AN EXPERIMENTAL STUDY OF THE EFFECTS OF . : QUANTITATIVE MANAGEMENT INFORMATION ON THE ' GENERATION AND SELECTION OF ALTERNATIVE ACTIONS IN INDIVIDUAL DECISION MAKING UNDER- * ‘ ' UNCERTAINTV ‘ ._ i . Dissertation for "the Degree Of Ph, D. MICHIGAN STATE UNIVERSITY; ’ CHARLES ALEXANDER DAVIS 1975 IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 3 1293 103915 ABSTRACT AN EXPERIMENTAL STUDY OF THE EFFECTS OF QUANTITATIVE MANAGEMENT INFORMATION ON THE GENERATION AND SELECTION OF ALTERNATIVE ACTIONS IN INDIVIDUAL DECISION MAKING UNDER UNCERTAINTY By Charles Alexander Davis Administrators of colleges and universities are experiencing increased pressure from legislators, trustees, taxpayers, and faculty to improve the management of the resources of higher education institutions. University administrators have responded by attempt- ing to adapt computer-based tools and techniques of management tech- nology, based on industrial models, to the operation of their insti- tutions. All of these computer-based tools and techniques and their associated data bases were characterized as Management Information Systems (M15) in this study. Early experience with M15 in higher education institutions revealed considerable difficulty integrating new management technology into the established management structure. Available literature has made valuable contributions to explaining the problem of implementa- tion, but reveals little evidence of basic research aimed at under- standing the effect of new management systems and technology on the individual manager. Charles Alexander Davis The present research was intended to explore the effects of quantitative management information on the individual manager in the performance of one of his most crucial functions--decision making. Because of the lack of research results directly related to indi- vidual decision making in the current literature, this study was considered exploratory. The problem addressed by this research was the effect of quan- titative management information about the state of an uncertain envi- ronment on the generation and selection of alternative actions in the individual decision-making process in that environment. An uncertain environment is one in which all possible actions, the outcomes of possible actions, and the possibilities of such outcomes are not known to the decision maker. The background for the present research was established by reviewing literature related to MIS applications in higher education. A brief review of the history of decision making was presented and a model, based on decision-making theory, was developed to establish the theoretical framework for the study. Five basic research questions were formulated relative to the effects of quantitative management information on the individual decision maker in an uncertain environment. An experiment was designed to provide the data from which answers to the research questions could be derived. Subjects used in the experiment were chosen from seniors and graduate students in the College of Business at Western Michigan Uni- versity. Subjects were randomly assigned to treatment groups to Charles Alexander Davis minimize the effect of confounding uncontrolled variables with the information treatment. Orthogonal planned comparisons of the mean number of alternatives generated and selected and nonparametric analy- sis of the distribution of alternatives generated and selected were used to extract information from the raw data. The methodology used in the study was experimentation. Internal validity was controlled by choice of sample size, confidence levels, and random assignment of subjects to treatment groups. Gen- eralization of the results of a single experiment, a single decision problem, and a given population to all decision situations is not the claim of the researcher. It is hoped that a contribution has been made to understanding individual decision making in an uncertain envi- ronment. Findings of the study were: l. Quantitative management information had no effect on the number of alternatives generated by individual decision makers in an uncertain environment. 2. There was no evidence that quantitative management infor- mation had any effect on the number of alternative actions selected by individual decision makers in an uncertain environment. 3. The number of alternatives generated showed high posi- tive correlation with the number of alternatives selected by the same individual except when the information treatment was quantitative management information. Charles Alexander Davis 4. The distribution of alternatives selected by individual decision makers in an uncertain environment was affected by the combination of quantitative and nonquantitative management informa- tion when compared to the use of quantitative management information alone in an uncertain environment. AN EXPERIMENTAL STUDY OF THE EFFECTS OF QUANTITATIVE MANAGEMENT INFORMATION ON THE GENERATION AND SELECTION OF ALTERNATIVE ACTIONS IN INDIVIDUAL DECISION MAKING UNDER UNCERTAINTY By Charles Alexander Davis A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Administration and Higher Education 1975 ACKNOWLEDGMENTS I express my sincere appreciation to Professor Russell Kleis, my committee chairman, for the inspiration and guidance provided during the early formation of my doctoral plans. His commitment to continuing education had a great influence on my decision to pursue a career in the profession. Many thanks are due to Dr. Glenn Keeney, my dissertation director and friend. I am grateful for the confidence that Dr. Richard Featherstone had in me and for his willingness to work with me in writing my pro- posal and conducting the experiment. I thank Dr. James Nelson for his contributions as a member of my committee and for providing timely advice when it was needed. Finally, I thank my guidance committee for the cooperative spirit and professionalism that they maintained throughout this entire experience. Without them, none of this would have been pos- sible. ii TABLE OF CONTENTS Page LIST OF TABLES ......................... v LIST OF FIGURES ........................ vii Chapter I. INTRODUCTION ...................... 1 Introductory Statement ................ l Functions of the Administrator in Higher Education . . 7 Statement of the Problem ............... l3 Purpose of the Study ................. l4 Methodology ..................... 15 Definition of Terms ................. l8 Limitations of the Study ............... 23 Importance of the Study ............... 23 Overview of the Dissertation ............. 24 II. SELECTED REVIEW OF LITERATURE ............. 26 Current Management Information Systems ........ 26 Brief History of Decision Theory ........... 4O Theories of Individual Decision Making ........ 5l Related Studies in Individual Decision Making Under Uncertainty ................. 58 III. THE EXPERIMENTAL DESIGN ................ 66 Theoretical Model of Individual Decision-Making Activity ...................... 67 Research Questions .................. 75 Experimental Procedure ................ 77 Techniques for Analysis of the Data ......... 78 Statistical Computations ............... 84 Possible Outcomes of the Experiment ......... 92 Instrument for Collection of the Data ........ 95 Chapter IV. V. ANALYSIS OF THE RESULTS ................ Objective of the Research .............. Raw Data ....................... Planned Comparisons Analysis ............. RIDIT Analysis of the Distributions of Alternatives Characteristics of the Sampled Population ...... Correlation Matrix .................. Summary ....................... THE PROBLEM, FINDINGS, CONCLUSIONS, AND RECOMMENDATIONS FOR FUTURE RESEARCH ......... The Problem ..................... Findings ....................... Conclusions ..................... Recommendations for Future Research ......... Reflections of the Researcher ............ APPENDICES ........................... A. B. INSTRUMENT FOR THE COLLECTION OF THE DATA ....... CALCULATIONS . . . . . . . . . . . . . ......... SELECTED BIBLIOGRAPHY ..................... iv 156 Table 01th \mem 10. 11. 12. 13. 14. 15. 16. 17. 18. LIST OF TABLES Student Input Characteristics ......... Output Measurement .............. What You Put Into CAMPUS VII ......... What You Get Out of CAMPUS VII ........ Orthogonal Weighting Coefficients for Comparison of Means .................. Calculation of RIDIT Values .......... Set of Alternative Actions Generated by Subjects Raw Data for Alternatives Generated ...... Raw Data for Alternatives Selected ...... Number of Alternatives Generated and Selected by Subjects .................. Weighting Coefficients for Planned Comparisons Calculated Values of VAR(D) .......... Test Statistic for Planned Comparisons . . . . Confidence Intervals for Planned Comparisons . Frequency Distribution of Alternatives Generated Selected by Treatment Group 1 ........ Frequency Distribution of Alternatives Generated Selected by Treatment Group 2 ........ Frequency Distribution of Alternatives Generated Selected by Treatment Group 3 ...... '. . Frequency Distribution of Alternatives Generated Selected by Treatment Group 4 ........ and and and and Page 30 3l 38 39 83 88 102 103 112 113 114 115 Table l9. 20. 21. 22. Page RIDIT Calculations for Alternatives Generated by Treatment Group l .................. ll6 Summary of Characteristics of the Sampled Population . l2l Pearson Correlation Coefficients by Treatment Groups . 124 Calculation of Rave for Treatment Group 2 ....... l61 vi Figure \JO‘U'I-hw SOC) 10. 11. 12. 13. 14. 15. 16. 17. 18. LIST OF FIGURES The University as a System .............. Typical Academic and Nonacademic Production Sectors . . . Sample Use of NCHEMS Instruments ........... NCHEMS CSM Data Format ................ NCHEMS CSM Data Format ................ Typical NCHEMS ICLM Data Format_ ............ Typical PLANTRAN 11 Input Data Sheet ......... Typical PLANTRAN II Output Data Sheet ......... Typical SEARCH Data Format .............. Two Sets of Not Mutually Exclusive Events ....... Decision Activity Matrix . . .' ............ Bayesian Model for Decision Making .......... Example of the Reduction of Original Alternatives . . . . Graph of Average Ridits Using Treatment Group l as a Reference ................... Possible Outcomes of the Experiment .......... Combinations of Information Given to Treatment Groups . . Rave vs. Treatment Groups for Alternatives Generated R vs. Treatment Groups for Alternatives Selected . . . ave vii Page 28 32 34 35 36 37 4l 42 43 54 68 7O 81 91 93 97 117 118 CHAPTER I INTRODUCTION Introductory Statement Higher education institutions will be difficult to manage even with the availability of the best planning and management systems information. Without such information, good management at a complex institution may be virtually impossible.1 The foregoing statement by Huff and Manning reflects the opin- ion of a new breed of management scientists whose mission is the appli- cation of the tools and techniques of management technology to higher education. The central element in most proposed applications is the electronic data processing capability of the computer. No standard- ized terminology has yet emerged, but the more highly recognized approaches to scientific management techniques are identified by the terms Management by Objectives, Program Planning and Budgeting Sys- tems, Management Information Systems, Computer-Based Planning Models, and Educational Simulation Systems. Despite the overlapping descriptions used to identify current approaches, all management tools and techniques are linked by one common denominator: a supportive, quantitative data base. This data base is most often associated with MIS, which Nelson defined as 1R. A. Huff and C. W. Manning, Higher Education Planning and Management Systems (Boulder, Colorado: National Center for Higher Education Management Systems, l972), p. 17. . that configuration of men, machines and methods which supports management in the collection, storage, processing, and transmission of information for operation, control, eval- uation and planning of a university. Robinson defined M15 in terms of its objectives: "The major objective of a management information system, therefore, is to provide useful, relevant information to management in the form and at the time when it will be most useful."3 The term M15 is used in this study to character- ize all of the tools and techniques of management technology and their associated data bases. While management scientists are adapting their technology to the management problems of higher education, there is increased pressure from public agencies to make higher education institutions more manage- able. Farmer described the financial pressure: For many years higher education has presented the bill for higher education to the public for support, and it usually was paid. Now, however, bond authorizations are frequently defeated at the polls, and state governments are drastically reducing per student funds. Although higher education used only 2.2 percent of the gross national product in 1965-67, the expenditures totaled $15.2 billion. By 1980 higher education will be consuming 2.5% of the GNP, some $32.5 billion. The public now has a large number of social programs--hunger, housing, medical care, transportation and pollution--competing for public funds. Educators are being asked to specifically describe their objectives, measure their performance, and determine costs.4 2C. A. Nelson, "Management Planning in Higher Education-- Concepts, Terminology and Techniques," Management Controls, January 1971. p. 5. 3D. D. Robinson, "Some Observations on the New Management for College and University," Management Controls, October 1970, p. 220. 