ABSTRACT ORGANIZATION SIZE AND ADMINISTRATION IN NATIONAL LABOR UNIONS: A COMPARATIVE ANALYSIS BY Joe Alan Miller In a descriptive study, the relationship between organization size and relative size of the administrative component was investigated, using 1964 data on 96 national labor unions in the United States. These were unions having (1) more than 5,000 members, (2) with two exceptions, a majority of members working for private employers, and (3) a structural similarity in being parent organizations of independently chartered local affiliates. Data for the study were obtained from (1) public records available from the U. S. Department of Labor, (2) union publications and reports, and (3) other published materials on particular unions and industries. A full dues-paying membership figure was used as the indicator of organization size. Other independent variables included (1) an industry type classification, based on the predominant product market characteristics of the industry(ies) in which a union holds jurisdiction, (2) a physical density index, based on the average size of local memberships, and (3) resource capacity, as indicated by a per capita receipts figure based on contributions by Joe Alan Miller members to the national organization. Relative size of the administrative component was determined by three measures: (1) the ratio of full-time officers and staff in the national organization to total membership--the total administrative component; (2) a similar ratio based on those officers and staff directly monitoring and servicing subnational units-- the direct administrative component; and (3) a third ratio, similarly calculated, based on the number of full-time national staff engaged in administrative tasks of an indirect nature--the supportive administrative component. The findings showed that the relative size of each aspect of the administrative component is much smaller among unions in local product market industries, as com- pared with their counterparts in industries where this market is national or broad in geographic scope. These findings concur with the suggested relationship in the literature on union organization to the effect that na- tional unions in predominantly local market industries are more administratively decentralized than are their counter- parts in national market industries. As for the organization size-administration rela- tionship, the results vary according to industry type. Only in the case of the interrelationship between organi- zation size and resource capacity were the results clear-cut. The findings generally indicated that, while resource capacity had a strong positive relationship with each Joe Alan Miller aspect of the administrative component, differences in resource capacity had very little effect on the organi- zation size-administration relationship. A moderately strong negative correlation was found between organization size and relative size of the direct administrative component among local market unions, but virtually no relationship existed between these variables among national market unions. However, among the national market unions, the physical density index was negatively correlated at a moderately high level with the relative size of the direct administrative component; among the local market unions, this correlation was also negative but very weak. These findings were interpreted as indicating that, for national market unions, the geographical extensiveness of the industrial environment makes the complexity in the administered system resulting from the way the membership is distributed among local unions a more important deter- minant of the size of the direct administrative component than organization size. Among local market unions, operating in an environment allowing considerable decentralization of administrative responsibilities to subnational levels, complexity is less of an administrative issue. The data also showed that local market unions are apparently able to utilize independently chartered and financed inter- mediate bodies to coordinate the activities of locals Joe Alan Miller operating in competing, though regionally limited, product market industries. The findings regarding the organization size- supportive component relationship were very similar in both the local market and national market unions. In each case there was very little correlation between organization size and relative size of the supportive administrative component, but scatter diagrams for these relationships suggested that organization size and the absolute size of the supportive component tend to increase at approximately similar rates. It was suggested that many supportive activities, in comparison to those of a direct administrative nature, are not easily decentralized or otherwise delegated to sub- national levels, regardless of differences in the indus— trial setting. The results of this study generally suggested that organization size and complexity have variable effects on the administrative component, according to (1) the way in which features of the environmental setting facilitate or constrain an organization's handling of the organiza- tion size—complexity issue, and (2) variations in the types of imperatives associated with different adminis- trative functions. r?— ‘A‘ _—‘————_—— ORGANIZATION SIZE AND ADMINISTRATION IN NATIONAL LABOR UNIONS: A COMPARATIVE ANALYSIS BY Joe Alan Miller A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Sociology 1969 ,1) ($1) L.) 2 / 5 3. ,/j ‘ /(:) ACKNOWLEDGMENTS The author is indebted to many individuals for assistance in developing and conducting this study. Special thanks go to Professor Philip Marcus, doctoral committee chairman, and to other members of this com- mittee--Professors William Form, John Howell, Donald Olmsted, and Christopher Sower. Gratitude is also expressed to several persons in the Institute of Labor and Industrial Relations, The University of Michigan-Wayne State University. Charles Rehmus and Ronald Haughton, co-directors of the Institute, were helpful in providing certain supportive expenses for travel and computer-processing of data. Louis Ferman, Institute research director, helped to clarify many ideas, though the final expression of these in the thesis is solely the author's reSponsibility. Claire Ben-Eliezer and Charles Betsey, members of the Institute staff, helped considerably in the collection of data, and William Murphy, also of the Institute, assisted in the preparation and submission of data for analysis by computer. The typing by Sharon Curtis is much appreciated, Particularly since this task often involved the careful translation of marginal notes. ii .___ q... __ _. .f‘, __.. _f A if ———_—— Finally, to my wife, Nancy, goes my deepest gratitude, for her understanding, encouragement, and unending patience. iii TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . . . . . viii LIST OF FIGURES . . . . . . . . . . . . . . . . . . x Chapter I. SUBJECT OF INVESTIGATION . . . . . . . . . . 1 General Objectives of the Study Rationale for the Selection of National Labor Unions for Study General Characteristics of the Study Specific Features of Union Organi— zation and Administration Conclusion II. A REVIEW OF LITERATURE . . . . . . . . . . . 31 A Review of Research Literature Conclusion Conceptual and Theoretical Implications of Previous Research Studies The Principle of Coordination and the Concept of the Administrative Component Organization Size, Complexity and Re- sources in Relation to the Adminis- trative Component Conclusion III. METHODOLOGY OF THE STUDY AND STATEMENT OF RELATIONSHIPS TO BE ANALYZED . . . . . . . . 