoo.llllu non-5o ooonuonon ooxonmou.lllll ooooonm.llllu nool nonnooonuou Ina—onow woo nonoooou dooooonon noonnuoon noon on know ononoonno oooao oononnoo ooo nooon no o>oo noooooo ooo ooow oconuoaan no... oonaonnon on» no gonna non .ao ooonnuosn uwonoon no: on .onoaonauo anomaou no woanonnoa unannooon unoo «wonuon .onnnucon noon on snow ononoonno coon: ooaonooo ooo nooon no ooo noonncu usn sung: non ooonnooon nnono no noooao «one» ozn on ooonnuoon on: no ooo snoo owonu con .oonnuoon nnono ooo oonn onol non woxuono on noownnnwon oooo on» ooo: oonao noon nu. noooxonooo noaon noonnwn nonnuuon oonnuow nononoao ooooa no: nononooo onooooo nouonooo oooooonon nononooo wnoonsunnao nonoAnnnoo noooooono nowno nnononooo onnnooooo unononooo anono nonnooouoo n« on noonocn ooonn tonnoo no .nozoonn uooono .xnono wnoa ogn sun: onnnn zoo wowonuon .oowonuxo ooon o>os sons: oonnnn oonnnownoa onoo woo wowonuon oooo o>os sons: aonnnn no monnonn «on on onon noonnonu anon osn onooonnow on no: noon any ooosn connoooooo on woomnoow ooonsool no oonnonooo eon woo .oooowooooonnoo .wonnnn wnouon .uonooox wnooon oonnoon gun: wo>no>on noo owonoou ooon «oonnono . .60 w u Uflflu .oonuuozn oo nonwonn o wow-nonnoo .noowono osn oo an non.— o nonlnonnon .noo'oo 2: no noowonn oon oo unconnw 9.3.8: on: .32: no nooo woono anaconnw unoa hunnoonwno ooononolo sool.:o- .oc oaa ooononaao ono ooohonooo noaon noonnn umnwmummm .ooononooo nooonasnnoon noono on» sun: noncon>noo uoo oonntnonnn noo ono moon smooon oo>on noolnnooow nnonoo on» on ooononooo nooon Inoouon osn owonoon .onaoooo onsn non .uooon onnoon oon oo on so: .oonononw noun-noonoooo on. on wooonooo .nooonnoo now snonoo on» on ooononooo muwonnnoooon 3m... onnon noonnonu can-n no... .ann no anon... noon. oonnoonnnooodu 221 nnooohonooo no noon nonnn nnozn nonnow o>nooon unnoo .oonnuoon anoo ooononooo noaon noonnw ow noooaoo oon onsnnl oonnoowono o nonononnoo .noawono ozn oo Anonononn son on. on nonwonoon noon nonononn noonnnoonnw nonxnoz oonn nnoon no noon woooo anon no Anoononnovo oonn «Hang oxooa noon no: .nn on: oooaonnso ono ooononoao nonon noonnn nonononh nooonnow .oonnocowo noonon no «noon on noononnooo no .<.n ooo: ooononooo zoos no: .on o>oo on wonownoooo on oonmow noononnooofi anonooo no .<.n o sun) ooonoa o woo oonnooowo , nannooow on: non o>nnsuoxo nonou noonon no onoon an ooo: on wonownoooo on oon on nnonon ooo ooohonolo oon no :oanooowo onoowonm noosoo owns o .wnoaoo Aoonnon noson noognna noonon no onoon no non-5o ononooo «on on non: .nnnnowonx monwznuxon noozoo snonoooono oonn wonoaoo «no oonnooowo noonon no onoo» nooobonnoo noaon noonnw nonnonoo nocwmnmol n Nun—s .. .. a u u a - - inoo ono oonnooowo «oonon no onoon nool.:o: .nn .oonnooon oonnoowono o nonononnoo .noowono can on nunuonnw nonxnoo oonn nnoon no nooo woman ooo ooononoso ono oooaonoao noaon noonno noson nooooa oonnooowo noonon nooooonon no oooonnon .wonooon ono oonon>nw can on ooononooo Anon In>nooaoooo nooo onosa nooon can on ooow oonononw :ooo smoonon onooon no nooooo osn noooo .Aonoono no>o Annnoonoo oo o>on ozav anunnnnoo ocn no sonnoa oon no onoxnoo mo>on ononoooo o oo nooou .odo>on mnnnoonoo no noaaoo osn nonnloo ooo: mononon>nw ado now ~mmnun>nw nooaooow .«ooon nunnonnoo woOuoo oon ononond uoo: no on Annooow on: now o>nnouooo nonoo nno>noonoonv oooho~1lo on» on annouon on: noownoonolouno o .onn . donoutuooaon woo obnuooono nongo osn ooo)» leono non .no>on Annnconoo nonnn osn oo no: onoAu ono Annnoonoo no onooon nooo no: .nn on on woonnow on as: o>nnoooxo nongu ogn oonn ozow wonooooo ono onooon unnnosno< .nonnooo no oooo o.nonunnno on» on wowonoon on: noononooo no unooow oon on nnooon oz) oooaoaoou .nononnno onn no nonnooo no oooo on» monnonoonou on oonoonwnonoo no .ono>on annnoznoo no noaooo oon nonnonoonou on ono>on ononoooo wonownoooo noo ono onoononooo no oonn: now oosh .nonunnno oon oo oonnnooo oooo oon nnoooo on wonownoooo ono onononnno on onoononooo no oonnooow .nonooon on FIIIII oono- nooonnnonnoo oOnofl .anou ooonnoo an _ II“ a. 8%000'3 onus-n nnnnoonao E Bonuoonnnooonu 222 .ouomonaao and. on non ozu non nononoan on vnaono bean .oon +no~oo woo noun: «ooo:- nouou an wovoauon ono ounnooon amonnn noxvoo, monsoon nu .Anc can» Iooovv oonnouoo woo noun: «anon. «anon no unnon oloo can no venou- on oasoao oonnouoo 19o non-a In N N IILI on u u a: n noun-«ow no. no a: ooo a . ..oonuna-c .au no nulls. coco unnuna o.o- oan on anon ono nonnonoo coo scoot goal .oo .ooonoaalo conga nooou no no sung: noon vo- o>nuouono nonno on» on nnooon on: 0.05» .cu noun-ooo an oooon . ooohoaolu can non conunooc onsn noaoo< no uonuoauu . oooou ooaonnlu non «undone ..nnoaon .nannn unoo aorood ouono>< onnnnn noonnn: :.ooaounlu nobnnooono nongo oz» on nnoaon oz: ooo non onnnoooa onannn nooo «cacao om- )honolo no snouoo «ooooo ononovo ozn on nos: .aa uno>o= ooo .oanoa no ouunocop omonnn noo _ .soxnca nonunooo no can» nnoaona on» noodncn on sou-one. on unaono nuuon manna: .ounnnopo non ovouuon .voxnn; onoo; no nooolo Hoonoo on» non QOuqu IA vanes. Anaenn unnna Anon-u Hanson ononooo an no; can» Iooo> ovoaucu .acooaou on» an unon an Ina-no: non annnonoauo, ooonu ooo .uoonn noou Aconcov an connovon ooosu .oouou anonhoo on so:- .:I~ ha connocon ooonn unnnooomkmummmm onnnooon owonnn no noon on» on ooonuau £655 .Lllllclu noooncaooo non-n. .oounooon oonu noonnv no nnoaoo dooooo onono»o osu on nos: .an. loovona o non-nonnoo .noovonn 0:» no nu oononmam mum ounnmmon ouonnn noou ~o:oo< owono>< 1.0:: 0.0:» on oonnonoo vouuoon non-a nooooc an I ..I I | I a ....uoonno using on: nu...» no noo- 33o. «noon-on on» no vononoio oooa «no: Anuonov oat ooononolo ono ooononnlo non-u noonna nun non o>nn=ooxo nongo one on unoaon o3: cosmono Ilo noon onoon no non-ac ouono>o osu on nos: .sn ounce-on OAn n; wounds-o noon ooo: ooononolo neg-n noon sum noun onoou no noaaao coon-so can on non: .on ummflmmumm c n0nax uncann canon noduooooaooo no oonnonnooooou nunnonuo< no nonnounuonnooo n.»-n ».og.«a moron noooon ounnoaia nude. ononooo «ooou nooanua moron nooaoa un.n.. «op-n o.oguna «ovou oooaou oooonnoomml nonu.unun..anu 223 .5333 2: 3 vooooo ouoanouol no coco doses: #33. .co 29.3 loan ononnonol .I I I I I I l I I I I no noon lonn 2.35 I 08.36 .uooosunan unnnoonvno no: as." a «noun no oooozunoo nonuono noon-ooo no: 2: w on .ouoolonnavon noonnoo non ooo-non: wolozunom no 15.-ooo on nonon 33:2... oooan. ..I I I I .I I II ~33 noon?— “noooo ouonnouol no oooosunoo o‘naoo ooo noun: Aopnuououo nonnu o5» Ionn ESu none—5o“; no.5.— hunnosuao nooaon oz» on no.5 .3 b—- «condos—u oonno 12.-3o 53 53:: Aofiuooouo nous» 0:» .5: .53 9.3553 no»: munnoauao uooaon 23 on “on: .3 3 n Fill IIIIII .— udflunflnrflu N nun—=5 no noon.“ nouns Sou—=3 2.93:2: 425." ~$7.85.: voouun o5 u 23 no u , no.5: oonaunvcooxo unounoogn no on guano.“ on: n3 03260.3 nonau u no.3 oponnao :3 non...) $32.93 no}... on» aonn :33 22 on 3.8%... on: noovnoonnnou? o .01 us—nS—ouv «0.5.— »unnoru... goal-H «En on nos: .no Iflono no.“ 40>: nunnosuao non: can no on on voonnov on 0:: 953530 none—u 2.» Ionn ESv venue-3: on: 33>: 3205.5 noon noon manna. oonaanvoonuo nounmoo Houon. .3— .Honnoou no noon nanonunnno o5 can w non-33.5.3 no {425” hunnosunu no non uaolnognn II...- 23 nonuonagou 5 3.25.— anon-Boo as; «ooze.— vonogogu no: on: ouuouonooo no noun...— "noqoo oonounvooeqo goon-oz: ..ov 3E. ..nonunnno o5 no 5312. oooo ovonano son £03! 7558!. nongu on» lonn £51 on» 59.096 on vonovnoflou Ono ondnuunno Gonna-Soy no»: ounncauoo «col: 9: on nos .3 on Soon-«ooo no 3.3—Eon .uonooou on ooaoo nonvoo oluonnoo oonoo nooonnnonnon nouns—u:- nun oosunsm anus»..- 9:3..— hung—So uooluoobuu dong-u: nooouo 224 .u oo an onaooo Ion Don oz» unauo ooo o oo an nonoaoon vovou use woo «n oocoaoon :on osu onouo .oa on an oonn Iouzv on noaouo onu coca ." l ' ' .oon so nMo connooow ooo .ononooa. «macoonon gun: vo>no>cn. oowoa no nopaoc ozu 4o on oonumoaa .noaooa as» an mono; no nonanc none». as» oo nn acnuooac onouo. .oonavooono no noauoa mono Icon on» an coconucn ono oonavouonn noaaoonon coca .« ooo a ouoouoa onoanuov no nonuonnvno Iooouo no ooonmov «gonna: Inouzn onoum .n oo count Inrucouo no: on ncoaaoo on» nowuounVun nuoooo no ooo A oo vonnvnoocono on noooaoo oAn noun Ioon‘ou nosooo oo onoom naovnosool oonsvooona one no onoa ononopo no no ono ovno: noon no: .Nn noonou Iooono no Hoodoo oAu on ono oooon nool no: .nn .Iosu no duo on oonom oAn ooo .ouooool ooo noon onol ono ouonn nos: nona- nonvou noon-o: oz.IIIII 33a .3 III noonooooono no nooooo o onoa noognoo ooo ooo: .on non ”a IIIIII .noonononoo oouon «on o sun: ooonouaao onoa no oonzn no ooonn nooouonn nonuuoquonvaa onnIoonnoo nnoo ooauoon on voonnov ono ounoa on ouoononnnoo ooononolo oonan nooon no no ounoo ooo: ooo Anuooov on: non o>na=ooxo nonzo ago on .oooogonoo nannocnvno no: unonon cs) ooonouolo onu sonon ooononoao on .ao olonn no oooozonoo nonooco novnoooo ooo on .ouooaonnscon noonnou non oooozunoo nonnu: nnonn non no nooonooo on nonon oconuooov oooah oaooooono nonooa nnoaoo ooononouo oongu nooon no no onus: ooo: ooo Annoooo on: non obnnooooo nonso .Hooon nnnnonuoo voouoo ozu no one on uncoon on) ooononalo oAu no noon to: .on on Anuoooe on: non o>nuaooxo nongu one on ounooon on) unannoonoIoon> o .onolo Ixo non .~o>ou nunnosuoo nonnn ogu no on vocnnov on as: o>nuouoxo noncu on» ma oonn ooo: oonosaoo ono ono>on nnnnonua< n a .Honnoou no oooo o.~ononnno o£n ooonunsn N noou on eovouoon ono noon-«ooo no nnoooe on» on nopoa nooaoa o» unomon oz) ooonononu .nononnno osn u nnooo ononnonoo nooooo nonou no a no nonnoou no coma ozu nonoonounoo on nooo nova: oonunooo ooo odonnonol no oooosonon oonocnvnonao no .ouo>oa nnnnozuoo no non. ooonooo ooo sons: noonuouono nonao ooo oonn oooo Iooo osn ocnnoaounoo on ono>on ooonoooo noun-soon uooon nunnosnoo noooou on» on non: .no oonoonoooo ooo ono ouoouonooo no oonnoo loo oogh .nononnno osn oo conunooo oooo on» nnaouo on nonovnoooo ono ononunnno on ouoononooo no oonnoooo .uonooon on son» nous. documunonnon oonovooonn ..hUGOOOhh no.lon«oo«¢nouaouu lonuoawununon ooo} nonuoUnnnooono 22_s .x 33.: on» onouo .nnnononon sonny no: ono no nae-Jo 2: an no: .33 on noon—3 as» onoum .a oo 2 once: on» onouo .3 on an :35 I325 on nonnoo 9.: :25 I .I domnloml nln oonuooor ooo .3033 Icon of on 2632.: ooo oonauouonn “quomnWoloflsfl .~ voo « 5.03qu :33: «o oonuonnvnov Ioooo no noonnoa ouonooo Inonon «noum .n no you; Inovoouo no: on noooflou onu monuoouvon no>ooo oo v:- a oo noon—Koo Ioono on nooaloo 33 no: IonnuanIWoaooo no onoun ond—E 93100 on nouoooocnno monononnon oonuaonnou ouo Innonoo noonloo 23 :2. 23:: no: .5: ...u; .3 li'l‘llJ Zoo-non? no ouuonuoou oounnna no oonozou. ono ooonoualo non: «993.. of no bin .5: .8rII l n noelno meIolnurloM uloNuMofl ..u convul— na sonoso ono ooononolo noo- ao: .an . l:.l|-l.l.l.lllL noouon ooonouolo non!— uoonno ono noooovonn no...— nug .3 oz IIII. our non-ton...— 4 nod—non no vouon ooononolo nos: 33:. 9: .2 I’l'l'l’l .333: oz II . oon nlnon cannon «goo-n2. vonnvnoooouo o 25.. noon-co o5 32— .on :3 nooool sonovouonn Hooooonoo ooo.— onono>o oo oo ono ovno: noo- aoz .3. .....I on noonovooons Hoocoonoo no noon-- 2.» on ono oonoo noo- 30: .2 39.25 oz goon oon nonnovooonn ooooonoo no nooool o o»!- nooioo on» oooa .2 I IIJIIHIIII flue-Java 63:: o oo gnu-«non: ouonoooo :uoo noooo .oonuonuouoo 03:15:: on on onosu ooo noooloo o5 no 3:! gnu-5515 noononnno our. nnouonoooo ononnonoo on I: S.» no odouou noononnnv :25 _ . ouoonuooo " noolnou‘ . _ I I II I I I. .oonuoflon oonuoovonn o nonlnonnoa 5262.. 33 no nnuuonns mono—non on: nnoou no noo- oooso o...— ooonono' ono ooononlo nooou noon:— .I I l I I I I . . _ . noun-n nooooonon .85 no So on oonon o5 loo .ouoaol ooo no.3 95! 9: on!» :3! noon-l oonovoo Ions nooooonom oonuoonnnooowu ...—nal unconuuflnnoa 226 303-: oz vaium no» nnonuooo .nonnonoooonn anoonsu Iona no omooo nonnonuooo ooo oonon on oooooo connonnooo anouoo noo ooo: non noon: on .olnon zoo «no no oonon no ooo-no o ooo: noon-on ooo oooo .«m noun-unuoauoo no“ nonnnocon ooo oounono Iooono oonon noon-o0 nool non>os onnon nnnoooo .noouo Ioo «onnooo oonon no lonono o nnnonooou noon» oonnonnonoo nanooonaoeonn _ ooouooa unnol nnuooonoooonn.IIIII .oonnooon :onuoooonn .n ooo a oooauoo ooonn nonnonooo nnuooowoooonn o non-nonnoo .noooonn on» oo nunuonno onounnao oo oonnoonono nooononolo nooon noonno nonxnoa oonn nnoau no nooo ooooo on) Iooouo no ooonmoo ouonooo non ooooonoon nnouoo non onooa o5» on non: .no ooononooo ono ooononolo nooon noonno Inonon. onoou .« oo ooo: n... I .I I I I Inoooouo ooo o- noon—ooo H nooononeoo noon.— . .53» non noaon ooo nonnoonoon nooooo oo noonno non onoa» ono oooonn nnonoo nooo ooo .oa . noonno ooo non oooonn nnouoo no non-so ooo n oo oonnonoooono .uwuwuuuuauufiflIIwnonono on» oo on oonnoooc nonnnonon .o noon-cu ago «on» o.oo-so oz Ioonoon noaooo oo onooa ooouu oo» nonoooaoo conuoooono o ooo: noooooo ogu ooon .on soon-no oz.IIIII union no» .oonuuoon connoooono noononnoo nounnno ooo: noon-on o n ooo: . nooonuannuooo non oonunna onoan ono ooononolo noaon noonno noon on: non .no o monononnoo .nuooono 0:» oo nnnuonnv nonanoz oonu nnozu no nool ooooo as) ooonOHan ono ooononalo nonon noonnn nonnooo oonon oooonn nnonon onoooaoo oonnoooonn nononnon nooonuonnoooo non oonnnno onosn ono ooonoaolo nooo ooo non .oo nnnoou oonuoonoonno oo noon. ono ooononooo noon no: .no .ounonnooo oooa nonoa noon oosn noouon nunuoo nooooou no nouuoo o oo ooononn Ina nqaon>nocn on ouuaanuoooo nnnaanon on noon conuoononon oz» .nonxoon oonu Iolnonon oo oo oonnnmooo uncoon nooooo oz. no uoonuooo nonoo on» no naoo < .ooono unnoooa oononoao oun no noooooo nooonn onouxooa oonuoononon ono ooonouooo noon no: on .oo on» noooo oonuoononon oouunna noo oooao Ion on ooonnoo on ouonxooa connoononon ooonnonnuooo ooo onnoao oonuonnoonno { ...anooa gwaanvuflfl nooonoo no onoaooo ono ooonoaooo nooon noonno noon no: .no nooonoo no onoaooo ono ooono ooo noon-o0 nooo no. .uo mmomnmmum nu .oonuooon oonu Ioooono o non-nonnon .nuooona o3» oo nu Inoonno nonanoa olnn nnosu no nooI nooao ooo ooonoaooo ono ooonodooo nooou ooonna mudaa mumummnlnnoa ”gm-Imam 225 . nonnonaooonn Elna—5 oz swoon—S o2: no owooo «ozonuoou ooo vooum on» oar—on on oouuuo ooZonuoou :35 no: Sonuoou ooo: ...—5 2:: un .38 :0: .36 no olnon no Iouono ... go: nooolou on» 33 .3 533325.: on: «333:: ooo oounono Iooouo ounon noon-o... noon non>os onnon Ioo Hana—So olnon no sou-no o nanonooou nnnuooo . nonuo oooun oonnonuonoo nnuooonaonénn III vflouooa unnoo nnuooonlooonn .oonuuoon oonuoovonn .N can a :3...qu vooun nunnonooo nanooonlooonn o nonanonnon 52.60.... 2.» oo nnnuonno onounnuo no nonuonnuno nooonsolo no»: noon: nonxno) oflnu nnogu no noon noon: on) Iooono no ooonmoo ouonooa non ooooonoon nno oo non onooa of on no.5 .3 ooononolo ono ooonouolo nvoH noonnn Inonon unoum .N oo vonnv —I I I I I I Inoooouo no: on noon—ooo nooononoao nose.— . ‘33» non noaou can nonnoonoon noaooo oo noonnv non ononn ono oovonn nnouoo noon :0: 6m . noon: oz» non ooo-nu nnonoo no non...- 1oo H oo cannon-ooo». I I l I l onono>o on» oo oo oonuouoo ooh—noon on noooIou 05 no: voouooa oz '- ..oonaon non-no no 938 2:8 our non-Gono- oonuuovono o 23: noon—loo 05 noon .3 1.3-.... . oz soon on» .oonuonon canon-ooo:— noononnon nonunna ooo: nooolou am» noon . o mongonnon .uuooonn on: :o nnuuanno mono—no: as: :93 no nooa 25% 9.7 noflonunnnuooo nan :onunnh ooononaau ono ooononafio no»: ouonnn onogu ono ooonofiIo ooo-.— uuonno noon .6... non .3 . .onnonngo oooa 9:3 :1. no...» nofion nonnon noooaou no .533. o oo ooonoun Ioo 1:333: on 13:553.. nanoanon o: uni. conga-Eon: SE Jud—ooo oonu Ioanonon on on 3:35.... unoaon 15::- o;.. no uoonuooo oono: 05 no naou < .33: unnoooa ounonnao on.“ no noooaoo :53. 3: noono oonuoononon ouuunn: noo ooonu onogoon nonnolnonon ono ooonounlo noon .8... on. .3 Ian on ooonnoo on 301—ooo oonuoanonon nooonoonnuoov non oounnn: ono...» ono ooonouio noo- ooz non .3 nuns—u oonuoonoonno oo oo>nn ono ooononooo noon .5: .mo nooonoo no .3333 no: uni ono ooonoaolo noaon noon: noo- so: .3 loo—yon; o non-nonnon $02.62. on» no n.— Iuoonno mono—no: on: 5!.» no noo- too..- of ooonouono ono ooononio non!