4J. Farmer, Why Planning, Programming, Budgeting Systems for Higher Education? (Boulder, Colorado: Western Interstate Commission for Higher Education, 1970). Robinson wrote of college administrators' growing awareness of the need for sound management: There is, above all else, a growing realization that good man- agement is important in an institution of higher learning. This may sound self-evident, but the fact is that up until recently most academic administrators believed (or certainly behaved as if they believed) that colleges were not subject to the same kinds of management rules as are other organizations; that through some sort of marvelous beneficence, they were exempt from all or most of the consequences of bad management. This revelation, this insight, this slow coming of age has finally made possible the rational consideration of the need to fashion tools that will assist in meeting the management problems of colleges and uni- versities. Rourke and Brooks pointed out the political utility of a computer-based MIS: Finally, beyond considerations of efficiency and internal con- trol, the computer plays an important role as a showpiece to impress the outside world with the modernity of university admin- istration. As the struggle for legislative appropriations grows intense, most universities must draw upon any and all available strategies to insure economic survival. One such strategy is to give the public and the state legislature every possible reason to believe ghat the university is being operated with maximum efficiency. The concentration of pressures from public agencies, faculty, students, and even some university administrators on the administrative structure of colleges and universities to "do something" to improve the management of their resources has generated the awareness of an acute need to change the traditional methods by which universities have been operated in the past. This acute need, coincident with the desire of management scientists to apply management technology to higher 51bid., p. 217. 6F. E. Rourke and C. E. Brooks, "Computers and University Administration," Administrative Science Quarterly, December 1966, p. 600. education administration, has led to the rapid and sometimes unfortu- nate introduction of M15 on college campuses. Early experience with MIS applications in complex organiza- tions such as institutions of higher learning has led some observers to point out problems of such systems. Ackoff criticized the assump- tions on which MIS designs are usually based: Contrary to the impression produced by the growing litera- ture, few computerized management information systems have been put into operation. Of those I've seen that have been imple- mented, most have not matched expectations and some have been outright failures. I believe that these near-and-far misses could have been avoided if certain false (and usually implicit) assumgtions on which many systems have been erected had not been ma e. Thompson raised some thorny questions about the impact of such systems in the university resource allocation process: After the initial resource allocation pattern has been fixed by legislative action, it is at this point that nonelected officials take over the responsibility for successive reallo- cations down to the smallest organizational components affected. But will PPBS data necessarily affect the long established practices of organizational bureaucracy as they apply to this type of budgeting? Will the process ever be free from successful attempts to once again introduce inter- nal and external political considerations capable of affect- ing the eventual outcome? And will the eventual reaction of the agencies budgeted to the final pattern of resource allo- cation be any different than the present if they perceive the decision—making process as being basically unchanged, but merely dressed up with some new computerized budgeting gimmick? Argyris discussed some of the emotional problems that arose when a MIS was introduced into the management of a complex organization 7IR. A. Ackoff, "Management Misinformation Systems," Management Science 14 (December l967): 147. I80. L. Thompson, "PPBS: The Need for Experience," Journal of Higher Education 42 (January 1972): 686. 9 Observation of the technologists' by a team of M18 technologists. attempts to increase the rationality of management behavior revealed an intensification of the emotional responses of the management per- Isonnel. Argyris reported that the technologists deviated from the rational philosophy of their professional training under the stress of the corporate environment. The technologists tended toward the same emotional behavior exhibited by the managers of the complex organization. Ackoff, Thompson, and Argyris all addressed a broad general problem: the effect of quantitative data-based MIS on the established practices of managing complex organizations. Higher education institutions, among the most complex of modern organizations, are not exempt from the requirement to effec- tively integrate new management technology into their established patterns of administration. There has been considerable deliberation of the problem of implementing new management systems. Churchman and Schainblatt sum- marized the various opinions about how the efforts of researchers in management science should be implemented into traditional management 10 structures. They identified four alternative positions with respect to the relationship between the management scientist and the manager: 9C. Argyris, "Management Information Systems: The Challenge to Rationality and Emotionality," Management Science 17 (February 1971): 275-291. 10C. W. Churchman and A. N. Schainblatt, "The Researcher and the Manager: A Dialectic of Implementation," Management Science 11 (February 1965): 69-87. 1. Separate function position--management and management research are viewed as separate functions. Implementation consists of specifying the physical changes that must take place in an organi- zation in order for it to accommodate the optimal mathematical solu- tion. 2. Communication position--emphasizes the need for creating better lines of communication between manager and the management scientist. The communication is direct and independent of the per- sonality of the manager. The manager's understanding of the scientist is viewed as critical. 3. Persuasion position--emphasis is placed on the scientist's understanding enough about the manager to persuade the manager to accept recommended changes. 4. Mutual understanding position--embraces the position that management and management science cannot be separated. If science is to become a method of managing, then management must become the method of science. Argyris used the strategy of placing a member of line manage- ment on the MIS team to act as a liaison in the implementation process, 1] He implied that imple- but reported less than desirable results. mentation strategies based on rational solutions such as education and structural change alone would not work, and suggested emphasis on the utilization of behavioral science technology to increase inter- personal competence. Mitchell, Farmer, Nowbray, and Levine analyzed the implementation of various management science tools in higher ”Ibid. education institutions.12 Halter and Dean discussed the relationship between the analyst and the decision maker in complex agricultural situations.13 Although it has made valuable contributions to explaining the implementation problem, the available literature reveals little evidence of basic research aimed at understanding the effects of new management systems and technology on the individual manager, or how individual managers use such systems and technology. ' Functions of the Administrator in Higher Education Although various writers have described the functions of managers and administrators differently, few disagree with Corson's statement that: The administration of any enterprise involves the making and subsequently the execution of a succession of decisions. In a manufacturing concern, these decisions involve the hiring of workers, the purchasing of raw materials, the determination of methods of production and volume of output, the setting of prices, and a myriad of related decisions of greater and lesser significance to the accomplishments of the enterprise. In a government bureau, the decisions involve the proposal of legis- lation, the hiring and promotion of civil servants, the con- tracting with industry, the adjudication of cases, the formu- lation of budgets and work programs and the determination of what shall be said to the public in speeches and reports. In a university similar decisions are made and executed. Faculty members, administrators, coaches, secretaries, and 12E. E. Mitchell, "PPBS: Panacea or Pestilence," AEDS Monitor. February l970, pp. 4-13; J. Farmer, An Approach to Plannigg and Man- agement Systems Implementation (Los Angeles, California: California State Colleges Publications, 1971); G. Mowbray and J. B. Levine, "The Development and Implementation of CAMPUS: A Computer Based Planning and Budgeting System for Universities and Colleges," Educa- tional Technology, March 1971. 13A. Halter and G. Dean, Decisions Under Uncertainty (Chicago: Southwestern Publishing Co., l971),fip. 139. various other persons are hired and promoted--or not promoted. A curriculum is formulated and reformulated; courses are added and dropped.14 A distinguishing characteristic of complex enterprises or organizations is the process by which decisions are made and imple- mented. A manager or administrator seldom makes decisions without the involvement of both subordinates and superiors. Simon argued that: It should be perfectly apparent that almost no decision made in an organization is the task of a single individual. Even though the final responsibility for taking a particular action rests with some definite person, we shall always find, in studying the manner in which the decision was reached, that its various components can be traced through the formal and nonformal channels of communication to many individuals who have participated in forming its premises. 5 A major concern of administrators or managers in a complex organization is the maintenance and structuring of the decision-making process. One could argue further that a major responsibility of the administrator in a complex organization is to fulfill his role in the decision-making process of the organization. The definition of that role depends on the particular organization. Most research on complex organizations and the decision processes has been centered on govern- ment, business, and industrial firms. Administrative theorists and administrators are concerned that the results of such studies are not wholly applicable to colleges and universities. Millett, a leading spokesman for this viewpoint, declared that "Ideas drawn from business 14J. Corson, Governance of Colleges and Universities (New York: McGraw-Hill, 1960), p. 118. 15H. A. Simon, Administrative Theory (New York: Macmillan Co., 1961), p. 221. and public administration have only a very limited applicability to colleges and universities."16 Litchfield, a spokesman for the opposing view, believed that "Administration and the administrative process occur in substan- tially the same generalized form in industrial, commercial, civil, educational, military and hospital organization."17 The similarities and differences between the management of universities and government, business, and industry was pointed out by Baldridge, using two models of the university.I‘8 If the university is viewed as a bureaucracy, as described by Weber, a great similarity with business and industry can be claimed.19 Among the shared features are: l. The university is a complex organization chartered by the state. 2. The university has a formal hierarchy, with offices and a set of bylaws that specify the relationships between these offices. 3. There are formal channels of communication that must be respected. ‘5J. D. Millett, The Academic Community (New York: McGraw- Hill, 1962), p. 4. 17E. H. Litchfield, "Notes on a General Theory of Administra- tion," Administrative Science Quarterly 1 (January 1956): 28. 18J. V. Baldridge, Academic Governance (Berkeley: McCutchan Publishing Co., 1971). 19M. Weber, The Theory of Social and Economic Organizations, trans. T. Parsons and A. Henderson (New York: The Free Press, 1947). 10 There are definite bureaucratic authority relations, with some officials exercising authority over others. Formal policies and rules govern much of the institution's work. Decision processes are often highly bureaucratic, especially when rather routine types of decisions are at stake. If the university is viewed as a "collegium" or "community of scholars," some distinct differences between the university and business and industry can be observed. Prominent features are: 1. 3. Participation of all members of the academic community-- especially faculty--in the management of the university. Emphasis on authority based on "technical competence" rather than the ”official competence" resulting from one's office holding in the bureaucratic hierarchy. A ut0pian prescription for humanism in the educational process, unlike the impersonalism of the bureaucracy. Baldridge pointed out that an analysis of the decision process implied by these two views of the university reveals weaknesses in both: The bureaucratic model does not deal adequately with nonformal types of power and influence. The bureaucratic model explains formal structure but does not deal adequately with the processes that give dynamism to the structure. 