75 Sources of Data iv f AAA III. Continued Page The Bureau of Labor Statistics Office of Labor-Management and Welfare— Pension Reports Union Constitutions Union Convention Proceedings, Officers' Reports and Union Periodicals Selection of Unions to be Studied Selection of a Potential Study Group Selection of the Final Study Group Key Characteristics and Limitations of the Final Study Group Concluding Comments about the National Labor Unions Selected for Study Data Collection Independent and Dependent Variables In- cluded in the Analysis Independent Variables Dependent Variables Examination of Relationships in the Analysis of Data The General Mode of Analysis Tests of Relationships Conclusion IV. PRESENTATION OF FINDINGS . . . . . . . . . . 124 Industry Type as a Control Variable and Its Relationship to the Administrative Component Industry Type and the Administrative Component Industry Type and Other Independent Variables in the Analysis Organization Size and the Administrative Component Organization Size and the Total Ad- ministrative Component V IV. Continued Page Organization Size and the Direct Administrative Component Organization Size and the Suppor- tive Administrative Component Conclusion Resource Capacity and the Organization Size—Administration Relationship Resource Capacity and the Adminis- trative Component The Effects of Resource Capacity on the Organization Size-Administra- tion Relationship Physical Density Characteristics and the Organization Size-Administration Rela- tionship Physical Density Characteristics and the Administrative Component The Effects of Physical Density on the Organization Size-Administration Relationship Conclusion V. SUMMARY AND CONCLUSION . . . . . . . . . . . 213 General Features of the Analytical Design A Summary of Findings Industry Type and the Administrative Component The Organization Size-Administration Relationship General Comments about the Study A Summary Comparison The Total Administrative Component The Direct Administrative Component The Supportive Administrative Com- ponent Conclusion vi w—AAA r—v—rfr‘ _-———‘ __.— vii LIST OF TABLES Table 1. .10. 11. RESULTS OF APPLYING INCLUSION—EXCLUSION CRITERIA IN SELECTING A POTENTIAL STUDY GROUP . . . . . . . . . . . . . . REASONS FOR EXCLUDING OTHER UNIONS FROM THE POTENTIAL STUDY GROUP . . . . . . . . . . . NATIONAL LABOR UNIONS INCLUDED OR EXCLUDED FROM THE FINAL STUDY GROUP, BY MEMBERSHIP SIZE AND INDUSTRY TYPE CLASSIFICATION . NUMBER AND PERCENT OF FULL-TIME NATIONAL OFFI— CERS AND EMPLOYEES FOR NATIONAL LABOR UNIONS IN THE FINAL STUDY GROUP . . . . . . . MEAN SIZES OF DIFFERENT INDEXES OF THE ADMIN- ISTRATIVE COMPONENT: BY SEPARATE INDUSTRY TYPE . . . . . . . . . . . . . . . . . . FREQUENCY OF UNION-SPONSORED MEMBERSHIP BENEFIT PROGRAMS, BY INDUSTRY TYPE FREQUENCY AND PERCENTAGE OF UNIONS IN THE FINAL STUDY GROUP, BY ORGANIZATION SIZE AND INDUSTRY TYPE . . . . . . . . . . . . MEAN SIZE OF PHYSICAL DENSITY INDEX, BY I NDUS TRY TYPE 0 O O C C C O C C O O I O O O C I . MEAN SIZE OF RESOURCE CAPACITY INDEX, BY INDUSTRY TYPE 0 O O I O O O I O O O O C C O O ZERO-ORDER CORRELATIONS BETWEEN ORGANIZATION SIZE AND THE ADMINISTRATIVE COMPONENT INDEXES: BY I NDUS TRY TYPE 0 C C 0 O I O O O O I O I O O ZERO-ORDER CORRELATIONS BETWEEN RESOURCE CAPACITY AND THE ADMINISTRATIVE COMPONENT INDEXES: BY INDUSTRY TYPE . . . . . . . . viii Page 85 87 9O 94 129 135 138 139 141 145 171 Table Page 12. ORGANIZATION SIZE AND RESOURCE CAPACITY: RAILROAD IINIONS ONLY 0 I I O O O O O O O O O O 173 13. PARTIAL CORRELATIONS BETWEEN ORGANIZATION SIZE AND INDEXES OF THE ADMINISTRATIVE COM— PONENT, CONTROLLING FOR RESOURCE CAPACITY: BY INDUSTRY TYPE . . . . . . . . . . . . . . . 174 14. ZERO-ORDER CORRELATIONS BETWEEN PHYSICAL DENSITY AND THE ADMINISTRATIVE COMPONENT INDEXES: BY INDUSTRY TYPE . . . . . . . . . . . 181 15. PARTIAL CORRELATIONS BETWEEN ORGANIZATION SIZE AND INDEXES OF THE ADMINISTRATIVE COM- PONENT, CONTROLLING FOR PHYSICAL DENSITY: BY INDUSTRY TYPE . . . . . . . . . . . . . . . 188 16. PARTIAL CORRELATIONS BETWEEN PHYSICAL DENSITY AND INDEXES OF THE ADMINISTRATIVE COMPONENT, CONTROLLING FOR ORGANIZATION SIZE: BY INDUSTRY TYPE . . . . . . . . . . . . . . . . . 189 17. MEAN SIZE OF THE PHYSICAL DENSITY INDEX FOR DIFFERENT CATEGORIES OF ORGANIZATION SIZE: LOCAL MARKET AND NATIONAL MARKET UNIONS ONLY 193 18. ORGANIZATION SIZE, INTERMEDIATE BODIES, AND THE DIRECT ADMINISTRATIVE COMPONENT INDEX: BY INDUSTRY TYPE . . . . . . . . . . . . . . . 197 19. ORGANIZATION SIZE, INTERMEDIATE BODIES, AND THE SUPPORTIVE ADMINISTRATIVE COMPONENT INDEX: BY INDUSTRY “PE 0 O O O O O O O O I C C C O O 207 20. INTERCORRELATIONS BETWEEN ORGANIZATION SIZE, PHYSICAL DENSITY AND RESOURCE CAPACITY . . . . 268 21. DISTRIBUTION OF UNIONS IN FINAL STUDY GROUP BY METHODS OF DETERMINING MEMBERSHIP SIZE . . . 301 ix LIST OF FIGURES Figure 1. 10. Covariation of organization size and the total administrative component index: Local market unions 0 O O O O I O O I O O O O O O O O Covariation of organization size and the direct administrative component index: Local market unions . . . . . . . . . . . . . . . Covariation of organization size and the supportive administrative component index: Local market unions . . . . . . . . . . . . . . Covariation of organization size and the total administrative component index: National market unions 0 O 0 O O O I O O O O O O O O O O Covariation of organization size and the direct administrative component index: Na- tional market unions . . . . . . . . . . . . Covariation of organization size and the supportive administrative component index: National market unions . . . . . . . . . . . . Covariation of organization size and the total administrative component index: Rail- road unions only . . . . . . . . . . . . . . . Covariation of organization size and the direct administrative component index: Rail- road unions only . . . . . . . . . . . . . . . Covariation of organization size and the supportive administrative component index: Railroad unions only . . . . . . . . . . . . . Covariation of the physical density index and the total administrative component index: meal market unions I O I C C O C O O O O O O O Page 269 270 271 272 273 274 275 276 277 278 Figure Page 11. Covariation of the physical density index and the direct administrative component index: Local market unions . . . . . . . . . . 279 12. Covariation of the physical density index and the supportive administrative component index: Local market unions . . . . . . . . . . 280 13. Covariation of the physical density index and the total administrative component index: National market unions . . . . . . . . . . . . 281 14. Covariation of the physical density index and the direct administrative component index: National market unions . . . . . . . . . 282 15. Covariation of the physical density index and the supportive administrative component index: National market unions . . . . . . . . . 283 16. Covariation of the physical density index and the total administrative component index: Railroad unions only . . . . . . . . . . . . . 284 17. Covariation of the physical density index and the direct administrative component index: Railroad unions only . . . . . . . . . . 285 18. Covariation of the physical density index and the supportive administrative component index: Railroad unions only . . . . . . . . . . 286 19. Covariation of resource capacity and the total administrative component index: Local market unions . . . . . . . . . . . . . . . . . 287 20. Covariation of resource capacity and the direct administrative component index: Local market unions 0 O O O O O O O O I O O O O O O O 288 21. Covariation of resource capaCity and the supportive administrative component index: Local market unions . . . . . . . . . . . . . . 289 22. Covariation of resource capacity and the total administrative component index: National market unions 0 O I O O O O O I O O O O O O O O 290 xi Figure Page 23. Covariation of resource capacity and the direct administrative component index: National market unions . . . . . . . . . . . . 291 24. Covariation of resource capacity and the supportive administrative component index: National market unions . . . . . . . . . . . . 292 25. Covariation of resource capacity and the total administrative component index: Railroad unions only . . . . . . . . . . . . . . . . . . 