— noon:— nooonoo no onoao. ono ooonouolo noooloo noon won .3 ooonnoIa oonoa nooonnnonnoa an: conga HOfluflou EOE oooonn nnonom 32.2.... oonuoooonn anunaou ooonnnnnuooo non ounano nonnoonoonno 358.. non nolnonon I‘D-I'll. eon nannuooonu 227 onnou umnuou on Inna-n9 anion Avooonuocun nuooono on oouonoooooo ono .odonoooo Ioon one: on nooononoon ooo oo guns: non .oonn uuooono no noooono o non ooooo>on no\ooo ouoou .onouooo nunnnnnoooomon on oouononoouo ono anonoooooon one: on nooo Inpnoon unnnoooo o sung: non ooooo>on no \voo ouocu .ooooonoo nonno non ooooo>on noxooo ouoou wonnononoooo onouooo ooo .onoucoo noooono .onouoou nnnnnnnoooa Ioon ooo oouononoo ooooxoono nooonnooon ooh .oeoanoun .aoun «onus-con no can nonoooon no nonoa on» on ozoo ooxonn on non onouooo nonnonoaoooo ouoo nonoooonn .nonoou oonuonon Ioooo ouoo nonuooonn «ooonon>no oaoo on» non uonooonn on non ounooon oooo>on ooo .uoou .oonnonnnuoou .3onn :ooo .onoao Ixo non .nonooo oaoo ooo non oouoooono onnooon noononnno ononoou nonuononoooo onoo nonuooonn ouonoooo oo noonooou no: on .oooao>on noxooo ouoou oaoo can no nonuononwwo no o~o>on noononnnv onouoou noonoaonoooo ooooooon noxooo onooo ono onouooo nonnononoooo onoo nonoooonn noon so: non .oo ounuonoaoouo ouoo Honoooonn ouonoooo oo noonoooo .nouoou connoaoaouuo onoo «onocoonn o oo nooouuo ooo nnoo noooo noon; noo no ounooon nonooa no .oooa ooo unnono .uooo nonconn nannonnvno nonnoanuo noun so: .na no>o no: on .oooouuo oooooon no ooou ooo coca onon no oouoonoo ogu ooonu Ion onoano noon .ooooo>on no\ooo ouoou oononosouoo nooaaou eon guns: on ono: ono onouoou oonuonoaouoo onoo nonuooonn nouns: ans. Inonnonnooo on ooonnooo.ono ooonodono noon )o: .ca .Amo onon HooOnunonnov onnnuooo ooov ooon cocoonn no nonuoooooo gun: ooo InOHono nozno ”no ooo .Ao onon nooonnno Innoo nonooom on ooonnoo oov ounoo onao Inounonuooo nonuo no ooonoaoso nno .nonu noon: nova.» .onn.unn.nnn «no» noconononnoo onnnooa. on ooonnoo oou noonnno nonoooonn nonsu oonnuonon nunoooconn noonoouo nonnon oonnnno nooonoo ooo ooo: nouno no: .no mum-Ina _ on» no ooooon one: oz» no Aooon ouooonn no nonuoooouo o>os noon nozuosa no ooou oonononv ooononono duo ounoo onaonodnonn Iooo on ooounooo ooononono oo ooo—non oouou nononnnonnoo ononooo nonuouonoouo ouoo «onoooonn ounonon non>noo Ion ooanono Ino conunoaonn onoonnnoooo onaonounonu Iooo on ooono—o Ino oonnnooonn onnm zuhmrmnpn aauhloo oznhzooouo oonoono nonnon ooo onnn nonnnunnnnnnno 2228 .vonoooonn ono nuooooonn noo nun: onnooon noon ooo unnono no noon song: on oouonn ono>on 0on2» one no nouns: onn ocouno onn ooo oonoooono ono onnoaon noon ooo unnono no “ooo guns: on ono>on on» non mongoosronn osn no owono>u oz» nonnn on» onauo .no oooooo Ioon oooOu sun can non .x noaooo on» onouo .ooooon non ono onnooon nn .nnn noaooo one .onnoo ooosn no oooooono noonno ooo nnoo no noon on odooao ononnnoo on .onnoo anoononuonn Iooo oAn non oooooono onononouoo non oooo noon Inou as» non: .wononnon no oonuonuonnoo oo zooo .onnoo osn on oomnosu nnnonnoo ono oooooono noonoocnnov noonnoon no>o onouo .nnnoon ooooon ono ounoaon nn ooo “on noao Ioo ozn o.ouo .nnnnoon oooonn ono onnooon nn an noaooo on» onooo .nnxooz ooo-on ono ounooon nn “— nooooo ozn onooo .nnnoo ooooon ono onnooon nn .oonnoo oonu ooo on onoo no noonso on» noaooo nan onooo .ooonnoo onnn no anoooa ooo no nuooooonn nonoaooon onooooo non gonzo unnumm nn H o-o Inosa ooo .nooaonooo ans-nonnonnooo no nonnonuonoov ooo .oonnoaoo .oonnonoo oo :uoo .onnoo oan no oooooono noonno uno oooauon .onnoo ooooo on“ on ooomnooo no: ooon ouooonn no wonuoooooo nun: ooononono "no no oonnoaoo oz» ooo .An onon noounnnonnoo nonooou on ooo—now oov onnoo annononnonnooo noxno no oooooouo «no .nann.onn.nnn.nnn anon Hooonnnonnoo onnnuono on ooonnoo oon noonnno nonu .onunpnuo. on. nannnhnnaonodunnnnoo ado no noonnano «noon .oo onoo In>noooo oonquonnn noonooo Ioon ouoonx _____ ...-o Ionooon non ___.___ .— uoo.oxnwn.o nun..a ungu=ox an..un o..¢u.a~n no: nuooouonn .on oonoooonn ono nnnnnonoooooon no ooono oonn Ioonoon nnoAn non onnooon ooo“ ooo unnonn no nooo gonna nun. unannoun. ou-nano can on ..aa .n. Cedarfla Iooonn nongo osn no ooooos unoo osn no ooooonxo «no oonnnonnuo ooo onnoo onto Inonnonnooo ado no oooooono oo ooouoou .onnoo nonoo noononnno no oooa oonoo no onnonon onononoo ononoou connonoaoouo onoo nonoooonn ononoooo oo noooo non .oonn nuooono no .oono noonnoonwoow .non Ionooo no oaoo ooxonn .oonoo oonIInooo Inoo oo onnooon no noo o nonooo cannon Ionouuo onoo nonocoonn ooo oo noooo .ono Inoxo non .onnooon noononnno on wonnno Ioonu on ouoo noon anon noon so: no anon Ionomon .nonooo cannononouuo onoo nonuoo Ionn nooonnnooo ooo no oonoooo on onooso unco o non oonononoooo onoo «onooo Ionn oonaInoooaoo .nouoou nonnononouoo onoo «onuooonn ooo oonoonooou on oooo ,Io>on ooo onoou nonnonoaouuo no no: own: Inooaaoo ogn .onoooouono nonuooonn non: Inooonou ononono suns) oonoouaou ndo non .ononooo nonuoonnnooo ooo unnono non nononoaoooo ono ooooo>on ooo oaooo anon ooo ”ononoou oooo>on non oooooaon nnooA nononooo noon non oononoooooo ono onooo nuoo .ononooo nonn35nnuooo ooo .nnnonn .oonooon .noou noo oononoooo vooononn ononn «onoooonn no oaooxoono oar .Aouu ouan «ooonnnonnoo onnnoono ooo “ononooo noonnooo no son nonconoonn ononooo uflnofl mummnunflmnoa nooooronn nnoooo umnuoonnnooono 2239 .mpOn ouooonn ooo nonuooouuo won looxunoacoo no nunnOHoa o Jun) uncannnaov o on conga» nonno unoo on» on. an on no Iaouxxooa no .ucouo:ouoo .oovonoxo coca u>oa gong: connnn oonn Inconos coon ooo wovanuon noon ooo: gonna nonnnn no nonnonn on» on onofi ouoconn ooo wonuooouoo Inc» 0:» onooonnuo on an: noon 05H .oonnal Inoncn «anon-can no cannocnsonono ooo .connonona unonon .monaoooono .mononouun onn gun: oo>no>on on: nan-noon» ooo“ oooocnn ooo wonnoaoou< .onOH ooooonn no monuoooouo new: ooononaflo nosuo no ooosn ooo .ao anon «ooonnnonnoo monsoon on ooonnoo now Inna: unson04nonnoou nonuo no ononn .Aonn.cnn .nnd.nnn ouon «ooonnnonnoo unnnuouo on oocnnoo can .nounnno «onus-onn nonnu on» an ocean: one: use no ooo;u .oa none-ooo on nouns. ozn on octane Ion ooononnao can «no no nonnnn now as» ooonuzn .oonounvon on oonnnnnAnoooooon no «use. no ooouo can on cocononnno nooonnncxna o coon. noononnnu oonounuzoo on unooso canonn on» .oooo nosn on .onooon nunnosnao noononnnv oo noonoo nonnnn as» ooonoo conunn no“ on mono: unouonono nonnlno nunno no .nonooo .noncoe oanoa on» onooun .onononusk .nounnn ouonoooo oonoo onooou on no: canon. nanny non oaoo can no oovonn noononnnv .Aoaon no nonnonnoononnno nounnno>v odooon Anon-o no .Annnonuoo .oonono no none-numwr. xozn unannov .nonnonnoou sonnnu monoosonon nono: can oonuau.u.anau< non: .nonu nn< oomlwmoronnslnb ooo-unnunnb canon» novonoon Iona. umnnou an "undo: ans-nonnonnooo no oooxonolo no nonooo uncanns can on ooouoo one coconucn suns: onnnu aon canon. can nova: vonnnooono on: nuns: Ang- cnonnonnooo on» no ooononouo on“ no soon so: .non nouns: ansnnounonuooo no oooaonnlo nool use oooonnon song: onnnn ooo canono ozu nous: oonnnooono on: onnoo nus- nnounonnoou .An no cosmonaut can no pool so: .oon nonnoa .ngnnonnounoon on anonnno no nonwoo can ooo» nosnon Anson no nonnonucononnnv nouoounnosv anon noonno non can no oonnononono no oonmoo osn cananuuov on on cannooav «an no oooonoo ozn oucnm .onnnu son on; noooo no: ov .ooucooo nvoonno Oman» no“ can oooanon unnoo on non o.oomonaao no nos: .novno ono: on ouoononnnv anoo onn nn olo- osn vonoonoooo on vnaozo conunn non .unnso anca no .co«noomnooo no. .oonu-oon noooaoo .oono naunznonmoon 0 on ouoononon on cocononnno amoo osn nn noononnno canyon-ooo on no: onoogo non oonuoonn no .ooouona .uuovona noononnno o on oucononon on ouoanon Inna Anna can nn oo>u noononnnv oonovno noon on onoozo nonunn non .uoononnno conuonooou on onooso nonunn no“ on» can own-unvon on nonnnnnAnooouoon no unau- no ooouo «an on oooononnno noounnnouno o .cno>on Annnosnao nouncnnnv Anna Inovnoooo no on. undone: can. 0:» sun: nonunn non nn .noononnno cause on. ooonunoon ago on oooxonoao no ooonnooon noonoo as» nn oo>o .onnnu anon osn on ocucoou on vnooso nonn .no>on Annnoguao «so. onu anouoanxonnoo no ono monvnoa can. can sun: oonnnu no“ nu .voo: on non! nooaumoon onscnoonocoo .no>o loo: .onnnn so“ anonomuo o no oouoooo on an nuguoza oocnanonoo onnnn son on» no nononoa can .nonocom on .ooOnn -ocon nounonnno anonnoa anon no ooo-s aouoo onn unsung: o.oo ncsnon annol anon anon cos-namonu-no mono-on on» noon-s: so on nnooznau ugh .ooononoao no anon Ananoo ago on cocoononnnv on connnn non can sense: noon «on nonooon on snooze nuns-ooo on» .nnn>nnuonao no can. con non .oomnnn aon on ooooo unonnnv no nodal a» nous-on on» no anon scann. can can no noun-nuns“. no can... no. can: oo>noson on. neonn-ooc.ooonn non-n..- ono ocean. non nosnonnnv no.- .o: ... noonwuumm ummmm mflflflmnnonnoa noannn AOn Macao oonnonnoouooo counnu non no eons-nonun- noaad no nan-none nnnuonaoou nauoana. nun non-Au fl 230 .o oo ooooooon on» onooo .oonnooon ooo name on onosn 30:3 .n a. 2: 3m wNISwm. Iooov o» noncomoon voooo coo son oz» onouo .N no u on NN canuooow o» cocoa-on :on can con: .' " .aoonnouon o3» do» no noononueo anaonunnonnooo no noses: omono>o us» no snap con ooo non oconnaoov onooo .ooononoao no noaaoo anon on» o>oz new vzo an onennooocv noononoao hooosoo no unoneao uoo stun; on» sun: «connouon noonnou on» 0:» cup: noo»ooon ono Inna: on:- nnoqnonnooo onus: ooonnouoa noon-on noonono nooa on» oooanoa oonnn on ouoonono on» on »os: oooooo on» sun! nonuooon boos-on on» o» nomad-no ono ooononolo ans-nonnon»ooo on» no noon nos .oon .x no men ooo nu ooonnooav o» moocoooon souoo coo :on on» once. .a on Nu nannaoow o» ooooocon tomloh» loom? nooononuao noonnoo no noaaoo noosnn: on» a»n3 connooon noooloo osn o» ooownooo ono ouononnlo onsononnonuoou as» no noon so: .non uwnhhnflnhlfln .x no «on can» loose on oooooanon oovoo woo Ion can once. .N no .n .o on nan oonuoono o» ooooooon )on on» cos: .u no non oonnooav on ooaooooon novoo ooo son on» onooo .n no 0 on «an noun-oar o» ooovoon :on 0:» non: noooaonooo unannounonuooo no non-so uncanna voouoo as» n»n3 oonnouon soooaoo on» o» vooanooo ono ooohonnlo nnaononnonnoou onn no boo! no: .cOn L'Jllul .hOH I..J nooononnlo noon-o0 no nod-no noosunax .noononooso ooooonn no nonuouoooo oo ooo 32.81% 38 «3.3353. 3302.. coon coon ooooonn no non»o:oooo Ann: ooononnlo nnoonpnooooooo cons» noooa »o oo vaunmwo on nnoo onaonouuonnooo 4 .hconaoo as» an anon on noon nn cooo vooonocn on »on nooohonooo ans-nonuonnooo no nonloo snoozo ooaoonoo o no ounnno onoonno as» »oonuna oa» a»«a nonuoooa noon-co onn o» ooou«ooo.oooonoon on non onoogo ouoxonoflo soooaou ono ooononslo unannonnonnooo oa» no noon so: .non nos ono nnono ooozz ooo noooaoo on» can: oodau .mnvou on .333 .onoon nono»uon»:ou o oo ononooo song) nonocoao no ooonuouon och .uovonuon ca onooxo obn»u:=n aoooaou nonno moo no .oon»onuononaoo monsoon .sonoooon .oouoo .connonoonoooo non ooonnoooa .A»oonno o naouuov mooalou oz» 5; ununonuoou no ”cocoonoo oncononnonnooo ono ooonnooon noon oooao no: nos» :oobuoa noon on onosn nn n-oo oonononoo n onuooo on» no soul ooa »< .nou coon» nnoo onononoo oonoonoooo ono ooonnoooa conga nooonnnonnoo ooonnooon ooo-no; oooonono ooonnooou nooao oooAOAaao no nonuonnoouooo unz- unonnon»coo sun: ooonnoo Ion obnunuoono ooonnoood no nonnonnoononnno oononoaono noonnoonnooo sonuoonnnooonu 231 i , .x no «Ou oon»no:o o» noznco oz» on: o no wou con»no=o o» noznoo oz» snoun .nnoa onznnouuon»ooo o no nonnnuoac »uc: o: nu .9v>0uaao nnznnouuon» loo» nmozunz no nounnno uonooocnn nonzu oz» acuoz xuononvoaan uo>ou oz» no nozoao oz» no mOu ocu»no:v o» noaooo oz» ooo u no mou can»no:v o» noaaoo oz» onoun .xoon Ilo» oz» on »n:: nnznnou uuon»ooo buoo oz» ovooz ooaouaao onznnouuon» loo» »noznnz no nonunno unnoooonn nonzo oz» nu oouau nonvoo on .uooou h»nnoz»=o ooouon oz» no on an» Isaac nnz new o>n»=oono nonzo oz» 0» ounoaon oz: .um:0u a»unoz»:o nonnn oz» so oz 0» unannou nu oz: o>nnauoxo nOnzu oz» IOnn slow vonoze:o Ono ouo>ou annnoz»=< nouonnonolaono o .ouoooxo non .uonnooo no coon o.uononnno oz» on vovouuon onn »oo»nnnno no »»oooo oz» 0» »nooon oz: noomouoam .uononnno oz» no uon»ooo no coon oz» non»ouoouou on oo»oznvnozon no .ouooou a»nnoz»ao no nozaao oz» no«»ou:uuou on ouo>ou o»onooon nonounnooo »oc ono n»oo»nnnno no non»:ooo nozh .uouonnno oz» no connnnoo oaoo oz» anoooo o» wonovnnooo ono ouonounno o» onoo»onnno no non»=oov .uonooon ou uooouunoaou onoz» .unnooou cu .uoo oconnoo uonucoounooo moo z»n3 nouu can» onoac on oooon»ooa non a»nuuznooonnon ooon>no o>oz no: ounozn nounnno uono woos—n nonzo oz» 0» oonu a»»noz»oo ozh .unnn.on».~nn.nnn onan nuconunannon on» snoonn on ooonnov non noonnno unnoooonn nonzo oz» o» oonu >»nnoz»:o »oonno o on ooonouoso nnznnouuonnco» xuoo nonunoou .noo»ouoao onzanouuonnooo nnonn>noo soozoo no nozeoo unounou oz» no uooou oz» aonn a: uo>ou onnz» oz» no onon luonooan znna nnoo unnonnn»no oo o»nono .nono: noz»o on .»nooon nnonn>nooao ----------. .auuou»nom oooo dunno: unnoooonnooo oooz» no noooz oz» 0» »nooon »oo ouoozn nonoz nnoz» .non»u>n»oo onzonouuon»oou ononnoz noz» zonz: non »no: uooonnonnonnno no »ooouoo bz oaoo oozonz ono soz» nn non .huunoon» nooon oaoo ooxonz oz uuna onnoo onznnouuon»ooo mongoonn oz» ooza o» nonnonooon oz» non nnznooono ozoz» on nonoa no oonz» nun noomouoeo anonn>noosnooo oz» uuo »no: onznnouuon»oou uo>ouu»no:ou uo noonnnvoo oo oo »o=oo .no»ncnvnoz=n onzo Inouuonnooo n o>oz no: on nnonn>nooan ouzonouuon»oou onooooz auouoo n»noo onzo .onnoo non»onooo o» oozuo»»o ono oooaouolo onzonouuonuooo nn «cu con»noav on »oz» ooz» noaou oz »zn«o uooou nnzh nvo»ooou ono onzn Inouuon»ooo on oonnooon snoonno oooza ooououooo onoza uooou hnnnoz»:o »oo:ou oz» on »oz3 .cuu — . . _ _ . . . . . . _ . . . _ . . . _ . _ _ . . _ . . . . . . no»no: onznnouuonuooo uooouu»oo:ou oooz» on uooooonoo uoooun»oooou oz» ono uopou »»nnoz»:o »oz3 »< .mOu r‘-"'-- unouuon»oou no unnuoso »oo ov neonooonn onzonouuon»ooo uo>oun»no3ou neon nu .»:noo »noon3 o»n o» no nonoouuoo on on:»uon»n oz» .oano» ouofluo cu .uo>ou noz» oo nouoz no oonoon» ono nooaouoao anonn>noaanooo no nozaoo »nownou oz» no uo>ou oz» zouoz nuo>ou oo nooAOunao uuo .onu< .ooxouoao Anonu>nooonooo o no oo»oon» on noohouoao >nonn>non nonooo no nozooo »nonnou oz» no uo>ou oz» o>ozo uo>ou oz» nouoz ononnonooao hon .noOn»oonv onoz» no onoonoo oz» non .oooaouoio unannonooaoooo no nozaao »non Inou oz» no uo>ou oz» o>ozo Auonouooaan uo>ou oz» oo oz »noa nnonn>noooo ozh .uo>ou »oz» souoz no nonooou ono ooo uaouooo zuon no nozaoo »oounou oz» onoza uoonu oz» no oz »n:l oooaouooo anon»? noaonooo ozh .uo ouon uooOn»nonnoo uono nnoooooo oz» on onoz» ono noon» »nos onznnouuonnoo» no «on» uonoooo o»noo onzonouuon»ooo uooouunoooou boo- oou .oou noon»numm o on nuns onzonouuonnooo uooou|»oooou o oouau uooOnummmnoa ouooou h»nnoz»o< connon» noononnno nae: oon»oonnuooouu 2252 nonouoo .zuozo ooo onoun non .on»nuoouoo on nonooz oonn eozoozu Anoooonnoo on! no .owono nnznoz .nooononnno no ”onon un»oh no nonno unnnonooo no onouob onon onzonz »oov nooonnon uunnnno on onzu 8.3. 5.2.8 an ll , — a — _ _ _ nunonon uuou — a n _ a _ nnnoaon connooouu a _ n _ u _ nnnunzo»nnona nuooonn — _ _ a n _ nonnoou nnnnoao a u n _ a _ noooono» concuoau a _ n _ — _ nonnoason nooo no »nnono »ooouon noonunnuax — a n _ n _ Nun no noon: .nuoonunnno a _ n _ — _ onnn no oonnn . a n _ — _ nnuooooo no oaouo> .nnnooz nonsmon oonn noooonno> — _ n a — _ onnmvaz ouzuaoun n a — a — . nnnnonm »oonmon oo»oxooz m _ _ _ _ onooo noonnon nououoon n on mum » : nnomou no omnn non» »nu on: non nn»nooon unnnon on nunonu on»oonvon nz» o» vnon>ono nno nonnonnonon no noon» nonrouuon nz» no zonza .nuu nooonnno nononocn» nunnu on. on nooon nnnnozoaq “on: no .nnn mammmuumm .won»owooz on wonnvouonOn noooo »oo on .uo>ou nnnnoz»:o cocoon oz» oo on An»onov nnz nov n>n»oooxo nonzo oz» o» n»nooon oz) nonunnononoouo o .nuoanxo non .uonou n»nnoz»oo nonun oz» on oz 0» vnonnoo on oz: nonnooooo nonz» oz» nn coon oonnzaoo ono nuoonu nnnnoz»:< .uonnooo no coon n.unn»nnno oz» on vovouoon ono noonnnnno no n»=ooo oz» 0» »nooon oz: noonouoam .uonunnno or» no uon»ooo no anon oz» mounououuo» on nonoononozoo no .ouo>ou n»nnoz»:o no nozeo: oz» won»ououuoo on nno>ou ononooon oonoo unncoo »oc ono nnonunnnno no non»oooo oozk .unnonnno oz» no cannnnoo noon oz» nooooo o» vonoonnooo ono nuonounno o» nnoonnnnno no nonnonoo .uonocum ou .noonnno uouooocnn nonzo oz» onnov unnooo oz tuna: nnuuon»:cu oz» .nouuon» loo» oz» onoa nonoononozon nnz no ooo nu .noonnno unnoooonn nonzo oz» oonoo snoooo oz nuunnooon no: ouooo ooooonn ooo oon»on»nnonooo no nonozo on noon lunonnsoon> o .ouoaoxo non .zon nnz no noon oz» a: ozos mannooouuo can oocoCun no ucou no zon nnz no »noo »cocn20vonn n ononn»nooo nos ooo coununcnnnon nnz noon: nz non nonnnouono nnoo »:z zoa onz no nnoo noooueooona o o»=»n»ncoo »oo ouoozn noono unnuoocnnooo nnz»o ooo oonnonnnnonnon unnoooo .uonnooouoo ooo ooo-onn nnn nonnoo nnonnno nnoz: unnonnno nooonoo noozanz oz» nz on nonnonnooo on nonouooo onznnouuonnooo gonzo»; no noonnno nonugucnn nuns» can onuou unconnnonnnn nonnnnnonnnnnno »noonn nnnonuolou nonnnn i an “IUdNflIICdO 2253 .noooloo on »o«uo »oo oooo zo>o~ ooza <2 »nooou .aoooz n»nnoz»oo oooooo oz» so on Annanoo anz no» o>nnauouo nonzu oz» o» o»nooon oz: »soo»aononou»> o .ozaaoxo non .zo>oz n»»noz»:o »on»n oz» co oz o» ooonuov a» oz: o>»»:ooxo nouzu oz» aonu choc oonozaoo ono ozo>o~ n»nnoz»o< on» nope; nope» gun _ _ no>on a». z _ noose» no>on zoo _ . nu>an can z _ «oooz znn _ _ uo>oa vow _ _ nvo»oooono muzoonnh» o»nooon ooon ooo unuono .« noJmoo oz» - ono oaooou n»«noz»oo nowaouzou oz» no zonz: on .ozz onooo .oozuozo o» Ho>oz oo ooz: .oozoozo Ho>oz nooaou oz» no nozaoo oz» can ooo nun ooouuoooo o» onoaooo oz» onoom .noooooo on »o»no »oo ooov Hoooz ooz: <2 »noocn an» ~.