11 3. The collegial model does not deal adequately with the problem of conflict. The argument that major decisions are made primarily by consensus ignores power plays, conflict, and the politics of a large university. It appears decision theorists should initiate research efforts aimed at understanding the university as a complex organization, different than business and industrial firms. Baldridge conducted a research project to study the decision- making process in a large American university.20 Results indicated that most members of the university community claimed some partici- pation in the decision-making process at some level. Many people were involved in decision making at the department level, but only a small number participated in college or all-university decisions. Further analysis indicated that decision-making influence was frag- mented, with different groups being strong in different spheres of influence and no single group dominating everything. The groups defined in the study were trustees, central administration, deans, college faculty, department faculty, and individual faculty members. One finding of the study was that trustees had very little influence on decision making in curriculum matters, whereas deans had great influence in decision making where faculty promotion was concerned. In another study of 115 colleges and universities in the United States, Baldridge reported a strong association between increasing institution size and the following: 20J. V. Baldridge, Power and Conflict in the University (New York: John Wiley and Sons, 1971). 12 l.- a center Specialized in mediating those external rela- tions that are crucial to the maintenance and development of institutional legitimacy and material support. 2. a powerful faculty senate and subject matter departments with more autonomy over matters of particular concern to them.21 The results of these research efforts imply that the large university is a loose federation of administrative units with wide participation in decision making at the lower levels and a centralized officialdom at the upper level involving very few persons in the decision process. It appears that the administrator in an institution of higher education is more likely to be in the role of mediation or consensus formulation than is his counterpart in business and industry. As Baldridge so wisely observed, however, the "collegial consensus" is often nothing more than the ascendancy of one group over another. In such situations, even at the departmental level, the administrator often decides which group will prevail. In large universities, at higher levels, the decision-making process involves few individuals and, to a large extent, administrative officials participate in the decision making with recommendations from committees or councils. If one is to make a research contribution to the effective implemen- tation of quantitative management information systems in complex organizations such as higher education institutions, such research 2IBaldridge, Academic Governance, p. 58. 13 might best study the effect of such systems on the decision-making process. One approach would be to investigate the effect of M15 on collegial consensus at the department level in the university. Another approach would be to investigate the effect of M15 on decision making by administrators in higher education functioning in the absence of consensus. The administrative role in such situations is more nearly like that of an administrator in business or industrial organi- zations. The present study takes the latter approach. Statement of the Problem The problem addressed by the present research was derived from the previously mentioned need to integrate quantitative management information systems and techniques into the traditional management patterns of complex organizations such as higher education institu- tions. Investigation of the problem focuses on the decision-making activities of individuals in an uncertain environment. The problem investigated was the effect of quantitative man- agement information about the state of an uncertain environment on the generation and selection of alternative actions in the individual decision-making process in that environment. The theoretical framework for the investigation is embodied in the Bayesian model derived from decision-making theory, as dis- cussed in Chapter II of this dissertation. Soelberg's research, based on an expanded model of Simon's characterization of the decision process, indicated that early activity focused on the search for alternative actions and the reduction to two or more acceptable 14 alternatives before termination of the search.22 In this activity the decision maker examines the environment to obtain information about present and possible states of nature, and tests hypotheses about the probability of likely states of nature in the environment as a result of the decision to be made. Purpose of the Study The purpose of the study was to seek answers to five basic questions about individual decision making under uncertainty and the effects of quantitative management information on the generation and selection of alternatives. The questions were: 1. Is there any difference in the number of alternative actions generated by the individual decision maker in an uncertain environment as a result of quantitative management information as compared to nonquantitative management information or no management information? 2. Is there any difference in the number of alternative actions selected by the individual decision maker in an uncertain environment as a result of quantative management information as com- pared to nonquantitative management information or no management information? 3. Is there any difference in the number of alternative actions generated by an individual decision maker under uncertain 22F. Soelberg, "Unprogrammed Decision Making," in Studies in Managerial Process and Organization Behavior, ed. J. H. Turner, A. C. Filley and R. J. House (Glenview, 111.: Scott, Foresman and Co., 1972). 15 conditions as a result of the combination of quantitative and non- quantitative management information when compared to the use of either type separately? 4. Is there any difference in the number of alternative actions selected by an individual decision maker in an uncertain envi- ronment as a result of the combination of quantitative and nonquanti- tative management information as compared to either type when used separately? 5. Is there any difference in the distribution of alterna- tive actions selected by individual decision makers in an uncertain environment as a result of quantitative management information when compared to nonquantitative management information or no management information, when the distribution variate is a randomly ordered nomi- nal set of alternatives representative of a referenced distribution? These questions were addressed to a small part of the problem of the effects of management information on individual decision making: the generation and selection of alternatives. It is hoped that further- ing the understanding about this part of individual decision-making activity will contribute to the knowledge base from which future researchers may draw some guidance. Methodology So little research has been published on the subject of indi- vidual decision making under uncertainty that this study must be con- sidered exploratory. For this reason, the research hypotheses were stated in the form of questions rather than major null hypotheses. 16 The question framework for stating research hypotheses allows the researcher more flexibility to explore whatever results the data might reveal. The methodology for the study was experimentation as opposed to field observation or a case study. The use of the experimental method in research has long been accepted in the physical and biologi- cal sciences, but still evokes debate among researchers in the social and behavioral disciplines. Decision making under uncertainty is a complex process involving many variables. The researcher must exert some control over variables believed to be related to the research question if valid knowledge is to be derived from his work. The experi- mental method allows the experimenter to control some variables and minimize the effect of others through randomization. This enhances the internal validity of the experiment. Another advantage of the experimental method is that the researcher must state explicitly the conditions under which the results were obtained. The results can then be generalized to other situations in which similar conditions are observed. This enhances the external validity of the results of the experiment. Another issue that had to be resolved was the population from which subjects would be drawn. Cummings and Harnett cited several studies that supported the reasoning that students can be used as subjects in managerial decision-making studies and that their responses 23 will be essentially the same as those of active managers. In this 23L. L. Cummings and D. L. Harnett, "Managerial Problems and the Experimental Method," Business Horizons, April 1968, p. 41. 17 study there was the added requirement that the subject population be familiar with the format for presenting quantitative management information, such as charts, graphs, tables, percentages, and propor- tions. The population used in this study was college students in a single university; they were chosen from the College of Business, before instrumentation of the study. The rationale for selection of this population was that the students' backgrounds indicated the aforementioned familiarity with the style and format of quantitative management information. It was expected such familiarity would mini- mize response variance resulting from misinterpretation of the infor- mation, thus increasing the precision of the experiment. A large resource group was identified, which represented a cross-section of the population. Subjects were randomly selected from the larger resource group. A total of 80 subjects was used in the experiment. Twenty students were assigned to each treatment group. The criteria for participation of students in the study were: 1. Subjects must voluntarily agree to participate in the experiment. 2. Participants must possess sufficient knowledge of the problem situation of the study to be considered usable subjects. 3. Subjects musthe accessible to the researcher to facilitate the collection of data and completion of the study within a reasonable time, and thus minimize the effects of history and external influence on the responses. 18 Definition of Terms Management information: Management information is that infor- mation necessary to support the management of an institution in (a) planning what should be done, (b) operationalizing plans to get things done, (c) controlling operations to determine whether plans are being operationalized, and (d) evaluating whether planned outcomes have been achieved. Management information systems: An earlier reference was made to the lack of standardization in the terminology of management tech- nology and its application to higher education institutions. In this study, M15 is used as a general descriptor of all of the tools and techniques of management technology and their associated supportive data bases. Decision: Most of the literature on decision making in the journals of psychology, economics, and statistical mathematics avoids specific definitions of a "decision." Eilon quoted a definition given by Ofstad: To say that a person has made a decision may mean (1) that he has started a series of behavioral reactions in favor of some- thing, or it may mean (2) that he has made up his mind to do a certain action, which he has no doubts that he ought to do. But perhaps the most common use of the term is this: "To make a decision" means (3) to make a judgment regarding what one ought to do in a certain situation after having deliberated on some alternative courses of action.2 24S. Eilon, "What Is a Decision?" Management Science 16 (December 1966): 172. fitmfig‘mg-s 0.. . , . .c““‘ a. . . x. .. Ill .. r .I I. I. I. . 19 Knezevich stated that "A decision can be defined as a con- scious choice from among a well defined set of often competing alter- natives."25 Although many decision theorists would argue that decisions are often made with considerable doubt and that choices are not always well defined, there would be general agreement that a decision involves at least two alternatives, and that after deliberation a conscious choice is made from the alternatives. For purposes of this study, the following definition was used: Decision--a conscious choice of an alternative after deliberate consideration of at least two competing alternatives. Decision making: The majority of decision theorists tend to focus on the decision-making process. Knezevich said, "Decision making is a sequential process culminating in a single decision or a series of decisions."26 Blankenship and Miles wrote: Decision making may be visualized as a complex process in which an individual or a group of individuals moves through a series of interrelated substeps including (1) the recognition of a problem requiring some response, (ii) the investigation of the problem and its environment in an effort to collect relevant information and to generate solutions, and (iii) the selection of a course of action based on an analysis of the available information and solutions. 255. J. Knezevich, Administration of Public Education (New York: Harper and Row, 1969):’p. 10. 26 Ibid., p. 32. 27L. V. Blankenship and R. E. Miles, "Organizational Structure and Managerial Decision Making," Administrative Science Quarterly 13 (June 1968): 107. 