293 26. Covariation of resource capacity and the direct administrative component index: Rail- road unions only . . . . . . . . . . . . . . . 294 27. Covariation of resource capacity and the supportive administrative component index: Railroad unions only . . . . . . . . . . . . . 295 xii CHAPTER I SUBJECT OF INVESTIGATION It is widely assumed that large organizations tend to be overbureaucratized, that is, that an increase in organizational size is accompanied by a diSproportionate increase in administrative overhead; but the evidence does not support this assumption.1 Although everyone agrees that administrative per- sonnel are essential for a bureaucratic organiza- tion, the proportion of personnel who perform administrative tasks is in doubt. From a variety of organizational settings, positive, negative, and nonexistent relationships between organiza- tional size and relative size of the administra- tive component have been reported, leaving the question for organizations in general unanswered. 2 These Opening statements, both relatively recent, point up the uncertain condition of our knowledge in an important area of organizational study, namely, the rela- tionship between organization size and relative size of the administrative component in formal organizations. The popular belief alluded to in the statement by Blau and Scott stems largely from the work of C. N. Parkinson, who set forth "Parkinson's law" stipulating 1Peter M. Blau and W. Richard Scott, Formal Or- ganizations (San Francisco: Chandler Publishing’COmpany, 1962), p. 226. 2William A. Rushing, "Two Patterns of Industrial Administration," Human Or anization, 26 (Spring/Summer, 1967), 32. that increases in the size of an organization result in relatively greater increases in the size of its adminis- trative component.3 Research evidence in support of the Parkinsonian thesis is, however, quite scarce. The cumulative evidence of research studies dealing with the 4 has cast organization size—administration relationship doubt on its validity as a general prOposition. This negative comment on the general applicability of Parkinson's "law" does not imply that there has been a corresponding development of alternative prOpositions broadly applicable to organizations of varying type. This has not been the case, as both opening statements indicate. Summarily stated, our present knowledge concerning the effects of increasing organization size on the relative size of the administrative component is best described as generally nascent in character. It is possible to state some broad generalizations on the basis of previous studies, but these must be cast in a tentative mold; none can be regarded as definitive. 3C. NOrthcote Parkinson, Parkinson's Law and Other Studies in Administration (New York: Ballantine Books, 1964 edition), pp. 15-27. 4The phrase "organization size-administration relationship" is used here and at later points to refer to the relationship between organization size and rela- tive size of the administrative component. This short designation, which has some currency in recent literature, affords considerable convenience in expression as well as relief from needless redundancy in textual presentation. The rather rudimentary state of development in this area of study is probably attributable in part to the rela- tive recency of much of the research dealing centrally with the organization size-administration relationship and the fairly narrow range of organization types studied. The major portion of the research literature has resulted from studies conducted in the last decade and a half, mostly since 1960. The research literature is not voluminous, and the accumulation of empirical data from a broad range of organizational settings, a necessity for elaborate theoretical development, has been relatively slow-paced. As a result, continuity in empirical and conceptual inter- ests has only slowly emerged. In the study of organization size and administration, an evident need exists for additional research which ex- pands the boundaries of organization types studied while retaining a conceptual and empirical relatedness to pre- vious studies. The present study proceeds within the framework of such an objective. General Objectives of_the Study In order to place the general objectives of the current study into perspective, it is necessary first to take note of two major trends in research on the organizar tion size-administration relationship. (1) Although the pace of development has been relatively slow, research is being conducted in an expanding variety of organizational settings in an apparent attempt to identify hypotheses and theoretical principles having broad application. (2) More attention is being paid to the possible analytical sig- nificance of other organizational characteristics which may both condition the effects Of increasing organizational size and act as independent influences on quantitative as- pects of the administrative component. Generally speaking, there is a tendency to expand the number of empirical issues being investigated under the general rubric of research interest in organization size and administration. At present, much of the research along these dif- ferent lines must be regarded as largely exploratory. This study seeks to contribute both to our knowledge concerning the basic relationship between organization size and rela- tive size of the administrative component and to the ex- panding research base associated with the above-noted points. It was stated earlier that the study of organization size and administration has not yet emerged as a well- deveIOped area of analysis, from the standpoint of either conceptual usages or the statement of research hypotheses. The general principle underlying the conceptualization and statement of research questions has been that, as the size of an organization increases through the addition of new members, its administrative structure must be adjusted to accommodate correlative increases in demand for coor- dination, control, and communication in order to maintain internal stability and acceptable levels of goal-achievement. As a result, there is a corresponding increase in the size of the administrative component, which is the key coordinar tive unit for the total organization.5 From the standpoint of research interest in the organization size-administration relationship, a major issue has been whether this "corresponding increase" in the size of the administrative component is proportion- ately greater, generally the same, or less, in comparison to the rate of increase in organization size. In effect, these alternative possibilities have provided implicit hypotheses for research, but while the number of studies dealing centrally with the organization size-administration relationship has been increasing, this central issue re- mains clouded, as the opening quotation from Rushing has indicated. 5The assumption that the primary function of ad! ministrative personnel is that of coordination is one commonly made in research studies on organization size and administration. It does not mean, however, that each unit in the administrative component is directly involved in supervisory tasks; to the contrary, coor- dination is typically achieved through specialization of functions, so that administrative positions move through a range of functions, involving the exercise of executive control, technical expertise, and clerical routines, the last of which facilitate coordination in the sense of information processing. In this study the organization size-administration relationship is examined with respect to a particular class of organizations-~national and international labor unions in the United States.6 Interest is centered pri- marily on national labor unions whose members work in the private sector of the economy. In designing the study, it was decided that this narrowing of focus would provide a certain functional homogeneity, though at a general level, within the class of organizations studied. In general terms the analysis concentrates on ads ministrative characteristics of the national organization7 in each union studied, in relation to differences in memr bership size and other selected organizational attributes. The national organization of each union is treated as the key administrative unit for the total union--a position which corresponds to information derivable from general commentaries on American unionism and union literature 6The descriptive terms "national" and "international" are, for all practical purposes, synonymous. The distinc- tion depends on whether the parent union has local unions in Canada as well as in the United States. As a matter of convenience, the single term "national labor union" is used hereinafte r . 