>.a n.,un so. . “ ~.,on so. _ _ noosoa «ooo— zuo — _ zo>oz vnn — _ «oooz z»n — _ zoo-z vow — _ ~oo»oooono ooooooon »oo »oz onooo no o»oolo»o»o on .n...» nunnozuac nanaanno» oz» no gonna an .n» nona no nozaoo oz» 9» novazuon o» .oo» nozoloooo ono ooooo>on ooo ouooo zunza non .oe«» ooo noo »o ooonnono on azofi uo nozsoo zoonon» ozh .uonunooon ocwn »u=vono on mz«»noaon »uon»ooo no noono zoe noouooou .oznz »o:oono no »uooono o no» ooooo>on ooo o»ooo o»oz:a:uoo ono»oou a»«~»z«ooonoon oz» an oo>o .ono»ooo »o:vono oo ono»ooo n»»z«z»ooon :oon uoooo no: on .ono noz» :oz» 53oz cozonz nzoonu onoa oz no noooaoo oz» mo o»no: oz» nozno>o noz»no »ooa ono»ooo »oooono ooza .nonooo h»»znznooom soon o oo zuoo .no»:oo no ooa» Hanan» Iooou noz»o oaoo o» ocoooonnou »oo »o:fl no»coo uoovono < .»o:vona nz o»nooon wanno»oou:coa on oo»onz o»ooo nusvono o»onoooo no onoooono az ooahzooo oozoo on vo»o»z ooooo>on »o:vona o»onoooo noz»«o ono»oou »uovono no »==ou »oo on .ozqz »oovono no »uooono o no» ooaoo> ton ooo o»ooo oaoo noooz »o o»oz=aauoo »o:l no»cou oz» .no»:oo «o oo»» «noon» noo»onaloooo ooooooon ooo o»ooo on: ono»ooo »o:oona noon 3oz non .oau noo»onoloooo ooooooon ooo o»ooo ono ono»ooo »»nono noon ooz non .nnz oonuu umnooo noon» loos» noz»o oaoo o» voooono no .no»:ou »oooono o oo vo»oooo oz o» novno on .vo»o~:aoouo ono o»ooo ooo ooooooon znoz zonza non ononooo zo»»ozoloooo o»oo zono loooun no nozloo oz» »oozo on oonuooac oz» .oono: noz»o on .ono»ooo »uuono ooo onouooo oo»»oznn»oou z»oz ooonoon oozau nooounnonuoa onnooon ooo» ooo »«nono onnooon »ooo onoooooo no nonnoonuonnooooz ononooo noavono nonnnooonm ono»ooo uwuonn noduonoona noz»oonnnooonu 234 .o no »oun:z »o: co ooo .u no ooozonxo noz»onooo oaoo ooo nooco>on new» conozo oaoo .~ oo oomzoo nno noz»onooo ooo ooacu>on noz»onooo .m oo ooozoo woo zzo ooo oosco>on zzo onoum .oounozo »cono«unv oz» no» nonouo oz» Boo .znz oncoaoon oooou non .u .oooz»oa zoun»oaoz»ol ~>I\ no omooouon oz» onooo .voo: ono ovoz»oa moo |«»oaoz»oa ooo «ounzoonm z»oz coz: .o no oo»o~:o azoo »oo oanzooo«»ozon »«u Ionauoeozo>a»ooo ooo .N no ouoz»oa zou«»oaoz»oa .z oo nomooz »oo oo oooooooo uo«»onono oaoo ooooo>on non»unooo oooo oooooono noz»onozo ooooo>on on«»onooo ooozoano Ado oooco>on «no o—a—oo-oI—o—o—o-n .uuo .ononzooo non-nonuon ooszoon oooz»oo 1o»oo»»o«zoom .o»o .voz»oo sonazunz .lonmono no»»ooo o no »zu«o nz oonz o unnaono .oonnomo»oo ozzonno> ooo coon» zoo«»oooz»oo o»o« o»=ooouo oz» «0 oo«»ou«»»ooono zuunznonm z z 1333...... 2.32: 32.»!- basal A.o»onozozuv ~oo»ozouzoo nonzooon» noun .oooz»ol zoonuoooznol oo oooz»ol nozon »«nououolozo»|»ooo ono unooz zunza no .ouz oo»oou»onzooo duo ooo ovozuol zooqzoonu nouooz ooqoxaoooooo o oo ouoa>oo o noon ooo noooa »o sonooooz ono ooo»« zonzz. a huuoono .ozooz noz»o o-oo-no— fiq‘o‘uuo—oo—n—ou .u omwzooonnowon »»uono,olsao>1»ooo o»ozouzoo zoo oz — z _ oooz»oa zounzaonu onouo .aau onconoon ooooo non .o»n:: zocon»on»cowno ooo oocud »unvonn no» zooo z ooo onooz oonalnoooaoo no» u once» .o»noo oonz» oz» no» oonooo oz» l3- .oua ooo can oooooooon ooooo non ooonz »oooono no» — g ouooz oonaahcoooou o co m u "oo»onooaoo o»o«oo oo>ozoonz on< .oun o»noa zooon»onnoouno no» 9 z ooznz »o:v¢no now — _ oqooz ovnalncooioo o oo _ ~ »o»oooomaoo ouzoanob ooo noun» ono» no»onoooo o»ooo on< .auu .n n» ucn>nu ‘- Ian—ad. 8.00 flown—u oz» non oonooo oz» loo .huz ooooooon ooooo non uuqan unnoou nu A.»ooaozo »oou zooo non ouooz ooo nzoo zoozov _ _ z z z z z _ uooo ooozno>o .Auv noz»o — g n _ _ _ — . »ooo nozoz nannooonwzo zoouzzuo» o co ooooz o»ooo hip ~ _ z . Fl. »oco ozonno»o= onovoo»o oo o»on ooozno>o oonzooo oo o N Iml o- novnocou oooo zouza on .hv:»o ucnnooo nnnuoao .noz»o .o unmoo doonozoo» o co ooooz oo»oi»»oo no:»o nounooonwoo doouozuou .« ono o»ooo voozno>o zo»o» moo o»on n» onooo ~oo»no»o»= .u tnooooo Iolnnoo ooooznncnnooonaoo oo o. oooo~o>ov no: onooo onoozonu .o »« ooonos any o»ooo «oonno»onz zo voooz ~o»oolo~o onooo enouoono oo o»on aoozno>o vouzooo nooo nonoouzou oz» no» oooozoooo nzuoonon» onooo oo noonooou .nomvou nonoooo oz» on onoooono ono ooooz nonooazou oz» no zouza no .ndd vouonoonoooa oz ooo oooo oaooo onouoonu mmmmmuumm sung» naaonununnom nIIIIIIIIIIIAfiIIIIHHIHHHMuoo anonoolono ado noonooou. o»«oo Hooo«»on»=omno no» a a ucu»ouooz ' II. nuuono nooozoolnoou o»o»oo oooozoon- o»ooo onzounop ‘flo vouuh oaooo unouoovn ooovaozooh oouaosoooo flowuouuuuoooau 235 ouzoonzzao oo o n.« n N u a . a zouvou. onoum «quz wanes» “coauozo oz» onoom .u ooo z ooo3»oz ozoouo loo oo noz»onnzan»oou no ooouoo o»onvoano»c» onoum .« oo v;»»zon»oooov o» noouaoo oz» monnoo Izod» noonoo om ooo a no qonnzon»ooo o» nnooaou oz» uo«»oo»uo» noaooo on onoum 1 .x on can zo»»oo:c o» oooooooon oovou ooo Son oz» ooo an oo»»ao:v o» noaoso oz» oo n~z oucooaon 3o» oz» onooo .Aozon Hoso»»»onnoo onumooao oouv o»»oo onzonoz~0n»ooo moan» :«zoow on ono onoz» coz3 b---- "'-|-"" ---- nunoozn .noz»o _ _ »oooonoool uo»»onooo ooo zoozoonoo .ozoza o oo nooosoo oz» »oo ouzonozzonnooo uo>ouunozu»z ooo3»oz ooo»>»o . . »az nooaaou oz» uo »ooowoo oaoo «o oo»» »ooooz »oolomocoo noz»onooo — . s»>»»oo ozzonozzon»ooo no» ozznoooooon unoo zooooonoa onzonozzon»ooo zooozunozwnz n _ nunnoszno a» zuzza Au ozon zoooznnozuoo »»nooon »»oo zonooouv »»z= ouzanozqonuooo no on». no» o.nozuon»ooo Honnooo oz» :oHoz ono zouzz o»»oo uooao o ano o»»c: a»zonoz~on»ooo onozh auzonozzonuooo oooz» no ooooz oz» oo ooza oh .onu WIIIIIIIIIIII I nu .u I «momWAIn»»noz»oo oooooo oz» oo o» A>»3aoo onz now o>»»:oouo nowzo oz» 0» o»nooon oz: »zoo»oonosoo»> o .ouosoxo »»»oo o.oo»»on»ooo «on»ooo oz» non .zoao» n»»noz»oo »on»n oz» so oz oozoz onuo: onzonouzonnooo no noo »oozo»z oz» 0» noo»uov o» oz: o>»»:ooxo nonzo oz» no ovooz oz» ono ~o>ou snunoz»:o »oz3 »< .n«~ Ion» osov vonozaoz ono oao>oz n»»noz»o< .zon»ooo no zooo o.uo»ozuuo oz» 6» oooozoo» ono »zo»o»o¢o no nnooov oz» 0» »nooon oz: ooonozoam .Hououuno oz» no Honncou no ooo. oz» mcmuozaonou z» oo»o=»v»oz=o no .o~o>oz ununoz»:o no nozzzz oz» - u:»»ozau~oo on ono>oz o»onooo» monoo .x oo «NH ooo aNH uoo»»noov o» onoaooo oz» onouo .ounooon a»»d anznozoqoun «o>ozuno:oz oo ono onoz» :oza .x no cad onooo .oo o» nNz con» noose o» nonaco oz» ooz: .ooooaoon oonozanz »oozm»z 1# oz» onooo .oozoozo o» oonozo ooo zoz» onoo ooz: .o oooozo>oo oo»»olno»o» oz ooo .« ono»on no o»on zoono»z» ooo oo~o> »oooono »oo .u anonon no o»on oznuoaoooo so» soznoz zu-znan onoo- .uu» common.» «one» non 833 on ozzoounooo »oo — _ unozoo »oo ono o»oo»o»ooo no oo»»=zov on a . oozh .nouo»uuo oz» no cowouoom oaoo ooh a . oz» amaooo o» oonooqocoo ono odououuuo ~o»nooon o» o»oo»o»ooo no oouuaooo .Honooou on n»»»»z»oooooon «ooozunoooq onoz» o» ooooo>on ooo ooooooxo odzonzon»ooo:o: Ion» vonouonwoo .oaoz: o ooaoooon ooo oooooooo ozzozzo»»ooo one .0» an .eua oo hooosoo oz» no» »oo »=z oo»»onooo o.noooloo oz» no nooouoo noo no» o»nooon on a _ nnnzuzuoooooon ovouoo» o» o»nooon con . _ n»»»»z»ooonoon nooouonooou »onono»ou poonooona o» »nooon oz» 5oz: no» noaozoa‘ . oz» hz ouzoazon»ooo no: ooooo>on no oonoooxo ooozozn Addonoaou ounooon huuuuznoooooon zooozanoaoz oz .nua vonoHoooo oo»»onnouo» on a nnzoooo .noz»o — “ ono»on no onon Hoono»ou — oszob »=ooono ooo a ono»on no o»on nonnoooooo — oounoo zoozhoo — .33 a: n: .... no?» 2558 no» 392mm... 3 3.8.. n is .2» Simona 3: :3»... “UAHOhuQOu oo»¢«>»o no n»n»nzn-aonaon no»»noaou acol»nonoo o.no»»on»ooo don»ooo no oooouno- hnwnozua< uo oouuoouuonuoou onzozuon»oooooo ooo.onzo-on»oou 3:83» 3:8 8:83:88 236 1ozoozu nozlsz nouoh .u no had non»oooo onooo .ono»:onoo noo oooz »oo oooo noooooo oz» oozn oozau nonoou nu no »no; »=onn2ozono o o»:»n»oaoo zoo zoo oon»Unoonn=n onz noon: oz nos monooooona o»oo noz zoa onz no unoz ncconsooonn o o»=»n»oooo »o: ozoozo ooono zonccoznnooc noz»o ooo oo»»on»ononaoo zonozou .mcn I»::oooo ooo ouooonn ono connoo Anoonna oooza nononnno nooauoo »oozmnz oz» oz 0» vonoonozoo on oonozaao onzonozzon»ooo »eozmnz no noonnno Honoooonn nonzo ozh .co»»oo:oo zoflnon no onoon cu o>oz o» oonovnozoo on oonmoo ooozoanoco no .<.n o z»»: ooonoo o ooo non»ouovo «oanon na.o»o .o»oo»o:oooo »ooov ooonouolo no onooh an o>oz o» oonoonoooo on o»o:v onzononnon»ooo nopous»oo3o» oz» no wonnovon Manon non» Hoozoo zwnz < .onoaco zoo»nomnoo Inoo ono oon»ooooo «oonon no onoon noon no: .Hnn ncnz monoozonov «oozoo nno»oooo«o oonn vo»¢:oo ono oonnoosvo zoanon no onoou no>oz Anonaooon» .noonnno nonoooonn nonzo .»zo»naoooo nonzo .nonnon»=o0v oononooo nnzonodzon»ooo »oozunz oz» oooo con»oo:oo nolnon no onoon soon now .an o»noo onzononzonuooo ado o» .Aonon Hooon»no»noo onnnoooo oonn voounooo ooonozooo o» onoo oonnozoo ooo oowo: .on» «a oon»ooov o» nooooo oz» on novazoon ooonouolo «no no oonnozoo oz» ooozoon nooooono u»oon»»o»o no zouncozuoo oeoo .oouzloooo no »Hooon oz» oo .nzzoon»oao»:o »:o oo»onnz ono oz «an: o»nooon ooosvoo oz» zonzavoonn ooooonon noono»on non noooaou oz» nz ooozoono o»oo no ooonnonoloooo ononooonouo» ooz» o»nooon ooononoooo»o oz» no connnooono naza .ouz noz»on onnooon ooonooo oz ozsozo ooozu »no:o zoono»on non»o:oooo »owooz nowooz nonooou non»coovuo »ooo nnonzo>cn oouzonoa 16o ouooe»nnzon1 oozzo>nooon ooo mznzznz uzcnzoa nnonooloo o oo .»noo on »ooo» »o .vosnonnoo ono .noo»>noo=o ooooznn no noz»oooooo oo ooo oon»non»oo donooonnn nonoouuon oz» no zonz: .ouz and.co».om ozon zooon»n=»noo onnnoooo ooov ozon ooooonn no uznncooooo z»»a ooonouio unanimous-ooo oonz» »ooou noon»v?n»oo .onnoo »o oo ooonnoo on »no= onzonoznon»ooo < onzononzon»oou o» oo»o>oonnooolou oz» hz.oonnnn»z . oln» no»aoooo «no no oon»nooono oz» on »rz3 .nn» .nooaloo oz» no ooonoono oz» oo oo»ooon ono»ooloo no oon» no»ooooo unoo noonooou ooon»oo oonou zooon»nonnoz U—H-ubuu ~o>on »oozwnz no»: »ooao~ non»ooooo noanon zooooanon no oooonnoo sou ooo non»oooon on. nozoznno»noou oo»»ooo»oo unooo- nonnoonnouooloo oon»oonoozoo: connoonnnooouu 237 .ooznoa onooz oan»»o>o no zon»nooona oz» »uoznon o» no»o:aoo oz onoozo oo»on nanooz .oan»no>o no» roanocn .oozno: onooz no ooxmwmou nm~.o»nnozoz o»»nocoz mummmm ouonnh »oou no::o< ouono>< nooz»n3 na.o»o .o»co»o:oooo »ooov ooononolo onzonouzon»ooo ~o>ozu»oooo~ oz» no nnonoo Hoocoo ouonovo oz» on »oz3 .snn »¢=oao noonoz oz» non oo»o»o oz onoozo zooon» Ioooo oooz» on oonnouoo condo». oomoa nooooo gonzo umnoom on uncooloo oz» nz vononolo oooz Anonoooon» .noonnno zonoooonn nonzo .»oo»oooooo nonzo .noznon»coon oonozoso onzo -noanon»zoo »oozmnz oz» ooz onoo» noon so: .on» nnzooaoo oz» nz oohoznlo oooz o>oz A.o»o .o»co»z:oooo »ooon ooonoaolo onzononzon»oou zo>odl»co3on oz» »oz» onooa no nozooo ozono>o oz» on »oz3 .nnd n»=oaaozoso no noon »onnn nnoz» nonnao o>nooon AnzoEnoo n.o»o .o»:o»o=o»oo »ooov ooonozoao onzononzon»:oo uo>onu»oo3oz oz» ow nooonoo oz» onz»na Amononon» zon oz» oo sonoonozn »oon nononon» zoonon no a»:o»o>noco oln» anon» ozoo) noon 3o: .cnu nooonuov noozo>naro no .4.» oooz o»»oa onzononnon»ooo o» coconooo ono oz: ooonozolo oz» no moo. to: .nnn mmmwmuumm nooozn»ooov Iouoo ovozuon .hoooaou oz» nz anon and noonoo »:z annno»zozo> onoz» ooo .»uon» noon Azonoo» nz connouon oooz» .ooxo» uzonnoo oo zuoo .uon nz connouon oooz» o»nnozoz ouonnn no »ooo oz» 3» ovanozn .noonnno donoooonn nonzo oz» oonoono ucoo oz ozooa nomzonuooo oz» .nonnon» nooo oz» ono: oo»oonvnoz=o onz no ooo nn .nooznno zonocoznn nonzu oz» oonou Inooou oz nzzonocom »cc ozao: oucoonn ooo oon»on»ononooo no onnozu on »ooo snoonoxoun> o .ozaaoxo non .zOH onz no »oon oz» 3: ozoo u:»»ooouuo ooo oucocnn oo wcoz no zoa onz no »noa »coznsovono o o»=»n»ocou non ooo zcn»onvonn=n onz noon: oz non wznaoouonn o»oo »:z zon onz no »noz »ooonoooono o o»ozn»oooo »oc snooze ooono «onocoonnooc noz»o ooo oon»on»ononaoo Honozoc .m:»»==ouoo ooo oucoznn ono oon»:o >noazno oooz) nononnno hcoQEOo »oozmnz oz» oz o» vonoonoooo on ooaonooo onzunoazon»ooo »oozm»z no noonnno nonozoonn nonzo ozh .zonan nacononunnoo onnnoooo ooov cm noz»ooav o» noanoo oz» on coconuon ooonozoso oz» «no nan ooonuov »oozo>»:oo ooo .<.n oz» »=ooo .noonnno nonuooonn nonzo oz» oonooncooo oz onooa noz»on» loo» oz» .nonzon»:oo oz» ono) oo»oono snozoo onz no ooo nn .noonnno zonoooo Inn nonzu oz» oonoonoooo oz nznonoooo »oo ozooa ooo-on» ooo oon»on»ononooo no ounozo on »oocnoono oono o .ouolouo non .zcn onz no »oon oz» no ozol no» a»oooooo ooo ooooonn oo noon oo zon onz cons: nooon»nonnoa uooon »oooon nnouou n...» ...zonz nosoz »oooo» oooonnoouu nonononk dfldug coda-ounnooodu 238 ." oaooz ooMOszu now, ..oazoz noonnAH ounnonon omonnm no o»nnocou a nooo zoooo< owono>< ooonnn »:oz»nz. nanonoooon» .noonnno zonoooann nonzo .»co»o=oooo. nonzo .nonnon»ooon oononolo nnzononuon»ooo. ».oz»»» can nounnnna. an. oz» .» no»: .nn». 3»... 2.38 2.3». I- I I I .. .. nu .noonnno nonooocnn nonzo oz» vonoono uzco oz unaoa nozzonzooo oz» .noznon» Icon oz» ono: oo»oznonozoo onz no ooo n» .noonnno zonuooznn nonzo oz» oonoo IzocOo oz mnzonoaou »oz ozooa ouaoonn ooo con»on»ononaoo no onnozo on »oov Inoono.oo»> o .ozoaoxo non .an onz no »oon oz» no ozoa «on»oooouo ooo oozoznn oo ween so zofi onz no »noo »coznaooonz o o»o»n»o:ou nos ooo oon»unoonn:q onz nova: oz no! wcnnoouono o»ov »oz zon onz no »noo »o=:»nooono o o»o»n»azoo »oo ozoozo ooono nonoooonnzoo noz»o ooo zon»on»ononooo Honoaoo .ncn»o:oooo ooo ooooonn ono oon»on anoanno oooz: nonunnno noooaou »oozmnz oz» oz o» oonovnoooo o» oonozooo onzononaon»ooo ooozmnz no noonnno Honoooonn nonzo ozh :.oonozouo no» o»nnoooz ooonnn »ooo zooooo ouonobo: »oo .ooooz no o»nnoooz ononnn ooozuna . .nnouoo nooooo ooonooo on non non» I oouau dozenunonnon n...» o.oznnz «ooo» »oo:on ounnoooz nno»n ouonooo «28» 2...»: 80¢ aflou «din-d0 APPENDIX B INTERPRETATION OF COMPONENTS .APPENDIX B INTERPRETATION OF COMPONENTS The four sections of Appendix B analyze the meaning and other characteristics of each variable in the dissertation. The variables are measured by component scores and henceforth are referred to as components.1 Each component's properties are first summarized in the following fbrmat: Component name: .......... Eigenvalue: ... Percent variance explained:2 ... Direction:3 positive or negative Consistency:“ ... Number of measurements: ... 1See pages 72-89 for a step-by—step description of the way com- ponents are derived. 2Eigenvalue and percent variance explained are discussed on pages 77-80. Their values for the components in the dissertation are listed in Table 4, page 74. 3The direction of a component may be either positive or negative and is the way the component varies with the variable it measures. If high component scores indicate large amounts of the variable, the com— ponent was defined to be positive. If low component scores indicate large amounts of the variable, the component was defined to be negative. The direction of the components was determined on the basis of the signs of the component loadings of the high-loading measurements. If most of such loadings were negative the component was determined to be negative, but if most of such loadings were positive the component was determined to be positive. In other words, if'mest of the more significant measure- ments increase as the component increases, the component is positive. If most of the more significant measurements decrease as the component in- creases, the component is negative. Negative components are not defec- tive in any way, but their associations with other components must be reversed in the interpetatian phase of the study. “Consistency is the proportion of the component loadings on a component which have the same sign as the majority of such loadings. If 239 240 Next, information on the component loadings1 of the measurements incor- porated in the respective components is presented in this fbrmat: NEASURE— LOADING .MENT NAME DESCRIPTION AND CALCULATION Large loadings which will be used to refine the meaning of components are italicized. A dagger (t) beside a measurement name indicates a measurement which varies in the direction opposite to that of the con- cept of interest. The loadings for such measurements must be interpreted as having the sign opposite to that listed. An asterisk (*) next to a loading indicates a loading with a sign that is inconsistent (taking into account daggers) with the direction of its component. Following the tabular infOrmation just described, the components are analyzed in narrative fOrm. Process Cbmponents Component name: Process sophistication Eigenvalue: 3.5 Percent variance explained: 50.5 Direction: positive Consistency: 85 NUmber of measurements: 7 all of the measurement loadings have the same sign, the component is 100 percent consistent. Even if the loadings are negative, the component is still 100 percent consistent since the component has negative direction. Consistency can therefbre never be less than 50 percent. The average consistency of the components isolated in this study is 83 percent. 1The nature of component loadings is discussed on pages 79-81. All the component loadings used in the dissertation are assembled in Table 20, pages 293—99. 