20 These two definitions represent the simplistic and the complex extremes of the concept of the decision-making process. The present study used the definition of decision making proposed by Blankenship and Miles. This preference is further demonstrated by the decision activity model discussed in Chapter III. Quantitative management information: Management information systems have generally been defined to include the tools and techniques of management technology and their supportive data bases. This defi- nition is adequate for purposes of discussion, but for designing the experiment, quantitative management information must be more precisely defined. A search of the literature failed to uncover any definition of these terms, so definitions are formulated here for purposes of this study. Quantitative management information must meet two basic criteria: 1. The variables on which data measurements are made must, in principle, be naturally quantifiable. 2. The format in which data are presented must not alter the information content. Any management information not meeting these two criteria is defined as nonquantitative management information. The first criterion differentiates between those variables that are naturally quantifiable and those on which quantitative scales are artificially imposed for purposes of measurement. Naturally quanti- fiable variables include dollars available for program support, number of student credit hours, number of jobs available, size of faculty, and projected size of the student body. Variables that might be 21 artificially quantified are satisfaction of graduates, unity of faculty, power of the student body, intelligence of students, and quality of programs. The second criterion is intended to distinguish between struc- turing the data format for purposes of presentation to user and alter— ing the information content by interpretation, inference, personal judgment, or axiological perturbation. The format for quantitative management information is generally void of prose and makes use of graphs, charts, statistical analyses, and tables. The second criterion does not prohibit ordinary statistical analyses such as determination of the mean, variance, mode, maximum and minimum values. Qualitative statistical analyses would violate the condition for quantitative management information. Techniques such as rank ordering, categorical grouping, and statistical inference convey information about the values and perceptions of individuals as well as the quantitative data. These two criteria were applied in the selection of management information to be used in the research instrument. Uncertain environment: The theories of decision making are related to the environment in which the decision is made. An uncer- tain environment is one in which all possible actions available to the decision maker are not known, the outcomes of such actions are not completely known, and the probabilities of known outcomes are in doubt. In summary, the following definitions were used in this study: Management information--that information necessary to support the management of an institution in planning what should be done, 22 operationalizing plans to get things done, controlling operations to determine whether plans are being operationalized and evaluating whether planned outcomes have been achieved. Management information systems--all of the tools and tech- niques of management technology and their associated supportive data bases. Decision--a conscious choice of an alternative after deliber- ate consideration of at least two competing alternatives. Decision making--a complex process in which an individual or a group of individuals moves through a series of interrelated substeps including the recognition of a problem requiring some response, the investigation of the problem and its environment in an effort to col- lect relevant information and to generate solutions, and the selection of a course of action based on an analysis of the available informa- tion and solutions. Quantitative management information--management information that meets the following criteria: 1. The variables on which data measurements are made must, in principle, be naturally quantifiable. 2. The format in which data are presented must not alter the information content. Uncertain environment--one in which all possible actions available to the decision maker are not known, the outcomes of possible actions are not completely known, and the probabilities of known out- comes are in doubt. 23 Limitations of the Stugy This study investigated only a portion of the decision-making activity of individuals in an uncertain environment--the generation and selection of alternative actions. It does not deal with the culminating decision activity, a final choice from selected alternatives. This study was experimental; as such, one must recognize the limitations of experimental studies in terms of validity and generaliz- ability of the results. Careful selection of the population and randomi- zation minimized the influence of variables other than the information treatment effects. A further limitation of the study was the use of students rather than university administrators as subjects in the experiment. The ability of such students to artificially assume the role of administrators for experimental purposes has been verified by management scientists but in any given decision situation, generaliza- tion of the results might be limited. Since the study was exploratory, the results are not expected to be the end but rather the means by which future research might be guided. Research questions were investigated. It was not expected that conclusions would come from this study, but that directions would be indicated for future investigation of the complex process of indi- vidual decision making under uncertainty and the effects of quantitative management information on that process. Importance of the Study Two factors have contributed to the increased application of computer-based management technology to the administration of higher 24 education institutions. Those factors are pressure from public agen- cies on higher education to be more accountable for the management of public resources and the eagerness of management scientists to adapt the tools and techniques that have proven successful in industry to higher education management problems. The problem now is one of "implementation"--the introduction of quantitative-data-oriented technology into the highly judgmental decision process of traditional educational administration. We must investigate the effects of such quantitative data on the process of decision making under uncertainty, to discover whether and how the introduction of such data contributes to more effective management. Research should start with the indi- vidual decision maker. Unless we increase our knowledge of decision making and the utility of new technology for improving it, the benefit that can be gained from the successful implementation of computer-based management information systems in colleges and universities will be lost. The present study is an attempt to contribute to that knowledge. Overview of the Dissertation Chapter I provided an introduction to the role of management information systems in higher education, a statement of the problem to be investigated in this study, and the purpose, methodology, and limi- tations of the study. Definitions of selected terms used in the disser- tation were also presented. Chapter II contains a review of literature and related research on management information systems and the decision-making theory on which the study was based. 25 A description of the model that forms the theoretical basis for the study, the design of the experiment, and the methods of analysis of the data are found in Chapter III. The raw data and the analysis of the data are presented in Chapter IV. Chapter V contains a summary of the results of the study, conclusions and recommendations for future investigation, and reflec- tions on the conduct of the study and its results. CHAPTER II SELECTED REVIEW OF LITERATURE The literature review of this study can be categorized in two distinct areas: (1) current quantitative management information systems and (2) decision theory. The review of current quantitative management information systems provides the background necessary to comprehend the kinds of data, formats, structures, and variables administrators in higher education institutions are likely to encounter now and in the near future in their search for information to assist them in decision making. Decision theory forms the theoretical base for the model used in the study. From decision theorists in psychology, statistics, economics, management, and business administration have come reports of research studies related to individual decision making under uncer- tainty. The results of some of those studies are presented in this chapter. Current Management Information Systems Reference was made in Chapter I to the lack of a standardized terminology of management technology and its application to higher education institutions. In this dissertation, M18 is used as a general descriptor of all the tools and techniques of management technology and the associated supportive data bases. This general usage causes some 26 27 confusion when one tries to talk about the state of the art of M15 in higher education, for it becomes difficult to categorize and survey the literature. Rourke and Brooks, in a survey of 436 institutions of post- secondary education to determine use of computerized systems in their administration, established four general areas of heavy usage: (1) student affairs, (2) financial management, (3) physical plant management, and (4) general policy planning.1 Student affairs activi- ties included registration, grading records, admissions, testing, and student records. Financial administration included payroll, general accounting, budget preparation, investment records, and general inven- tory. Physical plant management included space inventory, space cost analysis, classroom assignment, and office space assignment. Policy planning included long-range planning, institutional research, and simu- lation of institutional operations. In the overall operation of the university, none of these areas is totally independent of the other. Rourke and Brooks further analyzed the use of computerized systems in the four areas according to level of sophistication: routine, programmed procedures, management information, advanced programmed analysis, and nonprogrammed decision making.2 The interest in decision making under uncertainty, which will be discussed in detail in Chapter III, leads to more interest in those applications of M15 at the level of advanced analysis and nonprogrammed decision making, as defined by 1F. E. Rourke and C. E. Brooks, "Computers and University Administration," Administrative Science Quarterly, December 1966, p. 600. 2Ibid. 28 Rourke and Brooks, while still recognizing the necessity of lower level operating data systems to support higher level activities. No overall survey of the state of the art of M18 applications is attempted as a part of this study. In the words of Nelson, Discussions of the state of the art are often unsatisfac- tory because of the very different interests and perspectives of the participants. What are we talking about: Higher education planning?‘ Computer applications to university operations? Program planning and budgeting? Management information systems? Model building? A survey of literature to date leads one to believe that no university has a totally integrated MIS operating today. This review discusses the major characteristics of some of the more publicized systems dealt with in the literature, to give the reader some idea of the kinds of information being made available to the manager to assist in decision making. Most designers of M15 for higher education institutions view the university as a system as shown in Figure l. INPUTS Academic and Nonacademic + Production and Support Sectors I+ OUTPUTS Figure l.--The university as a system. 3C. A. Nelson, "Management Planning in Higher Education-- Concepts, Terminology and Techniques," Management Controls, January 1971, p. 6. 29 Vaj Wijk and Young defined inputs to include: 1. Student input--a description or measure of the student (or student body) being introduced into the system. 2. Financial input--a measure of resources being sent from the environment to the system to perpetuate its existence. The outputs they defined included: 1. Student output-~the behavioral change in the student input brought about as a consequence of the institution. 2. Nonstudent output--the impact of the educational process on the environment.4 Tables 1 and 2 show quantitative measures for input and output variables suggested by Hartley.5 Figure 2 shows the typical academic and nonacademic production and support structure of a university pro- posed by Van Wijk and Young.6 The three major functions--instruction, research, and public service--espoused by most university administra- tors are maintained by systems designers in their conceptual models of the university. One of the agencies that has been very active in the design, development, and implementation of management technology in higher 4A. P. Van Wijk and B. J. Young, Objectives, Program Structure and Evaluation in Higher Education: An Introduction, Research Report of the Institute for Policy Analysis (Toronto, Canada: Institute for Policy Analysis, 1971), p. 19. 5H. J. Hartley, Educational Planning, Programming and Budgeting: A Systems Approach (Englewood Cliffs, New Jersey: Prentice-Hall, 1968), p. 222, adapted from A. Astin, Who Goes Where to College? (Englewood Cliffs, New Jersey: Prentice-Hall, 1965), p. 26. 61pm., p. 15. 30 Table l.