7The term "national organization" is used here and afterwards in a restrictive sense. To avoid confusion, it is best to clarify its intended meaning at an early point. In this study it refers to the national officers and ap- pointed staff who participate in the day-to-day adminis- trative Operations of the national labor union, rather than to the national union as represented by its total member- ship or assembled convention body. itself.8 Units below the national level of organization are viewed as subsystems toward which the administrative functions of the national organization are primarily directed. Thus, the national labor union is conceptualized as a complex social system consisting not only of individual members but also of distinguishable subsystems toward which and through which administrative functions are per- formed on behalf of the membership. This conceptualization follows that offered by Ramsoy, who uses the concept of complex social system "to refer to the problematic rela- tionship between an inclusive system and its subsystems."9 The usefulness of this approach rests primarily in the 8A consistent theme in works dealing with union government and administration is the importance of the national organization as the key governmental and adminis- trative unit. Extensive documentation of this hardly seems necessary, but three generally representative works are particularly illustrative. For an early treatment of this theme, see Theodore W. Glocker, The.§9vernment.2§ .Irade,flnign§ (Baltimore: Johns Hopkins Press, 1913). Two more recent works are: Philip Taft,.2h§ Structure .agg Government_g£ Labor Unions (Cambridge: Harvard University Press, 1954); and William H. Leiserson, Trade Union Democracy (New York: Columbia University Press, 1959). In union literature, it is not uncommon to find references to the national organization as the administra- tive center of the total union; for example, the national office as the "nerve center" of the national union. The primary importance of the national level of organization is also attested to in the opening sections of most national union constitutions. 9Odd Ramsoy, Social Groups,g§ System and Subsystem (New York: The Free Press of Glencoe, 1963), p. 11. attention it directs toward the nature of subsystem dif- ferentiation in formal organizations, since, from the standpoint of the administrative function of coordination, differentiated subsystems into which individual members are aggregated for purposes of integrated activity are key organizational units. Within the framework of this basic analytical orientation, three general research objectives are pursued in this study. 1. Major emphasis is placed on analyzing the basic form of the organization size-administration relationship in the national labor unions studied. This particular phase of the investigation has two distinguishable facets: On the one hand, attention is focused on the relative Size of the officer and staff component in the national organi- zation in relation to differences in membership size. On the other hand, an examination is made of the relative rates of increase or decrease in the sizes of functionally differentiated parts of this total administrative component in relation to changing membership size. With respect to this latter analytical emphasis, the total administrative component is regarded as divisible into two different aspects: (a) a direct administrative component--consisting of national officers and appointed staff whose tasks variously involve the design of programs to carry out policies agreed upon in national conventions, formulation of policy, and provision of direct service functions to, and on behalf of, affiliated local unions; and (b) a supportive administrative component--consisting of national staff members whose positions involve primarily indirect, or supportive, administrative functions. 2. Emphasis is also placed on analyzing the effects of other organizational attributes on the organization size-administration relationship and on the relative size of the total administrative component and its functionally differentiated parts. It should be stated that at present there is no overall consensus as to what other organizar tional variables might profitably be incorporated into such an analysis. With the possible exception of organizational complexity, it cannot be said that any additional variable has received intensive or systematic attention. In the present case, the selection of additional independent variables for analysis was influenced both by practical considerations concerning the availability of reliable information on a particular organizational characteristic and the presumed analytical importance of the characteris- tic from the standpoint of its conceptual relatedness to the organization size-administration issue. G} Additional independent variables included in the analysis relate to the following organizational attributes of the unions studied: (a) The number and relative size (i.e., physical density characteristics) of affiliated local unions, as an indication of the degree of structural 10 complexity resulting from the differentiation of the union's membership into administered subsystems; (b) the financial resource capacity of the union, as an indication of its ability to develop the administrative apparatus in the national organization; and (c) industry type, as an indication of environmental sources of influence on administrative strategies. In addition to the above-mentioned four independent variables, two others are included in the analysis, and though formulated as independent variables, are not con- sidered as major analytical characteristics. They serve primarily to clarify and elaborate interpretations of re- lationships between the major independent and dependent variables. They are: (a) union-sponsored membership benefit prOgrams, other than the commonly found strike or unemployment benefits programs, which are presumably implicated in the comparative rates of growth in the supportive administrative component; and (b) the presence or absence of intermediate bodies standing midway between the national and local levels of organization, but financed and administered separately from each, which perform im- portant administrative functions in the national union. With respect to the latter, interest is expressed only in those intermediate bodies which have delegated functions in the national union for the coordination of affiliated locals' activities in the key areas of collective bargaining, 11 contract enforcement, grievance settlement and/or strike control. A more complete exposition of these variables and the importance of each in the analytical design of the study will be offered at a later point. 