241 MEASURE- LOADINC MEN T NAME DESCRIPTION AND CALCULATION 0.97 R4 SELLING PRICE 0.90 R5 MATERIALS COST 0.97 CSA R4-R5 (value added) 0.40 CSB (R4-RS)/R4 (proportion value added) -0.29* C6 R6/R4 (research emphasis) 0.01 C7 R6/R7 (research emphasis) 0.78 R8 PRODUCTION CYCLE-length With an eigenvalue of 3.5, this component is a fairly good ex- plainer of the original measurements. It has positive direction and is fairly consistent. Only one of the seven original measurements, researCh expenses divided by selling price, varies in the wrong direction, and that is explainable by the fact that selling price, a high positive- loading measurement, is in the denominator of the inconsistent measure- ment . The highest loading measurements are selling price and value added. The length of the production cycle is also fairly highly loaded. The lack of any strong association with current research expenditures is surprising and indicates that the component is very production-oriented. Component name: Process—output diversity Eigenvalue: 1.3 Percent variance explained: 65.2 Direction negative Consistency: 100 Number of measurements: 2 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION -O.81 R9 NUMBER PRODUCTS -O.8l R10 NUMBER CUSTOMERS "Process-output diversity" is a poor explainer of the two meas- urements, number of products and number of customers, with an eigenvalue of only 1.3. It is a 100 percent consistent negative component and is 242 equally loaded on the two measurements, number of products and number of customers . Overa Z Z S tmctura Z Com- p Zexi ty Components Component name: Company size Eigenvalue: 5.6 Percent variance explained: 93.9 Direction: negative Consistency: 100 Number of measurements: 6 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION -0.95 R12 NUMBER EMPLOYEES NOW -0.96 R14 NUMBER EMPLOYEES AVERAGE -0.99 R15 REVENUES -0.98 R16 EXPENSES -0.98 R17 .ASSETS -0.95 R18 EQUITY "Company size" is one of the best explainers of the original measurements in the study, having an eigenvalue of 5.6. It is a 100 percent consistent negative component. .All of the loadings on the original measurements are above 0.95. The financial measures of size are highly consistent with the employee measures of size for this sam: ple of companies. Component name: Company job structure complexity Eigenvalue: . Percent variance explained: 71.4 Direction negative Consistency: 33-100 Number of measurements: 3 243 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION -0.80 C19 R19/R14 (elaboration of jOb titles) -0.82 C20 (R20/R14)/(1/R19) (employee job concentration) -0.92 C21 (R21/[Rl4-R20])/(1/[R19-l]) (em- ployee jOb concentration) "Company job structure complexity" is a below-average explainer of its original measurements, with an eigenvalue of 2.1. It is most heavily loaded on the measures of employee job concentration (inequal- ity of numbers of employees in different jObs) but has a significant loading on elaboration of jOb titles. Conceptually, division of labor is most closely related to elaboration of jobs rather than concentratiOn of employees in jObs. Consequently, the definition of the component should be modified to "complexity of the job structure." In this sam: ple of companies, the concentration of employees in a few jObs increases as the number of jobs increases. "Company job structure complexity" was determined to be a nega- tive component solely due to its negative loading on elaboration of jdb titles. The consistency of the component is not a meaningful measure since there was no a priori notion of how employee jOb concentration should have varied with elaboration of job titles. Component name: Company geographical dispersion Eigenvalue: 2.8 Percent variance explained: 69.4 Direction negative Consistency 50-100 Number of measurements: 4 244 W AS URE- LOADING MEN T NAME DESCRIPTION AND CALCULATION -0.95 R22 LOCATIONS-number -0.58 C22 R22/Rl4 (differentiation of loca- tions) -0.90 C23 (R23/Rl4)/(l/R22) (geographical employee concentration) -0.85 R25 DISTANCE BETWEEN LOCATIONS "Company geographical dispersion" is an average explainer of its original measurements, having an eigenvalue of 2.8. It is most heavily loaded on number of locations but has significant loadings on distance between locations and concentration of employees at the largest locations. The component was determined to be negative due to its negative loadings on number of locations and distance between locations. Its consistency is questionable for the same reasons as "company job structure complex- ity": there was no a priori notion of how geographical employee concen- tration should have varied with number of locations. It is interesting that the concentration (inequality) measures were positively related to the absolute number measures (elaboration and differentiation) for both "company job structure complexity" and "company geographical dispersion." Component name: Company divisional differentiation Eigenvalue: 3.8 Percent variance explained: 63.6 Direction: negative Consistency: 83 Number of measurements: 6 245 MEASURE- LOADING MEN T NAME DESCRIPTION AND CALCULATION 0.15* C26 R26/Rl4 (lowest level unit dif- ferentiation) -0.83 R27 CHIEF SPAN OF CONTROL -0.97 R28 DIVISION HEADS-number -0.58 C28 R28/R27 (highest level unit dif- ferentiation) -0.97 R31 OPERATING DIVISIONS-number -0.94 R32 PRODUCTION DIVISIONS-number "Company divisional differentiation" is a relatively good ex- plainer of the original measurements, with an eigenvalue of 3.8. It is a consistent negative component since all of the significant loadings are negative. Only one of the original measurements, lowest level unit differentiation (number of first-line supervisors divided.by number of employees), was not a measure of divisional differentiation and its loading, though inconsistent, was nonsignificant. .All the measures of divisional differentiation had high loadings and consequently the definition of "company divisional differentiation" should be narrowed to ”divisional differentiation." Component name: Company divisional specialization Eigenvalue: 2.4 Percent variance explained: 60.3 Direction: negative Consistency: 75 Number of measurements: 4 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION 0.61* C29 R29/R26 (lowest level unit spe- cialization) 0.85 R30+ DIVISION RESPONSIBILITIES-prod- ucts, functions, geography -0.73 C31 (R28-R3l)/R28 (proportion non- operating divisions) -0.89 C32 (R28-R32)/R28 (proportion non- production divisions) 246 "Company divisional specialization” is a somewhat poor explainer of the original measurements, having an eigenvalue of 2.4. It has neg- ative direction since the highest loading three of the four measurements are negatively associated with the component. The only inconsistency is the fourth measurement, lowest-level unit specialization, which varies positively with "company divisional specialization." "Company divisional specialization" primarily explains specialization on the divisional level. It is apparent from "company divisional differentiation" and "company divisional specialization" that there is no strong positive relationship (and perhaps no relationship at all) of differentiation and specializa- tion between the divisional and lowest levels of the sample companies. Company mechanization-general; Company mechanization-computers Component names: Eigenvalues: 4.0; 3.9 Percent variances explained: 28.8; 28.0 Directions: positive; positive Consistencies: 71; 79 Numbers of measurements: l4; l4 MEASURE- LOADINGS MENT NAME DESCRIPTION AND CALCULATION 0.36 ; 0.01 RSSA MECHANIZATION BULK 0.21 ; 0.15 R35B MECHANIZATION MOST 0.80 ; -0.34* C36A R36/R16 (energy intensiveness) 0.45 ; 0.70 C36B R36/R4 (energy intensiveness) -0.04*; 0.80 R37 NUMBER COMPUTERS -0.22*; 0.79 C38A R38/R16 (computerization) 0.37 ; 0.79 C38B R38/R4 (computerization) -0.30*; 0.77 C39A R39/R16 (computerization) 0.37 ; 0.79 C39B R39/R4 (computerization) -0.43*; -0.11* C40 R40/Rl4 (electric typewriters per employee) 0.92 ; 0.12 C41A R41/R14 (fixed capital per em- ployee) 0.87 ; 0.08 C4lB R41/Rl7 (proportion assets fixed) 0.79 ; -0.25* C43A R43/R42 (depreciation/wage and salary expense) -0.44 ; -O.28 C43B+ R15/R41 (capital asset turnover) 247 The first component extracted for the fourteen mechanization measurements explained only 31 percent of the total variance of those measurements. Consequently, a second component was added.which in- creased the total explanation (by the two components together) to 57 percent. The two components were then rotated using varimax rotation to produce the two components, "company mechanization-general" and "company mechanization-computers," used in this study. "Company mechanization-general" and "company mechanization—computers" are both good explainers, having eigenvalues of 4.0 and 3.9, respectively. "Company mechanization-general" is loaded heavily on four meas- ures of general (perhaps production-oriented) mechanization: energy intensiveness (proportion energy expenditures to total expenditures), fixed capital per employee, proportion assets fixed, and depreciation divided by wage and salary expense. The component was determined to be positive on the basis of the positive loadings on these fOur meas- urements. "Company mechanization-general" seems fairly consistent even though four of the measurements have negative loadings on the component. These loadings are relatively small, three of the inconsistent measure- ments being heavily associated with the second component, "company mechanization-computers." "Company mechanization-computers" is most heavily loaded on three measures involving use of computers: number of computers, propor- tion computer hardware expenses, and proportion computer total expenses. "Company mechanization-computers" was determined to be positive based on its positive loadings on these three measurements. "Company mechaniza- tion-computers" is fairly consistent, having negative loadings on only 248 three of the measurements. These loadings are small, and two of the three are heavily associated with the first component, "company mech- anization-general." control system components Component name: Company direct supervision Eigenvalue: 2.0 Percent variance explained: 67.2 Direction: positive Consistency: 100 Number of measurements: 3 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION 0.63 C44 R44/Rl4 (supervisory ratio) -0.89 R45+ FIRST LINE SPAN —0.91 C47+ R46/R47 (ratio employees first two levels) "Company direct supervision" is a reasonably good explainer of the three measurements, with an eigenvalue of 2.0. The maximum the eigenvalue could be is 3.0, and 67 percent of the variance is explained. It is a perfectly consistent positive component since all three loadings are positive. The strongest loadings are on two measurements of span of control on the lowest company levels, but overall supervisory ratio, the third measurement, is also significantly loaded. Component name: Company staff support Eigenvalue: 1.5 Percent variance explained: 48.6 Direction positive Consistency: 67 Number of measurements: 3 MEASURE- LOADING MENT NAME -0.27* C48 0.86 R49 0.80 C50 249 DESCRIPTION.AND CALCULATION (R14-R44-R48)/Rl4 (staff ratio) STAFF FUNCTIONS (differentiation of) R50/R14 (clerical ratio) "Company staff support" is a poor explainer of the three origi- nal measurements, having an eigenvalue of only 1.5 and explaining only 49 percent of the variance. urements, differentiation of staff functions and clerical ratio. It is heavily loaded on only the two meas- The direction of "company staff support" is determined to be positive based on its positive loadings on these two measurements. third measurement, staff ratio, though inconsistent, is small. The loading on the Staff ratio was estimated by a surrogate measure which may have been defective. This would account for the low explaining power and the inconsistency of the component. Component name: Eigenvalue: Percent variance explained: Direction: Consistency: Number of measurements: MEASURE- LOADING MENT NAME -0.93 RSlA -0.93 RSIB Company authority levels 1.7 86.2 negative 100 2 DESCRIPTION AND CALCULATION LEVELS MOST LEVELS AVERAGE Even though the eigenvalue of "company authority levels" is only 1.7, it is a good explainer of the two measurements of number of author- ity levels. measurements). The maximum the eigenvalue could be is 2.0 (the number of Eighty-six percent of the variance is explained. 250 "Company authority levels" is consistently negative. Both loadings are large and equal to one another. Component names: Company personnel quality-high level; Company personnel quality- low level Eigenvalues: 3.7; 3.3 Percent variances explained: 24.7; 22.3 Directions: positive; positive Consistencies: 73; 80 Numbers of measurements: 15; 15 MEASURE- LOADINGS MENT NAME DESCRIPTION AND CALCULATION 0.06 ; 0.64 R52 DIRECT LABOR EDUCATION 0.52 ; -0.03* R53 DIVISION HEAD EDUCATION 0.05 ; -0.65* C53 R53-R52 (difference high-low education) 0.40 ; -0.10* C54 R54/Rl4 (proportion BA degree employees) -0.02*; 0.48 R55 WEEKS TRAINING 0.12 ; 0.13 R56 DIRECT LABOR SENIORITY -0.49*; 0.49 R57 DIVISION HEAD SENIORITY -0.56*; 0.38 C57 R57-R56 (difference div. head and dir. lab. seniority) 0.00 ; 0.79 C58 R58A+R58B (direct labor compen- sation) 0.87 ; 0.35 C59DA R59A+R59B+R59C (division head compensation) 0.91 ; 0.18 C59DB C59DA-C58 (difference high-low compensation) 0.14 ; 0.78 C591A R42/Rl4 (average compensation) 0.42 ; 0.48 C591B R58B/C58 (proportion direct labor benefits) -0.35*; 0.48 C59IC R59B/C59DA (proportion division head benefits) 0.90 ; 0.11 C59ID R59C/C59DA (proportion division head‘bonus) The first component extracted for this group of fifteen measure— ments accounted for only 25.3 percent of the variance. Consequently, a second component was added which increased the total explanation (by the two components together) to 47 percent. Such a level of explanation by two components is inadequate but it was necessary to discard other 251 possible components in order to keep the number of variables down and the analysis simple. The two components were rotated using varimax rotation to pro- duce the two components, "company personnel quality-high level" and "company personnel quality-low level." "Company personnel quality-high level" and "company personnel quality-low level" are reasonably good explainers, having eigenvalues of 3.7 and 3.3, respectively. "Company personnel quality-high level" is loaded heavily on three measures of division head compensation: total division head com- pensation, difference division head and direct labor compensation, and proportion division head bonus. Thus its meaning should be narrowed to "quality of high-level personnel." "Company personnel quality-high level" was determined to be positive due to its positive loadings on these three measures. Four of the fifteen measurements have inconsistent loadings (negative) on "company personnel quality—high level." Two of these are nonsignificant, but two measures of division head seniority are significantly negative and thus "company personnel quality-high level" must be considered someWhat inconsistent. This inconsistency suggests that seniority is inversely related to salary Characteristics of division heads fer this sample of companies. The researcher is tempted to speculate that younger, more qualified division heads are hired at higher salaries but that more senior but less competent divi- sion heads are retained at lower salaries. Thus seniority is not a measure of personnel quality on the division level. The research re- sults provide no backing for this, however. 252 "Company personnel quality-low level" is loaded heavily on two measures of direct labor qualitye—education and compensation——and one measure of average compensation for the company as a.whole. Thus the meaning of "company personnel quality-low level" is narrowed to "quality of low-level personnel." "Company personnel quality-low level" was de- termined to be positive based on its positive loadings on these three measures. Only three measurements are negatively (inconsistently) loaded on the component, and two of the loadings are nonsignificant. The third negative loading, difference high-low education on "company personnel quality-low level," is explainable by the fact that direct labor education is subtracted from division head education. Thus "company personnel quality-low level" is fairly consistent. Component names: Company centralization-investment; Company centralization- purchasing Eigenvalues: 5.3; 3.6 Percent variances explained: 29.3; 20.2 Directions: negative; negative Consistencies: 100; 67 Numbers of measurements: 18; 18 253 MEASURE- LOADINGS MENT NAME DESCRIPTION AND CALCULATION -0.29 ; 0.54* C60A R60A/R42 (compensation concentra- tion-top 1%) -0.24 ; 0.64* C60B R60B/R42 (compensation concentra— tion-tOp 2%) -0.11 ; 0.15* C6OC R60C/R42 (compensation concentra— top 25%) 0.73 ; 0.22 C6lA+ R61A/R51A (decentralization- investment authority-$100) 0.87 ; 0.08 C61B+ R6lB/R51A (decentralization- investment authority-$1,000) 0.82 ; -0.07* C6lC+ R6lC/R51A (decentralization- investment authority-$10,000) 0.88 ; 0.09 C63A+ R63A/R51A (decentralization- investment authority-1%) 0.70 ; 0.10 C63B+ R63B/R51A (decentralization- investment authority-5%) 0.64 ; 0.08 C63C+ R63C/R51A (decentralization- investment authority-25%) 0.42 ; 0.17 C64+ R64/R51A (decentralization-pricing authority) 0.47 ; 0.51 C65At R65A/R51A (decentralization- purchase authority-$100) 0.56 ; 0.46 C65B+ R65B/R51A (decentralization- purchase authority-$1,000) 0.47 ; 0.74 C65C+ R65C/R51A (decentralization- purChase authority-$10,000) 0.35 ; 0.82 C67A+ R67A/R51A (decentralization- purChase authority-1%) 0.11 ; 0.82 C67B+ R67B/R51A (decentralization- purChase authority-5%) 0.24 ; 0.61 C67C+ R67C/R51A (decentralization- purchase authority-25%) 0.21 ; -0.02* C68+ R68/R28 (divisional budget participation) -0.57 ; 0.22* R69 BUDGET PROPOSALS BELOW DIVISION The first component extracted for this group of eighteen meas- urements accounted for only 34 percent of the variance. Consequently, a second component was added which increased the total explained vari- ance (by the two components together) to 50 percent. The two components were rotated using varimax rotation to produce the two components, "com- pany centralization-investment" and "company centralization-purchasing." 