--Student input characteristics. Student Input Variable How It Could Be Measured Past achievement in high school academic - scientific - artistic -'musical - literary - oral - social Education & vocational aspirations - highest degree planned - probable major field of study - decided or undecided about studies Socio-economic background parents' educational level father's occupation number of parents living ethnic origin size of high school class Sex high school grades, high school rank placing in a scientific contest art awards, exhibitions of own art work ratings in music contests awards for writing, number of own works published in literary magazines oratory awards, participation in plays awards for leadership, offices held in school graduate work, Ph.D. degree, professional degree open primary, secondary, college graduate, post-graduate open median high school class size Table 2.--Output measurement. 31 Variables How They Could Be Measured Student Output Quantity and quality Nonstudent Output Community involvement Library growth Research and scholarly publications Economic benefits standardized test results: per- formance of students on standardized tests given in the freshman and senior years and on graduate admis- sions tests number and type of degree granted the number of seniors admitted to graduate schools questionnaires filled out by alumni giving a personal history after receiving their degrees, listing positions, salaries, participation in community affairs and graduate studies expressed in terms of lectures, cultural events, art exhibits & urban and community projects the number of books in the library expressed in terms of research grants and research publications economic implications of investment in education 32 .mcouumm cowpuzvocq owsmumumco: use uremumum qu_qxpi-.m mcszu ‘3 gain it II. III}... lat-Iii... “lat-.13.. [slit-lo Ian-ale; III-18...... .8338 pg Jilin .5 id id said J! III-III lain-loll...“ ii. Henna-g til"! gull-5.9.5.31 lion-lilac.- .Dillgh id 33.- Ill... 88383.. g.— gulogai ain't-IQ... 4 . «8.5.. £881.36 iii-815.. {pl-:58... alt-80.188. gilt-aim... liq-.8 .88: £58.- iguw 18.-3i.— .goz .12.: o .i .15. 6E: iii plural. mules-lei . cling... 38.8833 Inc-lit... 1 35.3186 iigl it‘s! Ill-8.. gun-SMbalfl-aod lung-Ha luau... . _ 5316... III-lag.- 8. no! 363.2.— 513 mu. annulus”: if. gal-El?! moo-#3983! lop-c.1858“ glue-“328.1!!! lac 85.355 .818. 8.2.7 My I. a Egan-.388... l...- , 5.5!..58-4 g.- 33.3»11‘931 ill-8.1.126 8.. . “lain-.3.— Ina-83 gal 03!... .u kilns->3 92 :83 £5.38 .53!— .- sang-Eli Satin r5522: Us... 33 education is the National Center for Higher Education Management Systems at Boulder, Colorado. The collection of tools and techniques offered by NCHEMS is in two general categories: 1. those that are used to gather historical data, and 2. those that use the historical data as a point of departure to project future costs and assist in planning for future operation. Specific instruments are Program Classification Structure (PCS), Information Exchange Procedures (IEP), Student Flow Model, Induced Course Load Matrix (ICLM), Cost Simulation Model (CSM), Faculty Activity Analysis (FAA), Cost Estimation Model (CEM), Cost Finding Principles (CFP), and Resource Requirements Prediction Model (RRPM). Figure 3 shows how some of these instruments might be used in a systematic way. Raw data are organized according to the program classification structure chosen. Output data from PCS are used to allocate support costs to departments. Course requirements for each instructional major are determined and displayed, according to the department responsible for providing the courses, in the ICLM. Department support costs and planning parameters, PCS data, and ICLM data are used to calculate costs fOr each program. Costs for each major can be converted to costs per graduate or cost per credit in any major. Planning and budgeting costs based on desired outputs are cal- culated for planned programs based on the input data. The output data of Figure 3 would be calculations of the costs required to support the programs planned, based on the input data. 34 Support Cost Output Allocation Indicators PCS / I I ~———->——. Program Cost Planning and Accounting F'— CSM ‘+’ Budgeting I Support Data ICLM I Figure 3.--Sample use of NCHEMS instruments. Figures 4, 5, and 6 show specific data and formats for some of the above—named instruments. Although NCHEMS offers the user a collection of tools and tech- niques, several commercial computerized packaged systems perform the same functions. Typically, a computer software option is sold or leased to the user. The program is controlled from a computer teletype terminal. The user interacts with the computer by supplying requested input data and specifying the outputs. One such commercial package is CAMPUS VII, designed by Systems Research Group of Toronto, Canada. Table 3 shows the data the user puts into the system. Table 4 shows the information the user might request as output. Another commercial package is PLANTRAN II, designed by Midwest Research Institute of Kansas City, Missouri. This software package allows the user the flexibility of structuring his own model of the .umEcom spun 2mu m2m=oz-:.¢ mgamwu .Amnmr..memumzm wcmsmmmcmz :owumuaum cw;m_: com mecmu Pecowumz ”ocmcopou .cmcpaomv msmumxm ucwsmmmcmz use mcwccmpa :owumuzum cmsmw: .mcwccmz .3 .u use ems: .< .m ”muczom ”“0322 mun— ._.m00 ._9 meOO .52: > _ m2_._n=0w_o 000.0 005.0 000.9 000.0 m0 A PZMEHEKEMO mwn. A 00PM. 000.0 00nd 000.N ._.m00 I_<20_._.03m._.wz_ 00N.m 00rd 000$ 000$ 053550”. mmzwaxm g 00:0}. muEFm FmOnEDw x=>_ xz._..._30m<4> >._.I_30< mmwsw§m whzwsjgomzm awhoqumm < > ”oumsopou .mswssmz .3 .0 use mus: .< .m SEINI'IdIOSICI HO SiNBWiHVdBO ”mussom 37 AVERAGE STUDENT MAJOR INDUCED COURSE LOAD MATRIX PROGRAMS A B C D 1 Eil i312 2L4» .453 I a) (D I— UZJ -—- I: .. If; .53. 2 4.3 4.5 2.1 5.2 E 8 $53 2.6 5.7 3.8 2.1 . LIJ m ._____. C3 (3 . 4 3.0 1.6 5.7 3.5 16 15 14 15 Source: R. A. Huff and C. W. Manning, Higher Education Planning and Management Systems (Boulder, Colorado: Higher Education Management Systems, 1972). Figure 6.--Typica1 NCHEMS ICLM data format. National Center for 38 Table 3.--What you put into CAMPUS VII. On degree programs The enrollment, course load per student and transition rates for each year (achievement level). The cross-loading or induced course load matrix, distributing students with each program among the disciplines or depart- ments that teach them, by percentage of discipline/department load. On departments or disciplines Faculty assumptions and characteristics--how staff members are assigned to teach, hours per week, distribution of ranks in hiring, salary rates per rank. Characteristics of courses in these disciplines or teaching departments--number offered, average section size, average hours per week per section, student credits per course, distribution of enrollments by course type, teaching staff and space for each type of course. Departmental/discipline functional computations--rules for computing support staff needs in the teaching departments, and needs for supplies, fringe benefits, etc. On administrative units ioriprograma) All of the supportive resources that the staff in administrative work require is entered, including expense and revenue items-- e.g., salaries, fringe benefits, secretarial staff, agg_ institutional revenue items such as tuition; bases on which these are to be computed. Source: George Mowbray, member of the Systems Research Group, pre- sentation to the Conference on Management Science in Educa- tion, Michigan State University, East Lansing, Michigan, August 2, 1972. (Unpublished.) Table 4. 39 --What you get out of CAMPUS VII. Single:year reports for each simulation year Program enrollment for each year of the program Program loading on the teaching departments Departmental/discipline contact hours Teaching staff requirements Teaching space requirements (sq. ft. & stations) Departmental/discipline support resources needed Administrative support resources needed in nonteaching departments and units Multi:year reports for simulationiperiod Enrollment, total costs and cost per student for each year in For For each program; division between teaching and administrative costs; cost to graduate; class and lab space required, in total and per student. each teaching department, costs of salaries, support staff and other resources; revenue generated, if any; number of staff and other personnel required; square feet of class- room, laboratory, office and support space needed; FTE enrollment and total student courses; total cost per stu- dent; faculty cost per student; total space per student and teaching space per student; student/faculty ratio; and aggregate costs per academic year of student. each administrative department or unit, approximately the same information as above, for teaching departments. Summary reports for all teaching departments and all adminis- trative departments or units. Source: George Mowbray, member of the Systems Research Group, pre- sentation to the Conference on Management Science in Educa- tion, Michigan State University, East Lansing, Michigan, August 2, 1972. (Unpublished.) 40 university. Figure 7 shows a typical input data set and Figure 8 a typical output data set. Peat, Marwick, Mitchell and Co. markets a system called SEARCH. Casasco reported that ‘ SEARCH is a generalized simulation of a college or univer- sity as an interactive system. It encompasses students, pro- grams, faculty, facilities and finances, functionally relating each of these aspects to the others so that it can simulate the behavior of a college as an operating system. Beginning with the actual present state of the institution, it simulates its future state by yearly intervals for up to ten years, based on a continuation of present operating policies and decisions as well as alternative policies and decisions the planner wishes to explore.7 Figure 9 is a sample of the projected data format. These current MIS tools were reviewed to demonstrate the kinds of quantitative data requirements the systems need to operate and the data that are provided the manager to aid in making decisions. No attempt was made to describe each system in detail. It can generally be said that the computerized systems take raw data from the operating systems of the university, manipulate them according to programmable decision rules, simulate the operation of the university, and calculate the new values of selected variables at specific time intervals. Brief History_of Decision Theory The situation with which decision theory deals is this: Given two perceived states, A and B, into either one of which an individual may place himself, the individual chooses A in preference to B (or 7J. A. Casasco, PlanningiTechnigues for University Management (Washington, D.C.: American Council on Education, 1972). 41 .ooosm apes eggs? as zssezsss .eowaxh--.s oe=m_s .Ammmp .mgsuwumsH susummma ummzuwz ".oz .xpwo mumssxv HH ssspsmps op sowpusuospsH s< A; mm. + N “zosessos as; I as "zosesscm as a s3 ”zosessom ass» mus CON mmsmsuzs so. \ es _o. .53 ”zose<30s ms_o.ses ”zosessom m3-oos ”zosessom m sue m. mmsmsUZH s mus m.~ so smsmsszs ezmosms Am; I N40 \ _s “zosessom s rpm zs zssmm m. so mmsmsoso ezmumss s osm 2H wzszzsssm _ mmsmsUZH msms.omms.ooms.moms.owms.osmsses0 20HH wm0esssss Ssm emsmms sue msssm>< mezsemsmms szssusme >essuss osszss mes mezsemsmms szszsss» use >essumes=u< owsssoms >es=oessu< stm zosessm mg<¢m>< mm: Hzos ezmszem >ssmm3 onhmHmummo ”mussom m~ N_ F— 0— k0 Q'OSLD F mzH 20m 4h~mmm>mza >z< 42 .Amsm_ .opzpscmes soeeomos pauses: mm.0mmm¢m om.0mm0m_ mo.mmm0nm um.ammm mu.¢00P0m 0n.¢0m0¢ mm.m_mmu «v.0cmFNN mmmp 00.um0mum mm.mmmmmm oo.mmmmmp 00.0mpqu 00.N0-0m mm.00—¢0m 0m.0000 u_.mmom Rm.~000mm 00.0000mm —m.wuomm u0.00~mm m0.omwm¢ mm.mw—0m uu.mm000m mp.Nmn00N sump mnmp 20m uh<0 kzmmmzu .eoosm aces passao ss zssezsss _eossse--.m os=a_s ".oz .xuwu mumsuxv HH sssusmFQ up sowuusuospsH s< ”mussom 0m.u0uomm mm.mmmumm 00.nmmmpm mmzmsxm Hams 4h430FHmmm>sz >z< STUDENTS ENROLLED FRESHMAN SOPHOMORE JUNIOR SENIOR NO OF FACULTY-~TOTAL PROFESSORS ASSOCIATES ASSISTANTS INSTRUCTORS ADJUNCT COURSES FRESHMAN SOPHOMORE JUNIOR SENIOR SECTIONS FRESHMAN SOPHOMORE JUNIOR SENIOR FACULTY LOAD (HRS/YR) STU-FAC RATIO CREDIT HOURS PRODUCED AVE CLASS SIZE TUITION INCOME INSTRUCTION EXPENSE INSTR COST/CRED HR Figure 43 YEAR 1970 1971 1972 1973 1974 1975 770 815 832 856 856 856 293 293 293 293 293 293 194 234 234 234 234 234 158 147 174 174 174 174 125 141 131 155 155 155 55 55 55 55 55 55 9 9 9 9 9 9 16 16 16 16 16 16 23 23 23 23 23 23 7 7 7 7 7 7 0 0 0 0 0 0 158 158 158 158 158 158 25 25 25 25 25 25 25 25 25 25 25 25 51 51 51 51 51 51 57 57 57 57 57 57 226 226 226 226 226 226 64 64 64 64 64 64 51 51 51 51 51 51 52 52 52 52 52 52 59 59 59 59 59 59 26 26 26 26 26 26 14 14 15 15 15 15 21741 24983 25530 26226 26226 26226 17 16 17 17 17 17 1305 1407 1437 1479 1479 1479 708 708 713 716 719 724 26 28 27 27 27 27 9.--Typical SEARCH data format. 44 vice versa). Throughout the history of social research, great minds have sought to understand the behavior of individuals in such situa- tions. Bodies of theory have been accumulated throughout the years under such varying names as decision theory, value theory, utility theory, theory of chance, theory of decision making, and others. Kauder credited Aristotle with creating the concept of value in use, and cited readings from one of the great philosopher's best- known works, The Topics, as evidence of Aristotle's well-developed theory of utility in human choice.8 Aristotle's theory dealt primarily with the economic marginal utility of goods, wherein marginal utility was based on the value of the last piece of goods exchanged. Aristotle's thoughts about utility theory, like most of his other works, lay dormant for over one thousand years. In the thirteenth century, growing debate over market forms and just prices led medieval doctors to revive the Aristotelian theories. Urged on by philosophers such as Thomas Aquinas, Henry of Ghant, and Johannes Buridanus, a value theory developed that was a mixture of economic cost and sub- jective value; utility was based on the general welfare of the commu- nity and not the pleasure of the individual. Kauder traced the threads of Aristotelian influence through the centuries to the young Italian economist, Abbé Galiani,9 who developed a value theory based entirely on subjective estimation, wherein value was defined as the 8E. Kauder, "Genesis of the Marginal Utility Theory," Economic Journal 63 (September 1953): 638. 