3. The analytical design of the study is largely a descriptive one, including a number of exploratory interests within this purview, in that the empirical investigation is confined to comparisons of relationships within a single class of organizations. It can be noted, though, that the conceptual bases for the statement of variables and speci- fication of relationships to be examined are derived from previous research and commentary in the selected area of study. In general, then, rather than test refined hy- potheses regarding organization size and relative size of :5 the administrative component, the study seeks to contribute to the developing fund of knowledge in this area of analy- sis. Reflections on the current state of theoretical development relative to organization size and administra- tion suggested the profitability of designing the study in this manner.10 10In a recent essay on the comparative study of or- ganizations, Udy asserts that in view of the current state of development in organization theory, exploratory analy- sis probably offers, on balance, greater benefits than hypothesis testing in comparative organizational analysis. He cautions, however, against an "anything goes" explora- tory method, in that such an approach contributes little in methodological or conceptual continuity to the compara- tive study of organizations. See Stanley H. Udy, "The Comparative Analysis of Organizations," Handbook of 12 It can also be noted that the study is not intended to be a complete exposition of administration in national labor unions. Rather, it proceeds by selecting a set of relatively few--though presumably important--variables and analyzing their relationships within the particular setting of labor union organization. Rationale fog;the Selection of National Labor Unions for Study The choice of national labor unions for study was not fortuitous, although the availability of data on a considerable number of such organizations, thus permitting the use of a comparative framework, was not the least of all reasons. Blau has recently commented on the general value of the comparative approach, noting that while it becomes necessary to ignore many individual organizational peculiarities by sacrificing detail for breadth of cover- age, the theoretical benefits usually outweigh this loss.11 There are, however, more strategic reasons for accounting this selection a valuable one from the standr point of interest in organization size and the administrar tive component in formal organizations. In the following discussion, these are presented by grouping the several Or anizations, ed. James G. March (Chicago: Rand-McNally Book Co., 1965), pp. 678-709. 11Peter M. Blau, "The Comparative Study of Organi- zations," Industrial and Labor Relations Review, 18 (April 1965), 333 et passim. l3 illustrative points into two categories--one dealing with the general contributory value of the study design itself and the other with specific features of national labor unions as formal organizations. General Characteristics of the Study There are essentially two features of the present study which enhance its potential in contributing to em- pirical and theoretical interests in the organization size- administration relationship. 1. The national labor unions included in the study sample extend across a broad range of membership sizes, from unions having only a few thousand members to several having more than one-half million members. This feature of the study enables an analysis of relationships under widely varying conditions of organization size. 2. The degree to which it is possible to arrive at generalizations regarding the organization size-administrar tion relationship that are applicable to a range of organi- zation types poses a major question in this area of study. The national labor union represents a different class of organizations than studied thus far. It is true that some studies of organization size and administration have in- cluded local unions in their study samples, but to date none has focused on the national level of organization. Thus, the present study offers an opportunity to generate 14 research findings for comparison with those from studies conducted in other organizational settings. Specific Features of Union Organization ‘ and Administration The distinctiveness of the national labor union as an organization type, as alluded to in the second point above, is not attributable simply to differences associated with a nominal class identity--for example, the labor union as opposed to the business organization--although, pre- sumably, such identities do imply some homogeneity of function within the nominal class. More theoretically meaningful distinctions establishing the national labor union as a promising unit for study are derivable from careful consideration of a number of its essential charac- teristics as an organizational form. These are treated generally in the following points; more specific elabora- tions of their analytical significance will emerge at later points in the detailed analyses of data. 1. In labor unions the formal flow of authority is from the membership to the elected leaders. In principle, the controlling units in a national union are the constitu- ent locals, which send delegates to the national conven- tion-~held, usually, every two to four years, depending on the particular union. In the formal structure of the national union, the convention is the lawmaking body, passing on policies and directives which elected officials \'I 5". \ D t 15 are expected to carry out between conventions. However, it is generally true, as Miller and Form note, that "most unionists recognize that considerable discretion must be given to the officers if they are to have a permanent and effective organization."12 At a minimum, the formal principle of having an upward flow of authority and an official hierarchy built from the bottom up by pOpular election is an organizational characteristic distinguishing the labor union from other organizational types more closely approximating the imperatively-controlled organizational form envisioned in the traditional bureaucratic model. More important, however, are the potential implications of this feature of union organization for administrative operations. In labor unions administrative processes are in- fluenced not only by impersonal criteria of operating efficiency--although these cannot be ignored in any formal organization--but by the internal political necessities of mobilizing and maintaining members' support for union programs and activities.13 Lester notes, for example, that 12Delbert C. Miller and William H. Form, Industrial Sociology; The Sociology of Work Organizations (New York: Harper and Row, 1964),pp. 338-339. 13It should be noted that the necessity for dealing with such political undertows is an administrative problem in all membership-service organizations that are classi- fiable under the general label of voluntary association. For a good case analysis of this phenomenon in one reli- gious denomination, the American Baptist Convention, see Paul M. Harrison, Authority and Power in the Free Church Eradiyion (Princeton, N.J.: Princeton University Press, 16 while there are many influences operating on national labor unions today which promote greater preoccupation with rational and efficient administration, labor unions are still effectively viewed as political organizations pur- suing economic goals.14 In a recent statement Barbash has commented on the slowness with which bureaucratization, in its classical guise, has penetrated the labor movement: The political strain so permeates the fabric of union government that it has undoubtedly stunted the growth of the kind of classical bureaucracy which derives its stnategic position from 'specé ialization, a hierarchy of authority, a system of rules, and impersonality.‘ Its closest ap- proximations to bureaucracy are the two categories 1959). Whether or not the labor union is appropriately classified as a voluntary association is a matter of some debate. Yet, it seems correct to assert that its organi- zational form most closely approximates that typically classified under this label. It may be, though, that the various constraints placed on union administration in the United States by labor legislation have limited the degree to which internal political processes can freely influence administrative structures and strategies. 