254 "Company centralization-investment" is an excellent explainer, having an eigenvalue of 5.3. "Company centralization-purchasing" is a good ex- plainer, having an eigenvalue of 3.6. "Company centralization-investment" is loaded heavily on three measures of decentralization of investment authority. Consequently the definition of "company centralization-investment" is narrowed to "cen- tralization of investment auflhorization." "Company centralization- investment" was determined to be negative due to its negative loadings on these three measurements. ”Company centralization-investment" is a totally consistent negative component in that all loadings are in the opposite direction to that which would be expected. "Company centralization-purchasing" is loaded heavily on three measures of decentralization of materials purchase authority. Conse- quently, the definition of "company centralization-purchasing" is narrowed to "centralization of purchase authorization" or perhaps, more broadly, "centralization of operating decision-making." "Company cen- tralization-purChasing" was determined to be negative based on its neg- ative loadings on these three measurements. Six of the fifteen measure- ments are inconsistently related to "company centralization-purchasing," but only two have significant loadings in the wrong direction. These are two of the measures of compensation concentration, top 1 percent and top 2 percent of employees. This is quite surprising in that, for this sample of companies, the higher are the top personnel paid, the more decentralized are purchasing (and.perhaps all operating) decisions. It is possible that other variables influence both compensation concen— tration and decentralization of purchasing decisions. High salaries are 255 paid not only to employees with administrative skills but also to em- ployees with special expertise. Companies which employ a high proportion of staff experts will tend to have more compensation concentration. Such companies might also tend to be more progressive in that they allow de- cisions to be made by lower-level personnel. Another possibility is that high-quality and highly paid top-level administrators perceive advantages to decentralization of operating authority. This might suggest that decentralization rather than centralization contributes to the control and coordination process. It also might suggest that decentralization and quality of high-level administrative personnel are complementary rather than substitutable control subsystems. Also, this analysis would suggest that compensation concentration is not a measure of centralization of either investment or purchasing decisions. Instead, it may be a measure of the quality of high-level personnel. Component names: Company standardization-jObs; Company standardization-general Eigenvalues: 6.3; 4.2 Percent variances explained: 31.4; 21.1 Directions: positive; negative Consistencies: 60; 70 Numbers of measurements: 20; 20 256 MEASURE- LOADINGS MENT NAME DESCRIPTION AND CALCULATION 0.01*; 0.84 R70+ PROCEDURES MANUAL -0.22*; -0.85 C72 R71XR72 (total procedures manual words) O.33*; 0.75 R73+ PERSONNEL PROCEDURES MANUAL -0.04*; -0.85 C75 R74XR75 (total personnel proce- dures manual words) O.73*; 0.04 R76+ PERSONNEL RATING FORM 0.72*; -0.14 R77+ DIRECT LABOR RATING 0.93 ; 0.19* C79 R79/Rl4 (proportion written con- tract employees) 0.89 ; 0.23* C80 R80/R48 (proportion written con— tract direct employees) 0.26 ; -0.13 R81 NUMBER UNIONS 0.95 ; 0.21* C82 R82/Rl4 (proportion union members) 0.90 ; 0.21* C83 R83/R48 (proportion direct labor union members) -0.14*; -0.24 C84 R84/Rl4 (proportion information booklet employees) —0.33*; -0.24 C85 R85/Rl4 (proportion organization chart employees) -0.78*; -0.18 C86 R86/R14 (proportion jOb descrip- tion employees) -0.69*; -0.12 C87 R87/R48 (proportion direct jdb description employees) -0.07 ; 0.50 R88+ WRITTEN POLICIES 0.19*; 0.26 R89+ PRODUCTION SCHEDULE -0.53 ; -0.49* R91 WAGE INCREASE BASIS-seniority, merit, negotiation 0.26* 0.73 R92+ FORMS CONTROL -0.09 ; 0.20 R93+ TIME AND NDTION STUDIES The first component extracted for this group of twenty measure- ments accounted for only 37 percent of the variance. Therefore, a sec— ond component was added which increased the total explained variance to 52 percent. The two components were rotated using varimax rotation to produce the two components "company standardization-jObs" and "company standardization-general." "Company standardization-jObs" is an excel- lent explainer (eigenvalue 6.3), while "company standardization-general" is a very good explainer (eigenvalue 4.2). 257 "Company standardization-jobs" is heavily loaded on four meas- ures involving the proportion of employees covered.by written contract and the proportion of union members. It appears that "company standard— ization-jobs" is a measure of unionization rather than standardization. "Company standardization-jobs" is determined to be positive based on the directions of loadings on these four measurements. An astonishing twelve of the remaining sixteen measurements are inconsistently loaded on the component. Three of these are significant: personnel rating form, di- rect labor rating, and proportion of jOb description employees. These inconsistencies can be explained by the possibility that unions take over many of the personnel control functions formerly performed by the company. Thus personnel rating is not necessary (nor tolerated by the union). Also, the proportion of supervisory employees necessary to con- trol workers might be less. Such supervisors in nonunionized companies might be in the category of employees with jOb descriptions (salaried employees). Thus unionized companies would tend to have a lower propor- tion of job description employees. The inconsistencies between the unionization and the explicit standardization measures suggest that unionization and standardization are substitutes for one another. "Company standardization—general" is heavily loaded on four measures of the existence and extensiveness of the procedures manuals and one measure of the existence of forms control. Thus "company standardization-general" is confirmed as a.measure of traditional standardization of procedures. It is determined to be negative based on its negative loadings on the above five measures. Five measurements are inconsistent with the direction of the component. Four of these are 258 nonsignificant. It is interesting that these are the four high-loading measurements on "company standardization-jobs," measures of unionization and proportions of employees covered by written contract. This tends to confirm the substitutability of unionization and standardization. The fifth inconsistent measurement is wage increase basis. Mere standard- ized companies (based on "company standardization-general" scores) tend to give wage increases more on merit than standardized criteria such as seniority or negotiation. This is explainable by the fact that more standardized companies tend to be less unionized. Unions would prOhibit management discretion over wage increases. Thus wage increases based on merit can occur only in standardized, nonunionized companies. Accounting System components Component names: Accounting size—infOrmation output; .Accounting size-resource input Eigenvalues: 1.9; 1.8 Percent variances explained: 27.4; 26.0 Directions: positive; positive Consistencies: 86; 57 Numbers of measurements: 7; 7 MEASURE- LOADINGS MENT NAME DESCRIPTION AND CALCULATION 0.26 ; 0.84 C94 R94/R14 (proportion controllership employees) 0.72 ; 0.09 C95 R95/Rl4 (proportion employees receiving reports) 0.75 ; -0.l3* R96 DATA CENTERS 0.58 ; 0.01 C96 R96/(R9+R44) (data center elabor- ation) -0.10 ; 0.42* C97Ar (R97A+R97B+R97C)/3 (average report frequency) 0.08 ; 0.88 C98A R98/Rl6 (proportion controllership expenses) -0.64*; -0.38* C98B R98/R4 (controllership expense emphasis) 259 The first component extracted for this group of seven measure- ments accounted for only 33 percent of the variance. Therefore, a second component was added which increased the total explained variance to 53 percent. The two components were rotated using varimax rotation to produce the two components "accounting size-information output" and "accounting size-resource input." Both "accounting size-information output" and "accounting size-resource input" are poor explainers, having eigenvalues of only 1.9 and 1.8, respectively. The poor explanatory powers of these two components are disappointing. Size of the account- ing system is an extremely important concept in this study. The fact that it took two components to explain a reasonable proportion of the variance of only seven measurements is very poor. "Accounting size-information output" is heavily loaded on three measurements: proportion of employees who receive reports, number of data centers, and controllership expense emphasis (controllership ex— penses divided by selling price of product). Unfortunately, the direc- tion of the controllership expense emphasis loading is not consistent ‘with the loadings on the other two measurements. Apparently selling price (the denominator) varies more than controllership expenses (the numerator). Selling price has a high positive loading on "process sophistication," and "process sophistication” (as is shown in the sub- sequent analysis) is heavily and positively associated with "accounting size-information output." High selling price causes controllership ex- pense emphasis to be low and thus its loading to be inconsistent. On the basis of this reasoning, the inconsistent measurement is disre- garded and the direction of the component is determined to be positive, 260 based on the positive loadings of the other two high-loading measure- ments. Since all other measurements are consistent with the direction of the component, it may be considered fairly consistent. The high- loading measurements indicate the meaning of the component has mostly to do with the output of the accounting system, the extensiveness of report dissemination, and the number of data centers. "Accounting size-resource input" is heavily loaded on proportion of controllership employees and.proportion of controllership expenses. Thus its meaning has mostly to do with the input to the accounting system, human and financial resources. The positive loadings on these two measurements indicate the component is positive. Three of the measurements have inconsistent loadings, none of which is very sig- nificant. .Average report frequency, the highest loading of the incon- sistent measurements, is conceptually more of an aspect of output than of input. Component name: Accounting job structure complexity Eigenvalue: 2.0 Percent variance explained: 50.8 Direction: negative Consistency: 75 Number of measurements: 4 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION 0.57 C99A R99/R94 (elaboration of jOb titles) -0.47 C99B R99/Rl9 (proportion controllership jOb titles) -0.89 C100 (R100/R94)/(1/R99) (employee jOb concentration) -0.83 C101 (R101/[R94-R100])/(l/[R99-l]) (employee jOb concentration) ”Accounting jOb structure complexity" is a below-average ex- plainer of its original measurements, having an eigenvalue of 2.0 and 261 and explaining only 51 percent of the variance. It is most heavily loaded on the measures of employee job concentration. The loadings of the two measures of relative number of job titles, elaboration of job titles and proportion controllership job titles, are not very signifi- cant and are inconsistent in direction with one another. As was men- tioned for "company job structure complexity," division of labor is conceptually more closely related to number of job titles than to em- ployee job concentration. However, the inconsistency prevented the use of the measures of relative number of job titles to determine the direc- tion of "accounting job structure complexity." Consequently, "account- ing jOb structure complexity" was determined to be negative on the basis of its negative loadings on the employee job concentration measures, which are in the same direction as the employee jOb concentration load- ings on "company job structure complexity." As with "company jOb struc- ture complexity," the heavy loadings on employee jOb concentration suggest that the meaning of ”accounting jOb structure complexity” should be expanded to "complexity of the accounting job structure." Since one of the two measures of relative number of job titles must be conceptually in the correct direction, the consistency of "accounting job structure complexity" is 75 percent. Component name: .Accounting geographical dispersion Eigenvalue: 3.4 Percent variance explained: 56.3 Direction: negative Consistency: 83-100 Number of measurements: 6 262 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION -0.86 R102 # CONTR LOCATIONS -0.84 ClOZA R102/R94 (differentiation of con- trollership locations) -0.72 ClOZB R102/R22 (proportion controller- ship locations) 0.28 C105 RIOS/R23)/(R94/R14) (geographical employee concentration) -0.71 R107 CONTR LOCATION DISTANCE -0.91 C107 R107/R25 (proportion contr loca- tion distance) "Accounting geographical dispersion" is a good explainer of the original measurements, with an eigenvalue of 3.4. It is most heavily loaded on proportion of controllership location distance, number of controllership locations, and differentiation of controllership loca- tions. It was determined to be negative based on its negative loadings on these three measurements. .All of the other measurements were con- sistent with the direction of the component except geographical employee concentration. .As was explained under "company geographical dispersion," there is no a priori notion of'how geographical employee concentration should vary with the other measures of geographical dispersion. Conse- quently, consistency, though good, cannot be determined exactly. It is interesting that the positive association of concentration measures with absolute number measures found fer the company as a whole ("company jOb structure complexity" and "company geographical dispersion") was not found for the accounting system ("accounting job structure complexity" and "accounting geographical dispersion"). 263 Component name: Accounting unit differentiation- horizontal Eigenvalue: 1.6 Percent variance explained: 54.4 Direction: positive Consistency: 100 Number of measurements: 3 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION 0.46 R108 LOWEST CONTR UNITS (number) 0.75 C108A R108/R94 (elaboration of lowest contr units) 0.93 C108B R108/R26 (elaboration of lowest contr units) Since all three of the original measurements for the principal components procedure which isolated "accounting unit differentiation- horizontal" are based on question 108 (page 231), the measurement rules which were used in interpreting responses to question 108 are especially important for determining the meaning of "accounting unit differentiation- horizontal." Question 108 asks about the number of lowest-level control- lership units in the company. One of the definitional rules fer question 108 stipulates that only accounting units in a direct authority line to the chief executive officer should be counted as lowest-level controller- ship units. This would exclude divisional accounting departments, re- sponsible to division heads. Consequently, "accounting unit differenti- ation-horizontal," whiCh is based on question 108, is defined as the horizontal differentiation of units within the central accounting depart- ment. Though unit specialization was not measured for the accounting system, it is probable that it would be positively related to "accounting unit differentiation-horizontal" since the multiple units of a central accounting department would prObably be assigned different functions. 264 “Accounting unit differentiation-horizontal" is a relatively poor explainer of the original measurements, having an eigenvalue of only 1.6. It is loaded most heavily on the two calculated measures of elaboration of lowest controllership units. The loadings on all three measurements are consistently positive, making the component positive. Since the three measures are closely related, the definition of the component is very tight, the horizontal elaboration of lowest-level controllership units within the central controller's department. Component name: .Authority levels Eigenvalue: 3.0 Percent variance explained: 73.9 Direction: negative Consistency: 100 Number of measurements: 4 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION -0.86 C109 R109/R51A (depth of lowest contr units) -0.90 C110 R110/R51A (depth of lowest contr emps) -0.81 ClllA (R110-R111)/R51A (vertical width contr fUnction) -0.87 ClllB (RlO9-Rlll)/R51A (vertical width contr function) ”Authority levels" is a good explainer of its original measure- ments, with an eigenvalue of 3.0. It explained 73.9 percent of the variance of its four measurements. It is loaded heavily and fairly equally on all the four measures, two of depth of the lowest controller- ship units and two of vertical width of the controllership function. It is a perfectly consistent negative component. 