91bid., p. 644. 45 ratio of utility and scarcity. For the first time, value had meaning for something other than economic goods. About the same time that Galiani was developing his subjec- tive theory, a young English minister, Thomas Bayes, was developing his solution to some of the problems of the doctrine of chance, based 10 on the mathematical work of Bernoulli. In an introduction to his essay, Bayes wrote: . that his design at first in thinking on the subject of it was to find out a method by which we might judge concerning the probability that an event has to happen, in given circumstances, upon supposition that we know nothing concerning it but that, under the same circumstances, it has happened a certain number of times, and failed a certain number of times.H The key to Bayes' mathematical work was the proposition that preceding an experiment, the chance of an event occurring could be estimated to lie within a probability interval based on the number of times the event had happened or failed to happen under similar circum- stances in the past. The relationship of Bayes' essay to the theory of choice is demonstrated by his proposition 2: If a person has an expectation depending on the happening of an event, the probability of the event is to the probability of its failure as his loss if it fails to his gain if it happens. Suppose a person has an expectation of receiving N, depend- ing on an event the probability of which is P/N. Then the value of his expectation is P, and therefore if the event fails, he loses that which is in value P; and if it happens he receives N, but his expectation ceases. His gain therefore is N-P. Like- wise, since the probability of the event is P/N, that of its failure is (N-P)/N. But P/N is to (N-P)/N as P is to N-P, i.e., 1OT. Bayes, "Essay Towards Solving a Problem in the Doctrine of Chances," Philosophical Transactions of the Royal Society 53 (December 1763): 370; reprinted in Biometrika 45 (1963): 293. 11 Ibid. 46 the probabilitonf the event is to the probability13f its failure as hlS loss if it fails to h1S gain If it happens. Bayes' concept of probability as determined by the frequency of occurrence of events formed the basis for the classical approach to the mathematical theory of chance that dominated thinking during the eighteenth and nineteenth centuries. The first half of the twentieth century saw the development of the classical decision theorists. This classical approach can be described as empirical, an attempt to justify propositions on the basis of data. The key to the expected success of the classical approach was the empirical verification of utility. Edwards expressed the view of that period: People choose the alternative, from among those Open to them, that leads to the greatest excess of positive over nega- tive utility. This notion of utIIity maximization is the essence of the utility theory of choice. Classical theorists used utility theory to establish the nature of the demand for various goods. Assuming that the utility of any good is a monotonically increasing, negatively accelerating func- tion of the amount of that good, theorists expected to show that the amounts of most goods a consumer would buy are decreasing functions of price--functions that are precisely specified once the shape of the utility curves is known. 12Ipio. 13W. Edwards, "The Theory of Decision Making," Psychological Bulletin 51 (April 1954): 381. 47 Edwards reviewed the numerous attempts by classical theorists to derive empirical marginal utility functions and some of the com- plexities that led to the eventual abandonment of the approach.M A new era in decision theory began with the publication in 1944 of Von Neumann and Morgenstern's book, Theory of Games and 15 The authors modified the classical approach to Economic Behavior. utility by requiring that an individual can completely order proba- bility combinations of states. This simple modification led to the concept of expected utility. For example, suppose an individual is indifferent between a certainty of $5.00 and a 60-40 chance of winning $7.00. It can be assumed that these two alternatives have the same utility. If the utility of $0.00 is defined as O utiles (units of utility) and the utility of $7.00 as seven utiles, we have assigned the two arbitrary constants in a linear utility transformation. Then the utility of $5.00 can be calculated by using the following concept of expectation: U($5.00) 0.6 U($7.00) + 0.4 U($0.00) 0.6(7) + 0.4(0) 4.2 By varying the odds and using the linear transformation already determined, it is possible, in principle, to determine the utility of any other amount of money. In a classical paper on decision theory, Edwards, Lindman, and Savage characterized the Von Nuemann and 14Ioio. 15J. Von Nuemann and O. Morgenstern, Theory of Games and Economic Behavior (Princeton, N.J.: Princeton University Press, 194411 48 Morgenstern theory as the "decision-theoretic formulation of statis- tical inference," and reviewed some of the scientists, such as Wald, 16 The mood of deci- Neyman, and Pearson, who championed that theory. sion theorists during that period was to act on the basis of a deci- sion determined from a point estimate of a parameter such as the expected value, x, of a variable. The decision-theoretic approach was often successful in predicting human choice where money was the exchange, but suffered deficiencies when other commodities were tested. Like most decision theory before that time, predictions were more successful in risky situations than in uncertain situations. The inaccuracy of the theories is generally attributed to the failure to consider personal or subjective probability. The most controversial new decision theory since about 1960 is grounded in Bayesian statistical inference. Bayesian statistics is said to have begun in 1959 with the publication of Probability and Statistics for Business Decisions by Robert Schlaifer. Edwards and his associates claimed that Bayesian statistics was a "reversion to the statistical spirit of the eighteenth and nineteenth centuries." Paradoxically, Bayesian statistics should not be confused with the theoretical viewpoint of the man for whom it is named, Thomas Bayes. 16W. Edwards, H. Lindman, and L. Savage, "Bayesian Statistical Inference for Psychological Research," Paychological Review 70 (May 1963): 193; A. Wald, "On the Principles of Statistical Inference," Notre Dame Mathematical Lectures, Vol. I (Ann Arbor: Edwards Brothers Press, 1942); J. Neyman, "Outline of a Theory of Statistical Estima- tion Based on the Classical Theory of Probability," Philosophical , Transactions of the Royal Society 236 (1937): 333-380; E. Pearson and L. Savage, The Foundations of Statistical Inference: A Discussion (New York: John Wiley and Sons, 1962). 49 Edwards et a1. stated that: "Bayesian statistics is so named for the rather inadequate reason that it has many more occasions to apply Bayes' Theorem than classical statistics has."17 The elements embraced by modern Bayesian decision theory, which distinguish it from earlier theory, are 1. the definition of probability as a particular measure of opinions of ideally consistent people, 2. the view of statistical inference as a modification of those opinions in light of evidence, with Bayes' Theorem specifying how such modifications should be made, and 3. the implication that the rules governing when data col- lection stops are irrelevant to data interpretation. This thought represents a radical departure from the classical concept of probability as the limit of the frequency of occurrence of events, the classical approach to hypothesis testing based on the out- come of an experiment without regard to prior probabilities, and clas- sical experimental design in which data collection is carefully planned and terminated before interpreting the data. This new philo- SOphic attitude opens the door to consideration of subjective or per- sonal probability, at least a priori. The potential for a workable theory of individual decision making under uncertainty is increased. The mathematical sophistication and experimental testing of this new approach to decision theory are still undeveloped. Classi- cal statisticians now recognize this new approach, even though it has 17Ibid. 50 not received unanimous endorsement.18 Neither the Bayesian approach nor any other decision theory to date adequately predicts human beha- vior in individual decision making under uncertainty. Enthusiasm for the success of decision-making theory in economic goods or consumer choice situations tends to obscure some of the apprehensions of behavioral scientists about the validity of expected utility, personal probability, and significance when value determinants are nonmonetary. Basic questions still remain, such as numerical combinations of probabilities and values, the assumption that individuals are always seeking to maximize utility, the psycho- logical impact of risk and uncertainty on human behavior, and the cost of incorrect decisions. This section presented a brief review of the chronological development of formal decision theories from the time of Aristotle, 400 B.C., to the present. The impact of economic choice is obvious throughout the history of decision theory. Although several publica- tions in the behavioral sciences, such as psychology, have been cited, the gain-loss variable in decision theory has still been mainly economic goods. The next section focuses on the assumptions of the theories of individual decision making. 18R. Kirk, Experimental Design: Procedures for the Behavioral Sciences (Belmont, California: Brooks/Cole Publishing Co., 1968), p. 32; W. Hays, Statistics (New York: Holt, Rinehart and Winston Publishers, 1963). 51 Theories of Individual Decision Making The method of theorists concerned with the theory of decision ‘9 They making was characterized by Edwards as the "armchair method." make assumptions and from them deduce theorems that presumably can be tested. One such theory is The Theory of Riskless Choices, sometimes called the theory of economic man. It is assumed that anyone who makes a decision to which this theory applies is an economic man. The properties of an economic man are: 1. He is completely informed about all courses of action open to him and what the outcome of each action will be. 2. He is infinitely sensitive. It is assumed that the alter- natives available to him are continuous and infinitely divisible, and that prices are also infinitely divisible. 3. He is rational. Rationality implies two things-—that the decision maker can weakly order the states of nature and that he makes his choices in order to maximize something. This property of ration- ality has led to a bulk of formal theorizing in economics, psychology, and management science. The theory of utility has developed in an effort to lend structure to the ordering of preferences. Sophisti- cated techniques, such as linear programming, have developed to facili- tate maximization. The Theory of Risky Choices is thought to have originated with Von Neumann and Morgenstern's Theory of Games and Economic Behavior, 19W. Edwards, "The Theory of Decision Making,” Psycholpgical Bulletin 51 (April 1954): 381. 52 published in 1944.20 Since, under conditions of risk, the outcome of actions is known to the decision maker only in a probabilistic way, the assumption of complete information is violated. The Theory of Risky Choices assumes: l. The decision maker can completely order probability com- binations of states of nature. 2. He will act so as to maximize the expected value (average value) of utility of outcomes. This theory has generated much activity, mostly under the title of game theory or gambling. Luce and Raffia made a classical presenta- tion of such theories.2] Halter and Dean presented empirical examples from agriculture and natural resources of risky decision making.22 The other category of individual decision making is decision making under uncertainty. Under uncertain conditions, the decision maker does not know the state of nature at the time of choice, the probability of a given state of nature, or what states of nature are associated with what available actions; in an extreme case, he may not even be conscious of all possible states of nature or alternative actions. The extreme situation would be one of mathematically "unbounded" variables and psychologically "complete ignorance." No consistent theory exists to predict behavior under these conditions. 20Ipid. 21R. D. Luce and H. Raffia, Games and Decisions (New York: John Wiley and Sons, 1965). 22A. Halter and G. Dean, Decisions Under Uncertainiy (Chicago: Southwestern Publishing Co., 1971)} p. 139. 