14Richard A. Lester, As Unions Mature: An Analysis of the Evolution of American Unionism (Princeton, N.J.: Princeton University Press, 1958), p. 23. Lester's com- ments are generally representative of many which view the political aspects of union organization as a basic distin- guishing characteristic in comparison to business organi- zations. For example, Ross notes the following: "The union is not a business enterprise selling labor. It is a political institution representing the sellers of labor, and there is no necessary reason to assume that it will automatically or mechanically behave in the same fashion as a profit-maximizing business enterprise." Arthur M. IRoss, "The Trade Union as a Wage-Fixing Institution," .Arsrigan Economig Reyiew, 38 (September, 1947), 568. I‘ I. 17 of union staff--the 'servicing' type of union representative, and the professional or tech- nical staff. This failure of a classical bureaucracy to develop is due, in the main, to the failure of an objective science of union administration to gain acceptance. (In striking contrast is the evolution of scientific management, which has been a major force for bureaucracy on the enterprise side.) Every issue in collective bargaining and in internal union government cuts across vital interests, and is therefore infused with a political essence which defies the objectivity of bureaucracy. Barbash's statement, of course, plays heavily on Weber's assertion that bureaucratic organization, in its {1”} purest sense, cannot develop fully when the governing of—‘Tg ficials are elected rather than appointed.16 It should be noted, however, that elected officials in formal organiza- tions--especially, perhaps, if elected to full-time positions--do have several sources of power; for example, the specialized knowledge growing out of experience and control over communications channels. It can be safely assumed that these are at least partly translatable into administrative control. Such an assumption is a key element in Michels' exposition of oligarchy17 and, 15Jack Barbash, Amgrigan Unions; Strugyure, Governmgnt,I ang Politics (New York: Random House, 1967), p. 141. 16Max Weber, From Max Weber: Essays in Sociology ed. H. H. Gerth and C. Wright Mills (New York: Oxford University Press, 1958), pp. 200-201. 17Robert Michels, Political Parties (Glencoe, 111.: {The Free Press, 1949 edition). —.£\’a 18 incidentally, in Weber's conceptualization of "bureaucratic control."18 It should also be noted that Barbash's statement assumes that the most compelling influences for adminis- trative formalization are contained within the internal structure of the organization, an assumption that is only partially correct. The primary function of administration involves not only the maintenance of internal stability andvflj:~ cohesion, but also the promotion of effective relations between the organization and its relevant environments. Although the foregoing advises caution against presuming a thoroughgoing bureaucratic tone in union or- ganization, it should not be taken as implying that the totality of administration in labor unions is necessarily a product either of whim or continuous political compromise. The debilitating effects of such administrative foundations would be such as to jeopardize the union's survival--or, for that matter, the survival of any formal organization-- much less its effectiveness in relation to employers and other unions. It does indicate, however, that there are influences operating in unionism generally which may temper bureaucratic tendencies. From the standpoint of operating administration, which is the central interest of this study, it suggests that decisions of union leaders‘figf' involving the allocation of resources for administrative 18Max Weber, The Theor of Social and Economic Or- ganization, trans. and ed., Talcott Parsons (New York: The IFree Press of Glencoe, 1964 edition), pp. 338-339. 19 purposes and the deployment of administrative staff are influenced both by internal political considerations and by criteria of rational efficiency. 2. Another important feature of union organization is the dependence of the elected leadership on membership contributions for the support of administrative and program Operations. Although national labor unions usually have alternative sources of income through investment programs, contributions by members in the form of per capita tax payments and regular or Special assessments are their primary sources of income. Increases in per capita tax rates or assessments are usually voted on by members' representatives in national conventions, although in some cases referendum voting procedures are used when issues involve financial matters. But, as numerous commentators on union organization have noted, union members are gen- erally penurious in granting such increases.19 This particular feature of union organization un— doubtedly affects the degree to which bureaucratic norms of technical efficiency govern administrative developments. Unlike business corporations, the labor union, generally Speaking, does not have a range of alternative means of raising capital to support its Operations. Its scale of 20It is the author's observation, based on many hours Spent reading national convention proceedings, that the lengthiest and most acrimonious debates on the conven- tion floor are usually those involving increases in per capita tax rates or the levying of Special assessments. 20 operations is dependent on the amount of resources it commands, which in turn is closely tied to the Size of its membership, the members' satisfaction with the performance of elected Officials, and, at least to some extent, the wage structure of the industry in which the union is located.20 Summarily stated, there are three key implications of this aspect of resource procurement in labor unions for adminis- trative operations. First, in the national labor union considerable emphasis must be placed on maintaining and, if possible, increasing the size of the membership because of the nexus ‘:¢ 21 This of organizational stability and resource level. implies that, as a matter of technical rationality, primary attention will be paid to the deployment of staff in posi- tions most directly related to membership maintenance or augmentation. In general, these are the positions that appear in the national organizations of labor unions under the various labels of international representative, field representative, or organizer. 20This, of course, is a characteristic shared with other membership-service-type organizations. It may well be, though, that some labor unions are in a more advanta- geous position than most other membership-service organi- zations, principally because of union security clauses in collectively bargained contracts providing for mandatory union membership within a given time period after initial employment and dues checkoff systems. 21For Weber, constancy of income was a necessary condition for the development of a stable and formalized, or bureaucratic, administrative structure. See Weber, From Max Weber: Essays in Sociology, pp, cit., p. 208. 21 Second, it might also be expected that, as a matter of political astuteness, in the development of the adminis- trative apparatus of the national organization, major con- sideration will be given to the development of those staff positions which are most visible to the membership, inas- much as affirmative votes on recommended increases in members' regular contributions relate closely to the sen- timents of the membership toward the elected leadership. By and large, it is the servicing type of representative mentioned in the preceding point which has this charac- teristic of visibility.22 There is a third obvious but important implication of resource procurement processes in labor unions for administrative structure, namely, that the level of re- sources available to the national organization places <33 limits on the size of its staff component. In brief, the complement of personnel and the specialization of their functions that might be considered desirable by union leaders on grounds of efficiency, will be constrained by the amount of resources available for these sorts of ads ministrative elaborations. 