265 Component name: Report differentiation Eigenvalue: 4.3 Percent variance explained: 61.5 Direction: negative Consistency: 85 Number of measurements: 7 MEAS URE- LOADING MENT NAME DESCRIPTION AND CALCULATION 0.10* C112 R112A+R112B+R112C (report diver- s itY) -0.84 R113 PROFIT CENTERS -0.58 C113A R113/R96 (proportion profit centers) -0.86 C113B R113/R44 (profit center elab- oration) -0.88 R114 PRODUCT CENTERS -0.92 C114A R114/R96 (proportion product centers) -0.95 Cll4B R114/R9 (product center elab- oration) "Report differentiation" is a good explainer of its original seven measurements, with an eigenvalue of 4.3. It is most heavily loaded on three measures of product center elaboration and two meas- ures of profit center elaboration. The component is reasonably con- sistent. The only inconsistent loading is on report diversity (differ- ent types of reports produced), and it was very insignificant. Never- theless, it is quite surprising that report diversity is not positively associated with elaboration of product and profit centers. "Report differentiation" was determined to be negative based on its negative loadings on the five top loading measurements. Component name: Sophistication of accounting techniques Eigenvalue: 3.3 Percent variance explained: 55.2 Direction: positive Consistency: 83 Number of measurements: 6 266 MEASURE- LOADING MENT NAME 0.81 C117 0.94 C118 0.86 C119 0.88 R120 -0.50* R121 0.07 R122 DESCRIPTION AND CALCULATION (R117A+R117B+R117C)/3 (standard cost usage) WEIGHTED FIXED-VARIABLE WEIGHTED BREAKEVEN (use of cost- volume-profit analysis) COST-VOLUME-PROFIT (sophistica- tion of) BUDGET ITEMS INCLUDED CAPITAL PROJECT INFORMATION "Sophistication of accounting techniques" is an above-average explainer of its six original measurements, having an eigenvalue of 3.3. It is most heavily loaded on four of the six measurements: classifica- tion of costs as fixed and variable, use of cost-volume-profit analysis, sophistication of cost-volume-profit techniques, and standard cost usage. "Sophistication of accounting techniques" was determined to be positive based on its positive loadings on these four measurements. The only inconsistent loading was for extensiveness of the budgetary system. This inconsistent 1oading was reasonably large. Component name: Eigenvalue: Percent variance explained: Direction: Consistency: Number of measurements MEASURE- LOADING MENT NAME 0.91 C115 0.18 C116 0.91 C97B Decentralization of accounts 1.7 56.3 positive 100 3 DESCRIPTION AND CALCULATION R115/R51A (depth lowest cost reports) Rll6/R51A (depth lowest profit reports) PROPORTION REPORT LEVELS "Decentralization of accounts" seems to be a relatively poor explainer of its original three measurements, with an eigenvalue of 1.7. 267 It is heavily loaded on two of the three measurements, depth of lowest cost reports and proportion of report levels to which reports are pre- sented. Though consistent in direction with the other two measures, depth of lowest profit reports is nonsignificantly loaded on "decentral- ization of accounts." This may indicate that profit reports are not generally presented to levels below the top two. Though it poorly explains the depth of profit reports, "decentralization of accounts" explains well the other two measurements. "Decentralization of ac- counts" is determined to be perfectly consistent and positive based on its positive loadings on all three measurements. Component name: Unit differentiation-vertical Eigenvalue: 2.0 Percent variance explained: 97.8 Direction: negative Consistency: 100 Number of measurements: 2 MEASURE- LOADING MEN T NAME DESCRIPTION AND CALCULATION 0.99 C125A+ RlZS/RSlA (reliance on central controller's dept) 0.99 C125B+ (R125-Rlll)/R51A (reliance on central contr's dept) The intent of the researcher was to measure centralization of the accounting function in a way similar to the way Simon et al. meas- ured it in their study.1 Their "centralization" was measured by whether lower-level accounting department heads report to higher-level account- ing department managers or to operating managers. Thus their "central- ization" is not centralization as it is conventionally defined, the 1HerbertA. Simon, George Kozmetsky, Harold Guetzkow, and Gordon Tyndall, centralization vs. Decentralization in Organizing the Controller's Department (New York: Controllership Foundation, Inc., 1954), pp. 8-9. Tex-*6 has 10 268 level at which key decisions are made. Instead, their "centralization" has to do with the level at whiCh the accounting structure is attached to the overall organization structure. The question which was intended to measure the Simon et al. "centralization" is question 126 on page 235 (the answer is recorded as measurement R126). It inquired about whether accounting department heads below the central controller's department report to higher-level accounting managers or to operating management. Unfortunately, only seven of the eighteen companies had accounting departments below the central accounting department which satisfied the definitional rule (discussed on page 235). The rest of the answers to question 126 were thus considered missing, and.measurement R126 was discarded. The remaining two measurements of accounting function central- ization were developed by the researcher. They have to do with reliance on the central accounting department, as measured by the number of levels between the central accounting department and the next-lowest-level accounting departments. For the eleven companies without lower-level accounting departments, the number of levels was the total number of levels in the overall company structure minus the level of the central accounting department. In essence, the measurements are of whether the company has any lower-level accounting departments. Thus their meaning is more akin to accounting department unit differentiation than to centralization of authority. Consequently, for purposes of this study, the definition of "accounting unit differentiation-vertical” is Changed to "accounting system, unit differentiation-vertical." 269 As is specified in the definitional rule for question 125 (page 235), the definition of lower-level accounting departments for the pur- pose of this question is restricted to accounting departments for seg- ments (most likely divisions) of the company, as opposed to specialized accounting departments for the company as a.whole (such as budgeting departments, accounts receivable departments, etc.). Consequently, the definition of "accounting unit differentiation-vertical," based on ques- tion 125, is refined to the "divisional differentiation of the account- ing function." "Accounting unit differentiation-vertical" is an almost perfect explainer of the two original measurements, having an eigenvalue of 2.0. The loadings of the two measurements of reliance on the central control- ler's department are both almost 1. The component is perfectly consist- ent. The direction of "accounting unit differentiation-vertical" is determined to be negative since high numbers indicate there are no lower-level accounting departments and thus no unit differentiation. Component name: Mechanization Eigenvalue: 2.4 Percent variance explained: 79.7 Direction: negative Consistency: 100 Number of measurements: 3 MEASURE- LOADING MENT NAME DESCRIPTION AND CALCULATION -0.92 R128 COMPUTERIZED FINANCIAL ACTIVITIES (number of) -0.88 R129 MECHANIZED REPORTS (proportion) 0.88 Cl30+ R130/R98 (capital-labor mix) “MeChanization” is a reasonably good explainer of its three original measurements, having an eigenvalue of 2.4 (maximum possible, 270 3.0) and explaining 80 percent of the variance. It is heavily loaded on all three measurements: number of computerized financial activities, proportion of mechanized reports, and capital-labor mix. "Mechaniza- tion” is a perfectly consistent negative component, as is indicated by the negative loadings on all three measurements. Component names: Personnel-education; Personnel-general Eigenvalues: 1.8; 1.5 Percent variances explained: 22.5; 19.3 Directions: positive; positive Consistencies: 75; 75 Numbers of measurements: 8; 8 MEA S URE- LOADI N GS MEN T NAME DESCRIPTION AND CALCULATION 0.86 ; 0.32 C132 (R132-R53)/R53 (highest educ compared to div heads) 0.90 ; -0.20* C133 (R133/R94)/(R54/Rl4) (BA's com- pared to overall company) -0.20*; -0.56* R134 WEEKS TRAINING-CONTR 0.02 ; 0.25 C135 (R135-R56)/R56 (low contr senior- ity compared to dir lab) -0.3l*; 0.72 C136 (R136-R57)/R57 (high contr senior- ity compared to div hds) 0.19 ; 0.45 C137B (C137A-C58)/C58 (low contr com- pens compared to dir lab) 0.26 ; 0.23 C138LA (R130/R94)/(R42/R14) (aver contr sal compared with company) 0.06 ; 0.50 C138IB (R137B/Cl37A)-(R58A/C58) (low benefits comp with dir lab) The first component extracted fOr this group of eight measure- ments accounted for only 23 percent of the variance. Consequently, a second component was added.which increased the total explained variance to 42 percent. This level of explanation is the lowest for any group of measurements in the study and is consistent with the other set of personnel components, "company personnel quality-high level" and "com- pany personnel quality—low level," which had the second lowest level of 271 explanation, 47 percent. Apparently the disparate and indirect nature of the personnel quality measurements prevents higher levels of explana- tion. Nevertheless, it was necessary to discard other possible compo- nents in order to keep the number of variables down and the analysis simple. The two components were rotated using varimax rotation to pro- duce the two components, "accounting personnel quality-education" and "accounting personnel quality-general." Both were poor explainers, having eigenvalues of only 1.8 and 1.5, respectively. "Accounting personnel quality-education" is loaded heavily on only two of the eight measurements: highest-level education compared to division heads, and BA degrees compared to overall company. The high loadings on the education measurements indicate the component has to do with the "educational qualifications" of accounting system employees. Positive loadings on the two high-loading measurements determine the direction to be positive. Two of the measurements are inconsistently loaded on the component: weeks training of controllership employees, and high-level controllership seniority compared to division heads. Neither of these inconsistent loadings is large, but both can.be ex- plained. Senior personnel tend to have less educational qualifications. Consequently, companies with more senior personnel tend to have person- nel with lower educational qualifications. .Also, companies which can hire educationally qualified personnel need training programs less. Thus "accounting personnel quality-education" can be considered rea- sonably consistent. 272 ”Accounting personnel quality-general" is loaded most highly on three measurements: weeks training of controllership employees, high- level controllership seniority compared to division heads, and low-level employees' fringe benefits compared with direct labor. The positive loadings on the seniority and fringe benefits measures determine the component to be positive, and this is confirmed.by the majority of positive loadings (six out of the eight loadings). Unfortunately, the weeks training of controllership employees is inconsistently and sig- nificantly loaded on "accounting personnel quality-general." This is consistent with training's inconsistent loading on "accounting personnel quality-education." The explanation is analogous: smoothly running accounting systems with experienced and well-paid personnel do not need fermal training programs. This explanation of training as just a re- medial device is only speculation which needs to be tested in other studies. The only other inconsistent loading is BA degrees compared to overall company, which is nonsignificant. Its significant associa- tion with "accounting personnel quality-education" suggests we can ignore it for "accounting personnel quality-general." Thus "accounting person- nel quality-general" can be considered reasonably consistent. APPENDIX C OTHER INFORMATION APPENDIX C OTHER.INFORMATION Dear Mr we are conducting a study to determine how accounting systems are inter— related with other managerial control systems in companies and how these relationships differ in firms with different characteristics. We are collecting infOrmation for this study by means of interviews with com- pany controllers and other company executives. The findings of the study may help us understand.more clearly the need for accounting con- trols and perhaps the "trade offs" involved in selecting between account- ing and other control devices. we would like to ask your assistance in this study by allowing us to visit with you. The enclosed short paper explains more about the project. .Mr. Rosenzweig will be the primary researcher and will be using the data as part of his doctoral disserta- tion here at Michigan State University. we are contacting you because your firm, as best we can determine, fits the size and manufacturing emphasis we are studying. we would like to obtain the research data needed from you through an interview at your firm. The questions are not concerned with attitudes but rather with the structural and financial characteristics of your company. You.may have to refer to your records or other officials fer some answers. we think you will find the questions interesting. The answers will not be identified with your company in the research re- port, and they will be kept in the strictest confidence. we anticipate having 20 to 25 companies participate. we will be happy to send you a copy of the research results as soon as they are assembled. we hope the results will be both interesting and useful to you and your company. we are restricting the current study to small or medium-size manufactur- ing companies (200 to 5,000 employees) which are not subsidiaries of other companies. we will call you in about a week. If you are willing to help us, we would like to set a date for the interview and answer any other questions you.may have at that time. Your cooperation will advance knowledge of the accounting control proc- ess. In addition, we hope you will also benefit from.participation in this research effort. Cordially yours, Harold Sollenberger Kenneth Rosenzweig Associate Professor of Accounting Doctoral Student 273 274 Table 19 Assignment of Measurements to Principal Components Procedures Measure- Description and ggéegf Dispositiona ment Name Calculation vations INT- ORDER COMPANY INTERVIEW ORDER 18 2 R1 ORGANIZATION CHART 18 2 R2 FINANCIAL STATEMENTS 18 2 R3 PERSONNEL LIST 18 Z R4 SELLING PRICE 18 Process sophistication R5 MATERIALS COST 18 Process sophistication C5A R4-R5 (value added) 18 Process sophistication CSB (R4-R5)/R4 (proportion 18 Process sophistication value added) R6 RESEARCH COSTS 18 X C6 R6/R4 (research 18 Process sophistication emphasis) R7 MANUFACTURING COSTS 18 X C7 R6/R7 (research 18 Process sophistication emphasis) R8 PRODUCTION CYCLE-length 18 Process sophistication R9 NUMBER PRODUCTS 18 Process-output diversity R10 NUMBER CUSTOMERS 18 Process-output diversity Rll NUMBER SUPPLIERS 18 iMaterials input diversity (used as is) ax = dropped once used fer calculation of new variable; Y = drOpped due to too few observations; 2 = requests for general infor- matione—not used in subsequent analysis. 