53 Approaches to developing theory have been based on the assumption that the set of states of nature forms a mutually exclusive and exhaustive list of those aspects of nature that are relevant to the particular choice problem about which the decision maker is uncertain. Earlier theorists attempted to modify the Theory of Decision Making Under Risk to predict decision making under uncertainty, by introducing the notion of subjective or personal probability.23 The notion attempts to use the decision maker's personal a priori probabil- ity over the states of nature in place of an objectively arrived at probability distribution function. The logical validity of the notion of subjective probability has drawn serious criticism. Bayesian decision theorists have accepted the use of subjective probability a priori and applied Bayes' theorem to calculate conditional probabilities to aid in a posteriori evalua- tion of the chance of events occurring. Empirical techniques for determining subjective probability functions have been slow to gain the approval of many decision theorists, and have failed to survive. One might begin to wonder at this point if it is not impos- sible to predict the behavior of individual decision makers under uncertainty, because of the ignorance of the decision maker about the possible states of nature and the probabilities associated with those states. Cannon and Kmietowicz discussed this problem and alterna- tives for conducting research in spite of the lack of an objectively 23S. V. Vail, "Alternative Calculi of Subjective Probability," in Decision Processes, ed. R. Thrall, C. H. Coombs and R. L. Davis (New York: John Wiley and Sons, 1955). 54 24 Researchers make the basic assumption that admin- verified theory. istrators, in an uncertain environment, do not operate in complete ignorance but bring some a priori knowledge to bear on the choice situation. Having made this assumption, one can draw on the work of theorists such as Luce and Raffia and Halter and Dean for a theoret- ical conceptual framework of decision making under partial ignorance and the use of a priori information.25 Halter and Dean used Bayes' formula in the decision-making framework, to provide a means of expressing conditional probability. Given two sets of not mutually exclusive events, A1 and A2, then P(A2/A]) = PISIA‘AZI I TI where A]((A2 is the union of the two sets (Figure 10). Figure lO.—-Two sets of not mutually exclusive events. 24C. M. Cannon and Z. W. Kmietowicz, "Decision Theory and Incomplete Knowledge," The Journal of Mana ement Studies, October 1974, p. 224 25Ibid. 55 Also P(Al/AZ) = P(ATIIAzI PIA2) Now if E is any subset of A made up of one or more of the subsets A1, and the subsets EI1A1, (i=l,n), are mutually exclusive and exhaustive of E, then n n P(E) = Z P(Er‘Ail = Z P(Ai) P(E/Ai) i=1 i=1 and in Bayes' formula = P(EFIA ) P(Aj/E) J and P(Aj/E) - P(Aj) P(E/Aj) n z P(Ai) P(E/A1) This form of Bayes' formula can be used in the development of decision theory under uncertainty with partial ignorance. A body of theory has developed, which is based on different strategies using the Bayesian a priori probability approach. This mathematical theory is useful if quantitative values for the probabilities can be obtained. The Bayesian approach can alSo be used to develop a conceptual activity model of individual decision making under uncertainty. An example of the application of Bayes' formula to which most readers can relate is deciding which card has been drawn from a stan- dard poker deck. If a standard deck of poker cards is shuffled thoroughly, spread on a table face down, and one card drawn from the deck, a participant might be required to determine which card has been 56 drawn. The participant knows the card drawn is one of a set of fifty-two cards. Let us call this set A. Based on this a priori information, the probability of the participant making the correct choice is 1/52. Set A can be further divided into subsets according to suit, color of suit, number on the card, face cards, or sex of face cards. A likely list of subsets might be: A1 = aces A2 = kings A3 = queens A4 = jacks A5 = tens A6 = nines A7 = eights A8 = sevens A9 = sixes A10 = fives All = fours A12 = threes A13 = twos These subsets are mutually exclusive. Any card chosen can belong to only one of the subsets. If mapped in set space such as in Figure 10, there would be no overlapping. At this point the participant has no information with which to improve his chance of making the correct choice. Let us assume the participant decides that the card drawn was the queen of hearts. This is a preliminary decision; the partici- pant is likely to seek more information before making a final choice. 57 Assuming that an unbiased observer peeks at the card and tells the participant the card drawn is a heart, the participant can now focus his attention on a subset,E,cfl’thelarger set, A, which contains only thirteen cards, namely hearts. A list of subsets, E, is E1 = hearts E2 = spades E3 = clubs E4 = diamonds The union of E] with subsets A1 to A13 is exhaustive of E], in that all of the elements of E1 are accounted for. Bayes' formula can be used to calculate the probability of the queen of hearts being the correct choice, given the new informa- tion. The probability is given by P(A3)P(E]/A3) P(A3/EI) ‘ 3 z P(Ai)P(E1/Ai) T 1 where A3 - queens E1 = hearts and P(Ai) = l/13 P(A3) = 1/13 P(EI/A3) = 1/4 P(El/Ai) = 1/4 The probability of the queen of hearts being the correct choice is 1/13) 1/4 _ 13(1/13I11/4) ‘ 1/13 58 At this point, the participant must choose between the queen of hearts as a final choice or seeking more information to increase the probability of his final choice being the correct one. Related Studies in Individual Decision Makipg_Under Uncertainty A search of the literature revealed much discussion of the theoretical and axiomatic approaches to prediction of decision pro- cesses, but a relative dearth of experimental or quasi-experimental studies on the subject. Soelberg, in a study of members of the gradu- ating class of the Sloan School of Management at MIT, investigated "how individuals make important, difficult, and highly judgemental 26 Soelberg used an expanded model of Simon's character- decisions." ization of the decision process with the following structural phases: 1. Participation--the decision maker (Dm) is induced to work in a given task environment, in which he is motivated to attain one or more nontrivial objectives. 2. Recognition and definition--Dm surveys his task environ- ment and then discovers, selects, or is provided with the particular problem he intends to devote his resources to solving. 3. Understanding--In a search for solution alternatives, Dm investigates his task environment to formulate and test 26F. Soelberg, ”Unprogrammed Decision Making," Studies in Managerial Process and Organization Behavior, ed. J. H. Turner, A. C. Filley and R. J. House (Glenview, Illinois: Scott, Foresman and Co., 1972). 59 hypotheses about the apparent cause-effect relationships in the environment. Design--Dm develops or searches for alternative courses of action for solving his problem. Evaluation--Dm assigns some sort of value measure to the estimated consequences of his perceived decision alter- natives. Reduction--Dm reduces his set of viable decision alter- natives to a single one. Implementation--Dm introduces and manages his decision solution in the task environment. Feedback and control--Dm receives and evaluates informa- tion from the task environment. Soelberg used the task of finding a job as the decision to be resolved. 1. Results of the study indicated that The search for new alternatives terminated before the time at which subjects reported having made a decision. Two or more acceptable alternatives were selected before terminating the search. Great uncertainty about the final choice existed at the termination of the search. Most subjects made final choices that had been indepen- dently identified as their choices before reporting the decision. 60 5. No subject reported a reduction in dissonance in his liking for accepted versus rejected alternatives imme- diately following the decision. Soelberg cited as implications of his research the multi- dimensionality of utility, personal probability, and the parallel and continuous form of the human decision process. His findings are con- sistent with the theoretical concepts on which the model of the present study is based. Soelberg's findings also imply that the decision maker tends to focus his activity on interacting with the real world to gain information from which to formulate alternatives at one time, and at other times he focuses on a simple decision strategy for the final choice. Morlock studied the effect of outcome desirability on the amount of information required to make a decision.27 In repeated decision tasks, subjects were allowed to acquire as much information as necessary to make a decision in which the subject knew in advance the outcome and payoff for a correct decision. The outcomes were known to be desirable or undesirable, with associated risks. Find- ings indicated'Umat,under moderate levels of difficulty, less infor- mation was required to decide that a desirable event would occur. These findings imply that the confidence the decision maker associates with quantitative information is related to the utility of such information as perceived by the individual. 27H. Morlock, "The Effect of Outcome Desirability on Infor- mation Required for Decisions," Behavior Science 12 (June 1967): 296. 61 Porat and Haas, in an experiment dealing with the effects of initial information and feedback on goal setting and performance, tested three hypotheses: l. The more specific information a decision maker has, the more accurate will be his levels of aspiration and decision. 2. The setting of goals will be a function of previous goals and comparative experiences in similar organizations. 3. Decision makers with less specific information initially will exhibit a higher rate of learning.28 The information was quantitative data about the firm in which the subject was presumably operating. Hypothesis 1 was not rejected, although there was an indication that the marginal utility of addi- tional information declined with an increase in quantity. Hypothesis 2 was not rejected, but difficulties with the data analysis made the outcome questionable. Hypothesis 3 was not rejected. Based on profit in the industrial firm, subjects with little initial information showed a greater improvement than did those with more initial information. This trend was most obvious when no information was compared with information, which led the researchers to point out the difficulty of assessing incremental information effects. The research cited tends to imply that quantitative management information has a measurable effect on the decision-making process of the individual under uncer- tainty. 28A. Porat and J. Haas, "Information Effects on Decision Making," Behavioral Science 14 (December 1969): 98. 62 Preston and Baratta, in an experimental study at the University of Pennsylvania, examined the relationship between the price indi- viduals were willing to pay for prizes and expected payoff of the prizes in an uncertain environment.29 A deck of forty-two cards was marked with six prize values and seven corresponding probabilities for each prize. Subjects were given a fixed sum of money at the beginning of the experiment. The cards were randomly auctioned to the highest bidder and the actual payoff determined by the roll of dice. Results of the data analysis indicated that subjects were willing to pay more than the mathematically calculated payoff for low-probability prizes and less than the mathematically calculated payoff for high— probability prizes. This behavior was independent of the cash value of the prizes, and was consistently demonstrated by college students, mathematicians, psychologists, and statisticians. The indifference point fell at approximately 0.20 on the probability scale. The researchers concluded that subjects conceived the probabilities asso- ciated with prize payoffs differently than the mathematically expected payoff. The difference was attributed to psychological probability, which the researchers defined as that probability which must be used to bring the price paid into rational relationship with the prize. Cummings and Harnett conducted experimental studies at the University of Indiana, aimed at determining the effects of information 29M. G. Preston and P. Baratta, "An Experimental Study of the Auction-Value of an Uncertain Outcome," American Journal of Paychology 61 (February 1948): 183. 63 on bargaining behavior in decision making.30 The researchers attempted to simulate the effects of real-world variables in a bargaining paradigm. The subjects were placed in a three-channel bargaining relationship such as manufacturer, wholesaler, and retailer, and were provided with different amounts of information under controlled communication con- ditions. Results indicated that: 1. For bargaining in a channel relationship, a bargainer had no profit advantage in being completely informed if his bargaining opponents were completely uninformed. 2. The less information available in the channel, the greater the impact of risk-taking propensity on bargaining behavior. 3. In the superior-subordinate relationship, a bargainer who was provided with complete information regarding the other party's possible monetary rewards had a more realistic bargaining attitude. 4. Allowing the bargainers to comnunicate with one another influenced bargaining in a manner similar to the effect of information. In a study at the University of Minnesota, Chervany and Dickson investigated the problem of information overload.31 Graduate students in the School of Business were provided with quantitative management information about the operation of an industrial manufacturing firm in 30L. L. Cummings and O. L. Harnett, "Managerial Problems and the Experimental Method," Business Horizons, April 1968. 31N. L. Chervany and G. W. Dickson, "An Experimental Evalua- tion of Information Overload in a Production Environment,“ Management Science 20 (June 1974): 1335. 