22Generally Speaking, the rationale presented to convention delegates by officers for increases in per capita tax rates is largely based on intended increases in the number of field-servicing staff and organizers. The author's observation, based on extensive reading of con- vention proceedings, is also that the granting of such increases is frequently tied to policy directives by the delegates that the new monies are to be used primarily for these direct administrative purposes. 22 The foregoing points make it clear that both tech- nical and political aspects of resource procurement in the labor union may combine to effect particular colorations in the development of the staff component. Presumably, then, the resource capacity of the union should be an im- portant analytical characteristic in understanding par- ticular aspects of its administrative development. 3. Andther distinctive feature of national labor unions is related to the fact that, from the standpoint of their key centers of administration, they are rather simply structured. The primary structural units are the national 23 Not organization and the constituent local unions. infrequently, there may be intermediate organizational units located midway between the national and local levels of organization, but these are not found universally in national labor unions.24 However this may be, in general terms it can be said that the national labor union dis- plays an organizational structure of rather flat shape. #3 23There are a few national labor unions which do not charter local unions as administrative units but they are a distinct minority. Only national labor unions having administratively separate locals are considered in this study. 24Some intermediate bodies are financed by per capita taxes received from affiliated local unions; others are supported by the national organization and are, in effect, extensions of the executive arm of the national organization. In general, intermediate bodies of the former type enjoy greater independence of action in re- lation to the national organization. 23 The relatively noncomplex structural configuration in the national labor union-~as contrasted with formal organizations having deeper hierarchical structures--is decidedly advantageous for one of the major analytical emphases of the study. In recent studies of organization size and administration, there is detectable a growing emphasis on the implications of organizational complexity c; for the organization size-administration relationship, with the complexity variable treated variously in terms of subsystem differentiation or role specialization with- in the organization system(s) being analyzed. From the vantage point of the national organization as the administrating unit, the key administered units in a national union are the local unions. At any point in time, the number of locals and their relative membership sizes are results of the aggregation of the organization's membership into differentiated subsystems arranged along AL) lateral, rather than hierarchical, dimensions.25 The resulting configuration provides an important and rela— tively uncomplicated means of determining the degree of 25This does not mean that the union controls the differentiation process, so that the number and relative sizes of local unions are completely matters of policy determination. The union has greater or lesser control over this process, depending largely upon the local union's geographic base--plant, community, or larger geographical area--but usually the union's Options are quite limited. From one standpoint, though, this is a question of little moment, since it is the end product of the differentiation process that is important, not the underlying reasons for a particular configuration. 24 structural complexity within the total administered system. 4. In the national labor union, the relationship between the national and local levels of organization de- parts somewhat from that commonly encountered in organizar tions having subunits which are fiscally and administra- tively under the executive control of a central organizar tion. Although subordinate in important ways to the national organization, local unions must be viewed as administratively separate units of organization having varying degrees of autonomy, and it is against this back- drop that the administrative characteristics of national labor unions must be studied. There are wide areas of activity in which the fiscal and administrative autonomy of local unions are formally assured by constitutional provisions. This point must not be overdrawn, however, for there are several mechanisms available to the national organiza- tion which enable it to exercise greater or lesser degrees of administrative control over activities in local unions. For example, although fiscal control of locals by the nar tional organization is largely confined to obligations for payment of per capita taxes and other assessments by the local, control over strike and emergency funds is typically 191aced in the hands of national officials. Under certain <3onditions, the national organization may exercise trustee- ship over a local's administration of its affairs. There 25 may also be provisions in the national constitution of the union permitting the national organization to review and sanction decisions made at the local level, although this varies among unions, and no single statement would be applicable without qualification. Finally, it should also be noted that the administrative scope of the national organization, both in terms of its broad knowledge of in- dustry developments and its complement of staff experts, contributes in no small way to its influence over locals' activities. 5. Partly as an extension of the previous point, mention should also be made of what is generally regarded as a major institutional trend in modern unionism, namely, the centralization of control at the national level of organization. In a sense the more recent developments in this direction may simply represent a continuation and perhaps acceleration of a longer-term trend.26 Typically, this centripetal tendency is seen as emanating from a combination of influences. The following quotation from Tannenbaum is generally representative of several commen- taries on this phenomenon. The centralization of control in national labor unions and the power of the leadership can be understood in large measure in terms Of the requisites of collective bargaining objectives and the political and economic environment within which these objectives must be pursued. Growing centralization in industry, industrywide 26Glocker,.gp. cit. 26 bargaining, government intervention, and the need for union experts, have all militated in the direction of more bureaucratic and centralized union organization.27 Administrative patterns do vary among labor unions, however, and conclusions regarding the centralizing ten- dencies in modern unionism must be placed into proper perspective. Generalizations relative to the pervasive— ness and scope of centralization can be qualified by taking note of two related aspects of administration in labor union organization. First, in the national labor union centralization of control is not necessarily tantamount to direct super- vision of activities at subnational levels by a represent- ative of the national organization. The deployment of personnel is partly a technical problem involving the efficient use of available resources, regardless of the criteria utilized in adjudging efficiency in any particular setting. Organizational control, as a processual aspect of the administrative function of coordination, may be achievable by means other than continuous or direct scrutiny of subsystem activities by an administrative functionary. Simon has noted that there are two very dif- ferent aspects of administrative centralization: 27Arnold S. Tannenbaum, "Unions," Handbook of Or- ganizations, ed. James G. March (Chicago: Rand-MONally Book Co., 1965), p. 752. References to the centralizing trend in unionism can be found in most works dealing with union structure and government. 