275 Table 19 (Cont'd.) Measure- Description and ggéegf Disposition ment Name Calculation vations R12 NUMBER EMPLOYEES NOW 18 Company size R13 EMPLOYEE FLUCTUATION 18 X R14 NUMBER EMPLOYEES AVERAGE 18 Company size R15 REVENUES 18 Company size R16 EXPENSES 18 Company size R17 ASSETS 18 Company size R18 EQUITY 18 Company size R19 JOB TITLES 18 X C19 R19/R14 (elaboration 18 Company job structure complexity of job titles) R20 HIGHEST JOB TITLE 18 X C20 R20/Rl4/1/R19 (employee 18 Company job structure complexity job concentration) R21 NEXT JOB TITLE 18 X C21 R21/(Rl4-R20)/l/(R19-l) 18 Company job structure complexity (employee job concen- tration) R22 LOCATIONS-number 18 Company geographical dispersion C22 R22/R14 (differentiation 18 Company geographical dispersion of locations) R23 HIGHEST LOCATION 18 X EMPLOYEES C23 R23/Rl4/l/R22 (geograph- 18 Company geographical dispersion ical employee concen- tration) 276 Table 19 (Cont'd.) . . No of Measure- Description and ‘ . . . . Obser- DispOSition ment Name Calculation vations R24 SECOND LOCATION 13 X EMPLOYEES C24 R24/(Rl4-R23)/1/(R22-l) 13 Y (geographical employee concentration) R25 DISTANCE BETWEEN 18 Company geographical dispersion LOCATIONS R26 FIRST LINE SUPERVISORS 18 X C26 R26/R14 (lowest level 18 Company divisional differenti- unit differentiation) ation R27 CHIEF SPAN OF CONTROL 18 Company divisional differenti- ation R28 DIVISION HEADS-number 18 Company divisional differenti- ation C28 R28/R27 (highest level 18 Company divisional differenti- unit differentiation) ation R29 PRODUCTION FUNCTIONS 18 X C29 R29/R26 (lowest level 18 Company divisional specialization unit specialization) R30 DIVISION RESPONSIBILI- 18 Company divisional specialization TIES-products, func- tions, geography R31 OPERATING DIVISIONS 18 Company divisional differenti- ation C31 (R28-R3l)/R28 (propor- 18 Company divisional specialization tion nonoperating divisions) R32 PRODUCTION DIVISIONS 18 Company divisional differenti- ation 277 R34 R35A R35B R36 C36A C36B R37 R38 C38A C38B R39 C39A C39B R40 tion nonproduction divisions) OUTPUT TO OTHER DIVISIONS MECHANIZATION BULK MECHANIZATION MOST ENERGY EXPENSES R36/Rl6 (energy inten- siveness) R36/R4 (energy inten- siveness) NUMBER.COMPUTERS COMPUTER EXPENSE R38/R16 (computeriza- tion) R38/R4 (computerization) COMPUTER.HARDWARE EXPENSE R39/R16 (computeriza- tion) R39/R4 (computerization) ELECTRIC TYPEWRITERS TABLE 19 (Cont'd.) Measure- Description and $3; of D' 't' ment Name Calculation er ISPOSI 10“ vations C32 (R28-R32)/R28 (propor- 18 Company divisional specialization 11 Y 18 Company mechanization-general; Company mechanization-computers 18 Company mechanization-general; Company mechanization-computers 18 X 18 Company mechanization-general; Company mechanization-computers 18 Company mechanization-general; Company mechanization-computers 18 Company mechanization-general; Company mechanization-computers 18 X 18 Company mechanization-general; Company mechanization-computers 18 Company mechanization-general; Company mechanization-computers 18 X 18 Company mechanization-general; Company mechanization-computers 18 Company mechanization-general; Company mechanization-computers 18 X 278 TABLE 19 (Cont'd.) .Measure- Description and EE' 0f , , . ment Name Calculation ser- D15P°51t1°n vations C40 R40/R14 (electric type- 18 Company mechanization-general; writers per employee) Company mechanization-computers R41 FIXED CAPITAL 18 X C4lA R41/R14 (fixed capital 18 Company mechanization-general; per employee) Company mechanization-computers C41B R41/R17 (proportion 18 Company mechanization-general; assets fixed) Company mechanization-computers R42 WAGE AND SALARY 18 X R43 DEPRECIATION 18 X C43A R43/R42 (depreciation/ 18 Company mechanization-general; wage and salary expense) Company mechanization-computers C43B R15/R41 (capital asset 18 Company mechanization-general; turnover) Company mechanization-computers R44 SUPERVISORS 18 X C44 R44/R14 (supervisory 18 Company direct supervision ratio) R45 FIRST LINE SPAN 18 Company direct supervision R46 LOWEST LEVEL EMPLOYEES 18 X R47 NEXT LOWEST LEVEL 18 X EMPLOYEES C47 R46/R47 (ratio employees 18 Company direct supervision lst 2 levels) R48 DIRECT LABOR EMPLOYEES 18 X C48 (R14-R44-R48)/Rl4 (staff 18 Company staff support ratio) R49 STAFF FUNCTIONS (differ— 18 Company staff support entiation of) 279 Tdfle19(&mfldJ Measure- Description and ggéegf Disposition ment Name Calculation vations R50 CLERICAL EMPLOYEES 18 X C50 R50/R14 (clerical ratio) 18 Company staff support R51A LEVELS MOST 18 Company authority levels RSlB LEVELS AVERAGE 18 Company authority levels R52 DIRECT LABOR EDUCATION 18 Company personnel quality-high level; Company personnel quality- low level R53 DIVISION HEAD EDUCATION 18 Company personnel quality-high level; Company personnel quality- low level C53 R53-R52 (difference 18 Company personnel quality-high high-low education) level; Company personnel quality- low level R54 BA DEGREES 18 X C54 R54/Rl4 (proportion BA 18 Company personnel quality-high degree employees) level; Company personnel quality- low level R55 WEEKS TRAINING 18 Company personnel quality—high level; Company personnel quality- low level R56 DIRECT LABOR SENIORITY 18 Company personnel quality-high level; Company personnel quality- low level R57 DIVISION HEAD SENIORITY 18 Company personnel quality-high level; Company personnel quality- low level C57 R57-RS6 (difference div 18 Company personnel quality-high head and dir lab senior— level; Company personnel quality- ity) low level R58A DIRECT LABOR WAGE 18 X 280 Table 19 (Cont'd.) ‘Measure- Description and Uggegf Disposition ment Name Calculation vations R58B DIRECT LABOR BENEFITS 18 X C58 R58A+R58B (direct labor 18 Company personnel quality—high compensation) level; Company personnel quality- low level RS9A 2ND LEVEL SALARIES 17 X R59B 2ND LEVEL BENEFITS 17 X R59C 2ND LEVEL BONUS 16 X C59DA R59A+R59B+R59C (division 15 Company personnel quality-high head compensation) level; Company personnel quality- low level C59DB C59DA-C58 (difference 15 Company personnel quality-high high-low compensation) level; Company personnel quality- low level CSQIA R42/R14 (average compen- 18 Company personnel quality-high sation) level; Company personnel quality- low level C59IB R58B/C58 (proportion 18 Company personnel quality-high direct labor benefits) level; Company personnel quality- low level CSQIC R59B/C59DA (proportion 15 Company personnel quality-high division head benefits) level; Company personnel quality- low level C591D R59C/C59DA (proportion 15 Company personnel quality-high division head bonus) level; Company personnel quality- low level R60A TOP 1% SALARIES 16 X C60A R60A/R42 (compensation 16 Company centralization-investment; concentration-top 1%) Company centralization-purChasing R60B TOP 2% SALARIES 16 X 281 Table 19 (Cont'd.) Measure- Description and 33se2f Disposition ment Name Calculation vations C60B R60B/R42 (compensation 16 Company centralization-investment; concentration-top 2%) Company centralization-purchasing R60C TOP 25% SALARIES 17 X C60C R60C/R42 (compensation 17 Company centralization-investment; concentration-top 25%) Company centralization-purchasing R6lA INVESTMENT AUTHORITY- 18 X $100 C61A R6lA/R51A (decentraliza- 18 Company centralization-investment; tion-investment auth- Company centralization-purchasing $100) R61B INVESTMENT AUTHORITY- 18 X $1,000 C6lB R6lB/R51A (decentraliza- 18 Company centralization-investment; tion-investment auth- Company centralization-purchasing $1,000) R61C INVESTMENT AUTHORITY- 18 X $10,000 C6lC R6lC/R51A (decentraliza- 18 Company centralization-investment; tion-investment auth- Company centralization-purchasing $10,000) R62 CAPITAL EXPENDITURES 18 X R63A INVESTMENT AUTHORITY-1% 18 X C63A R63A/R51A (decentraliza- 18 Company centralizationlinvestment; tion-investment auth-l%) Company centralization-purchasing R63B INVESTMENT AUTHORITY~5% 18 X C63B R63B/R51A (decentraliza- 18 Company centralization-investment; tion-investment auth-5%) Company centralization-purchasing R6 3C INVESTMENT AUTHORITY-25% 18 X 282 Tmfle19(&mfldJ . . No. of page. ”92:30:22..“ i-sioisiion vations C63C R63C/R51A (decentraliza- 18 Company centralization-investment; tion-investment auth- Company centralization-purchasing 25%) R64 PRICING AUTHORITY 18 X C64 R64/R51A (decentraliza- 18 Company centralization-investment; tion-pricing authority) Company centralization-purchasing R65A PURCHASE AUTHORITY-$100 18 X C65A R65A/R51A (decentraliza- 18 Company centralization-investment; tion-purchase auth-$100) Company centralization-purchasing R65B PURCHASE AUTHORITY- 18 X $1,000 C658 R65B/R51A (decentraliza- 18 Company centralization-investment; tion-purchase auth- Company centralization-purchasing $1,000) R65C PURCHASE AUTHORITY- 18 X $10,000) C65C R65C/R51A (decentraliza- 18 Company centralization-investment; tion-purchase auth- Company centralization-purchasing $10,000) R66 MATERIALS COST 18 X R67A PURCHASE AUTHORITY-1% 17 X C67A R67A/R51A (decentraliza- 17 Company centralization—investment; tion-purchase auth-l%) Company centralization-purchasing R678 PURCHASE AUTHORITY-5% 17 X C67B R67B/R51A (decentraliza- 17 Company centralization-investment; tion-purdhase auth-5%) Company centralization-purchasing R67C PURCHASE AUTHORITY-25% 17 X 283 Table 19 (Cont'd.) . . No. of Measure- Description and . . . . Obser- Disp051tion ment Name Calculation vations C67C R67C/R51A (decentraliza- 17 Company centralization-investment; tion-purchase auth-25%) Company centralization-purchasing R68 DIVISION BUDGET 18 X PROPOSALS C68 R68/R28 (divisional 18 Company centralization-investment; budget participation) Company centralization-purchasing R69 BUDGET PROPOSALS BELOW 18 Company centralization-investment; DIVISION Company centralization-purchasing R70 PROCEDURES MANUAL 18 Company standardization-jObs; Company standardization-general R71 PROCEDURES MANUAL PAGES 18 X R72 PROCEDURES MANUAL WORDS 8 X C72 R7IXR72 (total proce- 18 Company standardization-jobs; dures manual words) Company standardization-general R73 PERSONNEL PROCEDURES 18 Company standardization-jObs; MANUAL Company standardization-general R74 PERSONNEL PROCEDURES 18 X .MANUAL PAGES R75 PERSONNEL PROCEDURES 10 X MANUAL WORDS C75 R74XR75 (total person- 18 Company standardization-jObs; nel procedures manual Company standardization-general words) R76 PERSONNEL RATING FORM 18 Company standardization-jabs; Company standardization—general R77 DIRECT LABOR RATING 18 Company standardization-jobs; Company standardization-general R78 RATING FREQUENCY 6 Y 284 Table 19 (Cont'd.) Measure- Description and Ugéegf Disposition ment Name Calculation vations R79 WRITTEN EMPLOYMENT 18 X CONTRACTS C79 R79/R14 (proportion 18 Company standardization-jObs; written contract em- Company standardization-general ployees) R80 WRITTEN EMPLOYMENT 18 X CONTRACTS-DIRECT C80 R80/R48 (proportion 18 Company standardization-jObs; written contr direct Company standardization-general employees) R81 NUMBER.UNIONS 18 Company standardization-jObs; Company standardization-general R82 UNION MEMBERS 18 X C82 R82/R14 (proportion 18 Company standardization-jObs; union members) Company standardization-general R83 UNION MEMBERS-DIRECT 18 X C83 R83/R48 (proportion 18 Company standardization-jObs; direct labor union mem- Company standardization-general bers) R84 INFORMATION BOOKLETS 18 X C84 R84/R14 (proportion in- 18 Company standardization-jObs; formation booklet eme Company standardization-general ployees) R85 ORGANIZATION CHARTS 18 X C85 R85/R14 (proportion 18 Company standardization-jObs; organization chart Company standardization-general employees) R86 JOB DESCRIPTIONS 17 X 285 Table 19 (Cont'd.) . . No. of Measure- Description and . . . - Obser- DispOSition ment Name Calculation vations C86 R86/R14 (proportion job 17 Company standardization-jabs; description employees) Company standardization-general R87 JOB DESCRIPTIONS-DIRECT 17 X C87 R87/R48 (proportion 17 Company standardization-jObs; direct job description Company standardization-general employees) R88 WRITTEN POLICIES 18 Company standardization-jObs; Company standardization-general R89 PRODUCTION SCHEDULE 18 Company standardization-jObs; Company standardization-general R90 WAGE GRADES 13 Y R91 WAGE INCREASE BASIS- 18 Company standardization-jObs; seniority, merit, Company standardization-general negotiation R92 FORMS CONTROL 18 Company standardization-jObs; Company standardization—general R93 TIME AND MOTION STUDIES 18 Company standardization-jabs; Company standardization-general R94 CONTROLLERSHIP EMPLOYEES 18 X C94 R94/R14 (proportion con- 18 Accounting size—resource input; trollership employees) Accounting size-infOrmation output R95 EMPLOYEES RECEIVING 18 X REPORTS C95 R95/R14 (proportion em— 18 Accounting size-resource input; ployees receiving Accounting size-infOrmation reports) output R96 DATA CENTERS 18 Accounting size-resource input; Accounting size-infOrmation output 286 Table 19 (Cont'd.) Measure- Description and 33s of D' s't' ment Name Calculation er ISPO 1 ion vations C96 R96/(R9+R44) (data cen- 18 Accounting size-resource input; ter elaboration) Accounting size-information output R97A REPORT FREQUENCY-TOP 18 X R978 REPORT FREQUENCYAMIDDLE 12 X R97C REPORT FREQUENCY-BOTTOM 4 X C97A (R97A+R97B+R97C)/3 (av- 18 Accounting size-resource input; erage report frequency) Accounting size-infOrmation output C978 PROPORTION REPORT LEVELS l8 Decentralization of accounts R98 CONTROLLERSHIP EXPENSES 17 X C98A R98/R16 (proportion con- 17 Accounting size-resource input; trollership expenses) Accounting size-infOrmation output C988 R98/R4 (controllership 17 Accounting size-resource input; expense emphasis) .Accounting size-infOrmation output R99 CONTROLLERSHIP JOB 18 X TITLES C99A R99/R94 (elaboration of 18 Accounting job structure complex- job titles) ity C998 R99/R19 (proportion con- 18 Accounting jOb structure complex- trollership job titles) ity R100 HIGHEST JOB TITLE-CONTR 18 X C100 R100/R94/l/R99 (employee 18 Accounting job structure complex- job concentration) ity R101 NEXT JOB TITLE-CONTR 18 X 287 Table 19 (Cont'd.) Measure' Descripti‘m and ggéegf Dis osition ment Name Calculation . p vations C101 R101/(R94—RlOOYl/(R99-l) 18 Accounting job structure complex- (employee job concentra- ity tion) R102 # CONTR LOCATIONS 18 Accounting geographical disper- sion C102A R102/R94 (differentia- 18 Accounting geographical disper- tion of locations) sion C1028 R102/R22 (proportion 18 .Accounting geographical disper- controllership loca- sion tions) R103 HIGHEST CONTR LOCATION 10 X EMPLOYEES C103 R103/R94/l/R102 (geo- 10 Y graphical employee concentration) R104 SECOND CONTR LOCATION 7 X EMPLOYEES C104 R104/(R94-R103J/l/(R102- l) 7 Y (geog employee concen- tration) R105 HIGHEST LOCATION CONTR 18 X EMPLOYEES C105 RlOS/RZ3/R94/Rl4 (geo- 18 Accounting geographical disper- graphical employee sion concentration) R106 SECOND LOCATION CONTR 13 X EMPLOYEES C106 R106/R24/R94/Rl4 (geo- 13 Y graphical employee concentration) 288 Table 19 (Cont'd.) . . No. of Measure- Description and . . . . Obser— DispOSition ment Name Calculation vations R107 CONTR LOCATION DISTANCE 17 .Accounting geographical disper- sion C107 R107/R25 (proportion 17 Accounting geographical disper- contr location distance) sion R108 LOWEST CONTR.UNITS 18 Accounting unit differentiation- (number) horizontal C108A R108/R94 (elaboration of 18 .Accounting unit differentiation- lowest contr units) horizontal C1088 R108/R26 (elaboration of 18 Accounting unit differentiation- lowest contr units) horizontal R109 LEVEL LOWEST CONTR UNITS 18 X C109 R109/R51A (depth of 18 Accounting authority levels lowest contr units) R110 LEVEL LOWEST CONTR EMPS 18 X C110 RllO/RSlA (depth of 18 .Accounting authority levels lowest contr employees) Rlll LEVEL CHIEF FIN OFFICER 18 X ClllA (R110-Rlll)/R51A (ver- 18 Accounting authority levels tical width contr fUnction) C1118 (R109-Rlll)/R51A (ver- 18 Accounting authority levels tical width contr fUnction) RllZA INFO PROVIDED-TOP 18 X R1128 INFO PROVIDED-MIDDLE 18 X R112C INFO PROVIDED-BOTTOM 18 X C112 R112A+R1128+R112C (re- 18 Accounting report differentiation port diversity) 289 Table 19 (Cont'd.) . . No. of Measure- Description and . . . . Obser— Disp051tion ment Name Calculation vations R113 PROFIT CENTERS 18 .Accounting report differentiation Cll3A R113/R96 (proportion 18 Accounting report differentiation profit centers) C1138 R113/R44 (profit center 18 Accounting report differentiation elaboration) R114 PRODUCT CENTERS 18 .Accounting report differentiation C114A R114/R96 (proportion 18 Accounting report differentiation product centers) C1148 R114/R9 (product center 18 Accounting report differentiation elaboration) R115 LOWEST COST REPORT LEVEL 18 X C115 R115/R51A (depth lowest 18 Decentralization of accounts cost reports) R116 LOWEST PROFIT REPORT 18 X LEVEL C116 R116/R51A (depth lowest 18 Decentralization of accounts profit reports) R117A STANDARD MATERIAL COST 18 X R1178 STANDARD LABOR COST 18 X Rll7C STANDARD OVERHEAD COST 18 X C117 (R117A+Rll7B+Rll7C)/3 18 Sophistication of accounting (standard cost usage) techniques R118A FIXED4VARIABLE-COMPANY 18 X WIDE R1188 FIXED4VARIABLE-PRODUCT 18 X LINES R118C FIXEDéVARIABLE-UNITS 18 X 290 Table 19 Cont'd.) Measure- Description and ggéegf Dis osition ment Name Calculation . p vations C118 WEIGHTED FIXED4VARIA8LE 18 Sophistication of accounting techniques R119A BREAKEVEN-COMPANY WIDE 18 X R1198 BREAKEVEN-PRODUCT LINES 18 X R119C BREAKEVEN-UNITS 18 X C119 WEIGHTED BREAKEVEN (use 18 Sophistication of accounting of cost-vol-profit techniques analysis) R120 COST4VOLUME-PROFIT 18 Sophistication of accounting (sophistication of) techniques R121 BUDGET ITEMS INCLUDED l8 Sophistication of accounting techniques R122 CAPITAL PROJECT INFOR- 18 Sophistication of accounting MATION techniques R123 NONCONTROLLABLE EXPENSES 16 Y R124 NONCONTROLLABLE 7 Y SEPARATED R125 LEVEL CONTR UNIT HEADS 18 X C125A R125/R51A (reliance on 18 .Accounting unit differentiation- central controller's vertical dept) C1258 (R125-Rlll)/R51A (reli- 18 .Accounting unit differentiation- ance on cent contr dept) vertical R126 HEADS REPORT TO 7 Y R127 CONTR COMPUTER TIME 12 Y R128 COMPUTERIZED FINANCIAL 18 Accounting mechanization ACTIVITIES (number of) 291 Table 19 (Cont'd.) contr seniority comp to div heads) . . No. of IMeasure- Description and . . . ment Name Calculation ObSPT' DISPOSItlon vations R129 MECHANIZED REPORTS 18 Accounting mechanization (proportion) R130 CONTROLLERSHIP WAGES 17 X C130 R130/R98 (capital-labor 17 Accounting mechanization mix) R131 LOWEST CONTR EDUCATION 18 X C131 (Rl3l-R52)/R52 (lowest 13 Y educ compared to dir lab) R132 HIGHEST CONTR EDUCATION 17 X C132 (R132-R53)/R53 (highest 17 Accounting personnel quality- educ compared to div education; Accounting personnel heads) quality-general R133 CONTROLLERSHIP 8A 18 X DEGREES C133 R133/R94/R54/Rl4 (BA's 18 Accounting personnel quality- compared to overall education; Accounting personnel company) quality-general R134 WEEKS TRAINING-CONTR 18 Accounting personnel quality- education; Accounting personnel quality-general R135 CONTR SENIORITY 17 X C135 (R135-R56)/R56 (low 17 Accounting personnel quality- contr seniority comp education; Accounting personnel to dir lab) quality-general R136 HIGHEST CONTR SENIORITY 18 X C136 (R136-R57)/R57 (high 18 Accounting personnel quality- education; Accounting personnel quality-general 292 Table 19 (Cont'd.) . . No. of Measure- Description and . . . ment Name Calculation Obser- DISPOSItlon vations R137A CONTR SALARY-LOWEST 17 X R1378 CONTR BENEFITS-LOWEST 17 X C137A R137A+Rl378 (contr com- 17 X pensation-lowest) C1378 (Cl37A-C58)/C58 (low 17 Accounting personnel quality- contr compens comp education; Accounting personnel to dir lab) quality-general R138A CONTR SALARY-HIGHEST 16 X R1388 CONTR BENEFITS-HIGHEST 16 X R138C CONTR BONUS-HIGHEST 16 X C1380A R138A+Rl38B+R138C (con- 16 X troller's compensation) Cl38D8 (Cl3SDA-C59DA)/C59DA 14 Y (compared to division heads) Cl38IA Rl30/R94/R42/Rl4 (aver 17 Accounting personnel quality- contr sal comp with education; Accounting personnel company) quality-general Cl3818 (R137B/Cl37A)-(R58A/C58) 17 .Accounting personnel quality- (low benefits comp dir education; Accounting personnel lab) quality-general Cl38IC (R1388/Cl38DA)- 14 Y (R59B/C59DA) (high bens comp diV'hdS) C1381D (R138C/C138DA)- 14 Y (R59C/C59DA) (high bons comp div hds) 293 Table 20 Measurements Included in Componentsa . Measure- . . . Loading ment Name Description and Calculation Process sophistication 0.97 R4 SELLING PRICE 0.90 R5 MATERIALS COST 0.97 C5A R4-R5 (value added) 0.40 CSB (R4-R5)/R4 (proportion value added) -0.29* C6 R6/R4 (research emphasis) 0.01 C7 R6/R7 (research emphasis) 0. 