64 the typical raw data format and in the form of statistical summaries. During a weekend of intensive training and decision making, experimental gaming was used to evaluate the decision outcomes of groups receiving the two types of information. Minimization of operating costs, time required to make decisions, and subject confidence in the decisions made were the experimental measures. Results indicated that subjects who used the statistically summarized information had significantly lower operating costs than did the subjects who used raw data, showed less variance in operating costs than did subjects who used raw data, but had less confidence in their decisions than did subjects who used raw data. Subjects using the statistically summarized management information took more time to make decisions than did their counter- parts using the raw data. Most studies reported in the literature used the same type of management information with variation in the amount of information, exchange of information, or the search for information. The present study is the first known to examine the effect of different kinds of information--quantitative versus nonquantitative. The study of Chervany and Dickson is similar, but both information treatments were quanti- tative, according to the definition given in Chapter I. The studies reported here show a definite bias toward the use of the experimental method in research on the behavior of individuals in decision making under uncertainty. Cummings and Harnett urged more use of experimentation in research on management-type problems.32 The experimental design 3IIbid. 65 presented in Chapter III is aimed at investigating the effect of quantitative management information on the generation and selection of alternatives in an uncertain decision environment. CHAPTER III THE EXPERIMENTAL DESIGN At the founding conference of the Society for Management Information Systems held at the University of Minnesota, management information systems professionals were asked to rank twenty-six poten- tial research projects. The two projects receiving the highest rank- ing, according to Chervany and Dickson, were: 1. development of methods for determining what the content of an information system should be, and 2. investigation of the characteristics of decision makers that affect MIS design.1 The content of an information system is measurable on several dimensions. One of the dimensions is the quantitative or nonquanti- tative form in which operational data are presented to the user. The experiment of this dissertation was designed to evaluate the effect of information treatments on the generation and selection of alterna- tives in individual decision making in an uncertain environment. A theoretical activity model is developed, research questions posed, procedures described, data analysis techniques explained, and the data collection instrument presented in this chapter. 1N. L. Chervany and G. W. Dickson, "An Experimental Evalua- tion of Information Overload in a Production Environment," Management Science 20 (June 1974): 1335. 66 67 Theoretical Model of Individual Decision-Making Activity The research of this dissertation was grounded in the theory of decision making. Most of the theoretical developments in decision making have been reported in the journals of psychology, economics, statistics, management science, and business administration. A review of the history, theories, and experimental studies from the literature was presented in Chapter II. The field of decision making can be classified on two dimen- sions: l. decision making--individual or group 2. conditions--certainty, risk, or uncertainty The three conditions are: a. Certainty--if each action is known to lead invariably to a specific outcome. b. Risk-~if each action leads to one of a set of possible specific outcomes, each outcome occurring with a known probability. c. Uncertainty--if either action or both has as its conse- quences a set of possible specific outcomes whose proba- bilities are completely unknown or are not even meaningful. Figure 11 shows the resulting matrix of decision activities that could be studied. The present study is concerned with Type-C decisions, individual and uncertain. The rationale for this choice was twofold: 68 l. a belief that implementation of M15 requires an under- standing of the possible impact on administrative behavior, and 2. a belief that the condition of uncertainty is more des- criptive of the environment in which educational adminis- trators live and work. Certainty Risk Uncertainty Individual Type-A Type-B Type-C Group Type-D TYPe-E Type-F Figure ll.--Decision activity matrix. In summary, this study concerns individual decision making under conditions in which any_action has as its consequences a set of possible specific outcomes, but where the probabilities of these out- comes are completely unknown or are not even meaningful and the decision maker might not even be conscious of all possible outcomes. The example of the application of Bayes' formula to decision making in Chapter II was a Type-B decision--individual decision making under risk. The application of Bayes' formula in the decision-making process under uncertainty is not so easily or clearly demonstrated. A 69 model developed by this researcher to relate the formula to the decision-making process under uncertainty is shown in Figure 12. The model depicts the closed loop decision-making process an individual uses in an uncertain environment. The model allows for an unspecified number of possible states of nature in the real world, represented by s], $2, 53 ... Sn“ Through interaction with the real world, events occur whose outcomes 0], 02, 03, ... on provide information for the decision maker. The information may be acquired directly or provided by analysts, subordinates, administrators, or other sources. It may be quantitative or nonquantitative information. This information affects the decision maker's perceived probability of the possible states of nature, P(sj); his perceived probability of the outcome given the particular state of nature, P(oi/sj); the alternative actions available, a], a2, a3, ... an; and the utility (value) asso- ciated with the given actions for a particular state of nature. The individual combines the a priori probability, P(sj), and the condi- tional probability of the outcome of the event, P(oi/sj), according to Bayes' formula, to arrive at the new probability of the particular state of nature, P(sj/oi), based on the new information. The decision maker replaces the original a priori probability, P(sj), with the new probability, P(sj/oi). This probability is used in combination with alternative courses of action and the decision maker's utility of outcomes to form decision rules, or strategies,i%w~the action to be taken or recommended. The choice may be tentative, with the decision maker returning to the real world to seek new information before making 70 .mswxue sowmwumu sow Pmuoe sswmmaumii.m_ mssmws : ....m .N ._ IA... 0 0 0 0 mopsm sowmwumo MwmnhOVD 3___S= . ee...me.~e.se < msoZu< I mumssmu_< .n .F P +h w 3.. fit F =m....mm.mm.sm _I'II'I, 7&5: .P .n wsspmz so magnum u A o\ mvs UFL03 —mmm 0 A.mvs 71 a final choice, as was demonstrated in the card-choosing example of the previous section. The model is a closed loop, so there is not necessarily any starting point or sequential path to follow. The decision maker may be at any point in the process at any time, and indeed may move in either direction in the process of finalizing a decision. The primary function of Bayes' formula is to identify the modification of the decision maker's perceived probability of the states of nature hithe real world to account for information obtained from the outcomes of events occurring or having occurred before the time of the decision. The probabilities need not be objective. The model is also descriptive of subjective, or personal, probabilities. Dean and Halter showed evidence from empirical examples of individual decision making under uncertainty in oil well drilling, agricultural crop selection, turkey farming, and livestock management that a com- bination of objective and subjective probability is often used in real Tiie.2 Since this study concerns quantitative management information and its effects on the decision process, a partial list of points in the model where such information enters the process will demonstrate the usefulness of the model in the experiment. 1. The outcome of events occurring in the real world can be partially described in terms of quantitative parameters. In the 2A. Halter and G. Dean, Decisions Under Uncertainty (Chicago: Southwestern Publishing Co., 1971). 72 university setting, such parameters might be course demand, student flow, attrition rates, economic forecasts, and employment profiles. 2. The utility associated with particular states of nature can be quantified on some dimensions. In universities, parameters such as tuition income based on enrollment patterns, athletic profits based on sports activities, and legislative income based on FTE stu- dent enrollment are readily quantifiable. 3. Alternative actions are, for the most part, quantitatively presented. In the college environment, parameters such as admissions criteria, program changes, publicity budgets, recruiting, and staffing changes are very quantifiable. Areas in which quantitative information exerts influence in establishing parameters and their magnitudes but where the primary functions are more likely to be subjective are: 4. Formulating_probability functions foripossible states of pature, Although quantitative data on the outcome of events occurring in the real world are taken into consideration by the decision maker, intuition, judgment, experience, political perceptions, hunches, and "gut-level" feelings are likely to be heavily weighed in shaping per- sonal probabilities. 5. Formulating conditional_probabilities of states of nature is likely to involve considerable judgment on the part of the decision maker. This same problem is faced by researchers in the use of hypothesis testing based on experimentation. The decision maker, like the researcher, subjectively assigns confidence criteria to the empiri- cal information about the states of nature. 73 6. The application of Bayes' formula seems to be a highly objective process, but this is a deceptive viewpoint. Even if the decision maker had assigned quantitative values to each of the a priori and posterior probabilities in the formula, there is still the problem of which terms will be included in the summation of conditional proba- bilities in the denominator. In other words, how many states of nature are relevant to the matter about which the individual is uncertain? Also, how many states of nature can a human be cognitively aware of at one time? March and Simon proposed the concept of "limits of cog- nitive rationality," which implies that an individual can only attend to a limited number of things at any given time.3 This places limits on how many conditional probabilities are included in the weighting factor to determine posterior probability. 7. Decision rules (or strategies) can seldom be guanti- tatively stated and are likely not to be pure strategies but some form of mixed strategy. This is especially true in higher education, where goals and subgoals are seldom explicitly stated (or even known) by the decision maker. Some strengths and limitations of the model should be recog- nized. Limitations are: l. The model does not show the parallel or multidimensioned activity that the human mind is capable of engaging. 3J. G. March and H. A. Simon, Organizations (New York: John Wiley and Sons, 1958). 74 2. The model implies discrete paths for moving from one activity to another, when in actual practice the individual may skip haphazardly from point to point at will. 3. The model does not rank activities in terms of time spent or relative contribution to the decision process. Strengths are: l. The model is simple. It reduces a complex process to a relatively simple and manageable framework. 2. The model is explicit. Relationships among different activities in the decision process are functionally demon- strated. Discussion of the decision process is facilitated without the limitations of verbal communication. Grayson, in discussing the procedures on which the model is based, observed: . . But even if such procedures are not adopted in total now, the mere discussion of them may help operators (1) to realize the problems that they are now handling implicitly in their minds, and (2) to think about them in a more for- mal manner. 3. The model is continuous, so that no starting or ending point is implied. This is consistent with intuitive feel- ings about human behavior. Complex decisions seldom are discrete and independent of the past or future. Action is often stimulated by exogenous demands such as time dead- lines, resource depletion, or demands from superiors. 4C. J. Grayson, Decisions Under Uncertainiy: Drilling Decisions peril and Gas Operators (New York: Plimpton Press, 1960): 75 This model presents the theoretical-conceptual framework for the experiment. Research Questions So little research has been published on the subject of indi- vidual decision making under uncertainty that this study is considered exploratory. For this reason, the research hypotheses are stated in the form of questions rather than as null hypotheses. The question framework of stating research hypotheses allows more flexibility to explore whatever results the data might reveal. The problem being investigated in the study is the effect of quantitative management information about the state of an uncertain environment on the generation and selection