78 R8 PRODUCTION CYCLE-length Process-Output Diversity -0.81 R9 NUMBER PRODUCTS -0.8l R10 NUMBER CUSTOMERS Company Size -O.95 R12 NUMBER EMPLOYEES NOW -0.96 R14 NUMBER EMPLOYEES AVERAGE -0.99 R15 REVENUES -0.98 R16 EXPENSES -0.98 R17 ASSETS -0.9s R18 EQUITY Company Job Structure Complexity -0.80 C19 R19/R14 (elaboration of jdb titles) -0.82 C20 (R20/R14)/(l/R19) (employee jOb concentration) -0.92 C21 (R21/[Rl4-R20])/(l/[R19-1]) (employee jdb concentration) Company Geographical Disperion -0.95 R22 LOCATIONS-number -O.58 C22 R22/R14 (differentiation of locations) -0.90 C23 (R23/R14)/(l/R22) (geographical employee concentration) -0.85 R25 DISTANCE BETWEEN LOCATIONS aLarge loadings are italicized. * = The sign of the loading is inconsistent with the direction of the component. + = The measurement varies in the opposite direction from the concept of interest; interpret the loading as having the opposite sign. 294 Table 20 (Cont'd.) . Measure- . . . Loading ment Name Description and Calculation company Divisional Differentiation 0.15* C26 R26/R14 (lowest level unit differentiation) -0.83 R27 CHIEF SPAN OF CONTROL -0.97 R28 DIVISION HEADS—number -0.58 C28 R28/R27 (highest level unit differentiation) -0.97 R31 OPERATING DIVISIONS-number -0.94 R32 PRODUCTION DIVISIONS-number Company Divisional Specialization 0.6l* C29 R29/R26 (lowest level unit specialization) 0.85 R30+ DIVISION RESPONSIBILITIES-products, functions, geography -0.73 C31 (R28-R3l)/R28 (proportion nonoperating divisions) -0.89 C32 (R28-R32)/R28 (proportion nonproduction 36 , 0.01 21 , 0.15 80 , -0.34* .45 , 0.70 04*, 0.80 22*, 0.79 37 , 0.79 30*; 0.77 37 ; 0.79 43*; -0.11* 92 ; 0.12 87 ; 0.08 79 , -0.25* .44 ; -0.28 0.63 -0.89 -0.91 divisions) Company Mechanization-General Company Mechanization-computers R3&A R35B C36A C36B R37 C38A C388 C39A C39B C40 C4lA C41B C43A C43B+ C44 R45+ C47+ MECHANIZATION BULK MECHANIZATION MOST R36/Rl6 (energy intensiveness) R36/R4 (energy intensiveness) NUMBER.OOMPUTERS R38/R16 (computerization) R38/R4 (computerization) R39/R16 (computerization) R39/R4 (computerization) R40/R14 (electric typewriters per employee) R41/R14 (fixed capital per employee) R41/R17 (proportion assets fixed) R43/R42 (depreciation/wage and salary expense) R15/R41 (capital asset turnover) company Direct Supervision R44/R14 (supervisory ratio) FIRST LINE SPAN R46/R47 (ratio employees lst 2 levels) 295 Table 20 (Cont'd.) Loading 11226135111123“; Description and Calculation Company Staff'Support -0.27* C48 (R14—R44-R48)/R14 (staff ratio) 0.86 R49 STAFF FUNCTIONS (differentiation of) 0.80 C50 R50/R14 (clerical ratio) company Authority Levels -0.93 R51A LEVELS MOST -0.93 R518 LEVELS AVERAGE company Personnel Quality-High Level Company Personnel Quality-Low Level 0.06 ; 0.64 R52 DIRECT LABOR EDUCATION 0.52 ; -0.03* R53 DIVISION HEAD EDUCATION 0.05 ; -0.6S* CS3 R53-R52 (difference high-low education) 0.40 ; —0.10* C54 R54/R14 (proportion BA degree employees) -0.02*; 0.48 R55 WEEKS TRAINING 0.12 ; 0.13 R56 DIRECT LABOR SENIORITY -0.49*; 0.49 R57 DIVISION HEAD SENIORITY -0.56*; 0.38 C57 R57-R56 (difference div. head and dir. lab. seniority) 0.00 ; 0.79 C58 R58A+R58B (direct labor compensation) 0.87 ; 0.35 C59DA R59A+R598+R59C (division head compensation) 0.91 ; 0.18 C59DB C59DA-C58 (difference high-low compensation) 0.14 ; 0.78 C59IA R42/R14 (average compensation) 0.42 ; 0.48 C5918 R588/C58 (proportion direct labor benefits) -0.35*; 0.48 C59IC R598/C59DA (proportion division head benefits) 0.90 ; 0.11 C591D R59C/C59DA (proportion division head bonus) company centralization-Investment Company centralization-Purchasing -0.29 ; 0.54* C60A R60A/R42 (compensation concentration-top 1%) -0.24 ; 0.64* C608 R608/R42 (compensation concentration-top 2%) -0.11 ; 0.15* C60C R60C/R42 (compensation concentration-top 25%) 0.73 ; 0.22 C61At R6lA/R51A (decentralization-investment auth- $100) 0.87 ; 0.08 C6lB+ R618/R51A (decentralization-investment auth— $1,000) 0.82 ; -0.07* C61C+ R61C/R51A (decentralization-investment auth- $10,000) 296 Table 20 (Cont'd.) Loading INSRPNEEE: Description and Calculation 0.88 ; 0.09 C63A+ R63A/R51A (decentralization-investment auth- 1%) 0.70 ; 0.10 C638+ R63B/R51A (decentralization-investment auth- 5%) 0.64 ; 0.08 C63C+ R63C/R51A (decentralization-investment auth- 25%) 0.42 ; 0.17 C64+ R64/R51A (decentralization-pricing authority) 0.47 ; 0.51 C65A+ R6SA/R51A (decentralization-purchase auth- $100) 0.56 ; 0.46 C658+ R658/R51A (decentralization-purChase auth- $1,000) 0.47 ; 0.74 C65C+ R65C/R51A (decentralization-purChase auth- $10,000) 0.35 ; 0.82 C67A+ R67A/R51A (decentralization-purchase auth- 1%) 0.11 ; 0.82 C678+ R67B/R51A (decentralization-purchase auth- 5%) 0.24 ; 0.61 C67C+ R67C/R51A.(decentralization-purchase auth- 25%) 0.21 ; -0.02* C68+ R68/R28 (divisional budget participation) -0.57 ; 0.22* R69 BUDGET PROPOSALS BELOW DIVISION company Standardization-JObs Company Standardization-General 0.01*; 0.84 R70+ PROCEDURES MANUAL -0.22*; -0.85 C72 R7IXR72 (total procedures manual words) 0.33*; 0.75 R73+ PERSONNEL PROCEDURES MANUAL -0.04*; -0.85 C75 R74XR75 (total personnel procedures manual words) 0.73*; 0.04 R76+ PERSONNEL RATING FORM 0.72*; -0.14 R77+ DIRECT LABOR RATING 0.93 ; 0.19* C79 R79/R14 (proportion written contract employees) 0.89 ; 0.23* C80 R80/R48 (proportion written contract direct employees) 0.26 ; -0.13 R81 NUMBER UNIONS 0.95 ; 0.21* C82 R82/R14 (proportion union members) 0.90 ; 0.21* C83 R83/R48 (proportion direct labor union members) -0.l4*; -0.24 C84 R84/R14 (proportion infOrmation booklet employees) -0.33*; -0.24 C85 R85/R14 (proportion organization Chart employees 297 Table 20 (Cont'd.) Loading "glisligge Description and Calculation -0.78*; -0.18 C86 R86/R14 (proportion job description employees) -0.69*; -0.12 C87 R87/R48 (proportion direct job description employees) -0.07 ; 0.50 R88+ WRITTEN POLICIES 0.19*; 0.26 R89+ PRODUCTION SCHEDULE -0.53 ; -0.49* R91 WAGE INCREASE BASIS-seniority, merit, negotiation 0.26*; 0.73 R92+ FORMS CONTROL -0.09 ; 0.20 R93+ TIME AND MOTION STUDIES Accounting Size-Information Output Accounting Size-Resource Input 0.26 ; 0.84 C94 R94/R14 (proportion controllership employees) 0.72 ; 0.09 C95 R95/R14 (proportion employees receiving reports) 0.75 ; -0.13* R96 DATA CENTERS 0.58 ; 0.01 C96 R96/(R9+R44) (data center elaboration) -0.10 ; 0.42* C97A+ (R97A+R97B+R97C)/3 (average report frequency) 0.08 ; 0.88 C98A. R98/R16 (proportion controllership expenses) -0.64*; -0.38* C988 R98/R4 (controllership expense emphasis) Accounting Job Structure complexity 0.57 C99A R99/R94 (elaboration of jOb titles) —0.47 C99B R99/R19 (proportion controllership jOb titles) -0.89 C100 (R100/R94)/(l/R99) (employee job concen- tration) -0.83 C101 (R101/[R94-R100])/(l/[R99-l]) (employee jOb concentration) Accounting Geographical Dispersion -0.86 R102 # CONTR LOCATIONS -0.84 C102A R102/R94 (differentiation of controllership locations) -0.72 C1028 R102/R22 (proportion controllership locations) 0.28 C105 (RIOS/R23)/(R94/Rl4) (geographical employee concentration) -0.71 R107 CONTR LOCATION DISTANCE -0.91 C107 R107/R25 (proportion contr location distance) 298 Table 20 (Cont'd.) Loading ggfifinflzzé Description and Calculation Accounting Unit Differentiation-Horizontal 0.46 R108 LOWEST CONTR UNITS (number) 0.75 C108A R108/R94 (elaboration of lowest contr units) 0.93 C1088 R108/R26 (elaboration of lowest contr units) Accounting Authority Levels -0.86 C109 R109/R51A (depth of lowest contr units) -0.90 C110 R110/R51A (depth of lowest contr emps) -0.81 ClllA (R110-Rlll)/R51A (vertical width contr function) -0.87 ClllB (R109-Rlll)/R51A (vertical width contr function) Accounting Report Differentiation 0.10* C112 R112A+R1128+R112C (report diversity) -0.84 R113 PROFIT CENTERS —0.58 Cll3A R113/R96 (proportion profit centers) -0.86 C1138 R113/R44 (profit center elaboration) —0.88 R114 PRODUCT CENTERS -0.92 C114A R114/R96 (proportion product centers) -0.95 C1148 R114/R9 (product center elaboration) Decentralization of'Accounts 0.91 C115 R115/R51A (depth lowest cost reports) 0.18 C116 Rll6/R51A (depth lowest profit reports) 0.91 C978 PROPORTION REPORT LEVELS Sophistication of’Accounting Techniques 0.81 C117 (R117A+R117B+R117C)/3 (standard cost usage) 0.94 C118 WEIGHTED FIXED-VARIABLE 0.86 C119 WEIGHTED BREAKEVEN (use of cost-volume-profit analysis) 0.88 R120 COST-VOLUME-PROFIT (sophistication of) -0.50* R121 BUDGET ITEMS INCLUDED 0.07 R122 CAPITAL PROJECT INFORMATION 299 Table 20 (Cont'd.) . Measure- . . . Loading ment Name Description and Calculation Accounting Unit Differentiation-Vertical 0.99 C125A+ R125/R5LA (reliance on central controller's dept) 0.99 C1258+ (R125-R111)/R51A (reliance on central controller's dept) Accounting Mechanization -0.92 R128 COMPUTERIZED FINANCIAL ACTIVITIES (number of) -0.88 R129 MECHANIZED REPORTS (proportion) 0.88 Cl30+ R130/R98 (capital-labor mix) Accounting Personnel Quality-Education Accounting Personnel Quality-General 0.86 ; 0.32 C132 (R132-R53)/R53 (highest educ compared to div heads) 0.90 ; -0.20* C133 (R133/R94)/(R54/R14) (BA's compared to overall company) -0.20*; -0.56* R134 WEEKS TRAINING-CONTR 0.02 ; .25 C135 (R135-R56)/R56 (low contr seniority compared to dir lab) -0.3l*; 0.72 C136 (R136-R57)/R57 (high contr seniority compared to div heads) 0.19 ; 0.45 C1378 (Cl37A-C58)/C58 (low contr compens compared to dir lab) 0.26 ; 0.23 Cl38LA (R130/R94)/(R42/Rl4) (aver contr sal compared with company) 0.06 ; 0.50 Cl38IB (R137B/Cl37A)-(R58A/C58) (low benefits comp with dir lab) 0 300 Table 21 Zero-Order Pearson Correlations of Accounting Components on Each Other Acctg. Accounting Componentb Compo- nentb A B c D E F G H I J K L M A . . . . . . . 0 7 . . . . B . 0 5 . . . . . . . . . . C . . . . . 0 4 . . . . -0 6 . D . . . 0 7 . . . . . . . . E . . . . . . . . —0.6 . . . p . . . . . . . . . . . . G . . . . . . . . . . . . H . . . . . . . . . . -0 4 . I . . . . . . . . . . . . J . . . . . . . . . . . . K . . . . . . . . . . . . L . . . . . . . . . . . . M . . . . . . . . . . . . aCoefficient signs are adjusted for the direction of components. Coefficients are significant at the 0.10 level. bAccounting components: (A) Size-Information Output; (B) Size- Resource Input; (C) Job Structure Complexity; (D) Geographical Disper- sion; (E) Unit Differentiation4Vertical; (F) Unit Differentiation- Horizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decentralization of Accounts; (J) Sophistication of Techniques; (K) Mechanization; (L) Personnel Quality-Education; (M) Personnel Quality- General. 301 Table 22 Zero-Order Pearson Correlations of Accounting Components with Other Componentsa Org. Accounting Componentb Compo- nentC A B C D E F G H I J K L M Process 0.8 O O O O O O O 0.5 O O O O O O . —0.S . O 0 Overall Structural Complexity MNH o 0 U1 0 4 ° ° ° ° 0.4 ° 0.5 ° ° -0.4 ° ° ° 5 0.6 . . . . . . . . . . . . 6 . . . 0.5 . . . . . . . -0.4 . 7 ° - 0.5 0.6 0.5 - 0.4 - - o 0.4 - - 8 ° ° ° -0.4 ° -0.4 ° ° ° ° -0.5 ° ° 9 . . . . . . . . . -0.5 . . . 10 ° ° 0.5 ° 0.4 - - ° - - 0.5 ° ° control system 11 - - 0 0.6 0.5 ° 0.6 ° 0.5 -0.4 - o o 12 . 0.5 . . . . . . . . . -0_4 . 13 . . . . . . . . . . . . . 14 . . . . . . . . . . . . . 15 0.5 -0.4 -O.4 - o o -0.4 0.4 . . . . . l6 . . . . . . . 0.4 -0.5 . . . . 17 . . . . . . . . . . . . . 18 ° -0.5 ° -0.6 -0.6 ° ° ° ° ° ° ° ° 19 0,5 . . . . . . . 0.6 . . . . aCoefficient signs are adjusted for the direction of components. bAccounting components: (A) Size-Infbrmation Output; (B) Size- Resource Input; (C) Job Structure Complexity; (D) Geographical Disper- sion; (E) Unit Differentiation-Vertical; (F) Unit Differentiation- Horizontal; (G) Authority Levels; (H) Report Differentiation; (I) Decentralization of Accounts; (J) Sophistication of Techniques; (K) Mechanization; (L) Personnel Quality-Education; GM) Personnel Quality- General. COrganizational components: (1) Process Sophistication; (2) Process-Output Diversity; (3) Materials Input Diversity; (4) Company Size; (5) Company Job Structure Complexity; (6) Company Geographical Dispersion; (7) Company Divisional Differentiation; (8) Company Divi- sional Specialization; (9) Company Mechanization-General; (10) Company Mechanization-Computers; (11) Company Direct Supervision; (12) Company Staff Support; (13) Company Authority Levels; (14) Company Personnel Quality-High Level; (15) Company Personnel Quality-Low Level; (16) Com- pany Centralization-Investment; (17) Company Centralization-Purchasing; (18) Company Standardization-Jobs; (19) Company Standardization-General. SELECTED BIBLIOGRAPHY SELECTED BIBLIOGRAPHY Blau, Peter M., and Richard A. Schoenherr. The Structure of’Organiza- tions. New York: Basic Books, Inc., 1971. Caplan, Edwin H. "Behavioral Assumptions of Management Accounting," The Accounting Review, XLI (July, 1966), 496-509. ”Management Accounting and the Behavioral Sciences," Manage- ment Accounting, L (June, 1969), 41-45. . Management Accounting and Behavioral Science. Reading, Mass.: Addison-wesley Publishing Company, 1971. Cornfield, Jerome, and John W. Tukey. "Average values of Mean Squares in Factorials," The Annals of’Mathematical Statistics, XXVII (1956), 912-13. The Directory offlMichigan Manufacturers. Detroit: Manufacturer PUblish- ing Co., 1971. Draper, N. R., and H. Smith. Applied Regression Analysis. New York: John Wiley 8 Sons, 1966. Golembiewski, Rebert T. "Organization Structure and the New Accountancy: One Avenue of Revolution," The Quarterly Review of’Eeonomics and Business, III (Summer, 1963), 29-40. "Accountancy as a Function of Organization Theory," The Ac- counting Review, XXXIX (April, 1964), 333-41. Green, Paul E., and Donald S. Tull. Research for Marketing Decisions. Englewood Cliffs, N.J.: Prentice-Hall, 1970. Horngren, Charles T. Cost Accounting, A Managerial Emphasis. 3d ed. Englewood Cliffs, N.J.: Prentice-Hall, 1972. Nie, Norman E., C. Hadlai Hull, Jean G. Jenkins, Karin Steinbrenner, and Dale H. Bent. Statistical Package for the social Sciences. 2d ed. San Francisco: MbGrawaHill, 1975. Price, James L. Handbook of Organizational Measurement. Lexington, Mass.: D. C. Heath, 1972. 302 303 Pugh, D. S., D. J. Hickson, C. R" Hinings, K; M; Macdonald, C. TUrner, and T. Lupton. "A Conceptual Scheme for Organizational Analy- sis," Administrative Science Quarterly, VIII (December, 1963), 289-315. Pugh, D. S., D. J. Hickson, C. R. Hinings, and C. Turner. "Dimensions of Organization Structure," Administrative Science Quarterly, XIII (June, 1968), 65-105. "The Context of Organization Structures," Administrative Sci- ence Quarterly, XIV (March, 1969), 91-114. Simon, Herbert A. , George Kozmetsky, Harold Guetzkow, and Gordon Tyndall. centralization vs. Decentralization in Organizing the controller's Department. New York: Controllership Foundation, Inc., 1954. "7 1H0 H9 5'1