27 On the one hand, decision-making powers may be centralized by using general rules to limit the discretion of the subordinate. On the other hand, decision-making powers may be centralized by taking out of the hands of the subordinate the actual decision-making func- tion.2 At this point, it is important only to note that, in discussing the so-called "bureaucratic drift" in union organization, a distinction should be maintained between the decision-making and decision-implementing dimensions of this trend. In some unions, organizational control, in the sense of prescribed authority to establish standards for, and exercise review of, decisions made at subnational levels of organization, may reside in the national offices, while the pattern of operating administration for decision- implementation may be largely decentralized. Second, there is general agreement among students of the labor movement that an important influence on the degree of administrative centralization in the national labor union is the character of the product market in the industry (or industries) where jurisdiction is exercised.29 28Herbert A. Simon, Administrative Behavior (New York: The Free Press, 1957), p. 234. 29See, for example, the discussion in Marten Estey, Thg Unions; Stgugyure, Deyglopment and Managgmgnt (New York: Harcourt, Brace and World, Inc.), especially pp. 61- 65; and, Walter Galenson, A Primer on Employment and nges (New York: Vintage Books, 1964), pp. 50-61. Also, this theme is explored at various points in several of the papers contained in Arnold R. Weber, (ed.), The Structure of Collective Bargaining: Problems and Perspectives (New York: The Free Press of Glencoe, 1961). 28 Generally speaking, the geOgraphic scope of the product market is viewed as an environmental condition effecting interdependence among local unions in terms of the union goals of wage improvement and the elimination of wage differentials. In unions operating in industries where the product market is predominantly local or regionally limited in character, it is administratively feasible--and perhaps desirable, from the standpoint of efficiency--to permit considerable autonomy in locals' collective bargaining actions, in that the effective area Of wage comparison is relatively narrow in geographic scope. This does not mean, however, that the national organization does not monitor locals' actions. Such participation may be by indirection--for example, as noted in the previous point, by the setting of collective bargaining standards or by reviewing and sanctioning locally-negotiated contracts. On the other hand, in unions located in industries where the product market is predominantly national or broadly geOgraphic in scope, some form of administrative centralization may be necessary to protect against the debilitating effects of wage competition among affiliated locals. In this latter instance, the national organization is the administrative unit most capable, both from the standpoint of resources and administrative scope, to effect the necessary coordination of locals' bargaining objectives and activities. 29 In summary, the foregoing discussion has attempted to illustrate the potential contributions of this study to theoretical and empirical interests in the organization size-administration relationship. This has been done by alluding to the pragmatic values associated with the general study design and by concentrating on the distinc- tive qualities of the national labor union as an organi- zational form. In a general sense, it can be stated that the study affords a considerable number of Opportunities to compare findings generated by research in other or- ganizational settings with those from one of somewhat different dimensions. Conclusion This opening chapter has been primarily concerned with providing a general orientation to the research objec- tives and empirical focus of the study. Consequently, the problem area and chief concerns of the study have been sketched only in broad outline. In subsequent chapters detailed consideration is given to the conceptual, method-~ ological and analytical components of the study. Chapter II is concerned with a review of previous studies of organization Size and administration; in addi- tion, attention is also given to a number of theoretical considerations pertaining to the emergent results of these earlier studies. 30 Chapter III discusses the methodology of the study, with particular attention paid to problems of data reli- ability in connection with the usage of secondary data. In this chapter the several independent and dependent variables that are analyzed in the study are introduced, with emphasis placed on demonstrating their conceptual relatedness to other studies of the organization size- administration relationship and their adequacy in reflec- ting important aspects of union organization and adminis- tration. Chapter IV is devoted to data analysis. The ob- served relationships are interpreted both in terms of the particular setting of union organization and general theoretical interests in the study of organization Size and administration. Chapter V presents a summary of findings and dis- cusses the implications of this study for further concep- tual and empirical work in this general area of organiza- tional analysis. CHAPTER II A REVIEW OF LITERATURE As an integral part of this study, a review was made both of research studies that have focused attention on the organization size-administration relationship and other commentaries that are related, directly and indirec- tly, to conceptual concerns in this area of organizational analysis. This review provides the basis for the discus- sion in this chapter. The materials which follow are organized into two major sections. The first deals with the research liter- ature itself and recounts the findings of these earlier studies, though without making extensive comparisons among them. The second deals with a number of theoret- ical considerations about the organization size-adminis- tration relationship, basing the discussion on an inter- play of information derived from the empirical results of the research studies and more general commentaries relevant to the study of organization size and adminis- tration. 31 32 A Review of Research Literature Lack Of conceptual and empirical continuity in research studies dealing centrally with the organization size-administration relationship makes systematic organi- zation of a review discussion somewhat difficult. It is helpful, though, to note two essential features of the literature in this area of study. (1) The greatest amount of research has been conducted in the setting of business and industrial organization, and some systematization is achieved by grouping these studies for review purposes; as for the remainder, there is no overall convenient categorization. (2) Some order can also be achieved by delimiting three varying emphases that run throughout the studies: the relationship between organization size and relative size of the administrative component, with no additional independent variables considered and no dif- ferentiation made between different groups or parts comprising the administrative component; the effects of increasing organization size on differentiated groups or parts comprising the administrative component; and the effects of other organizational characteristics, considered as independent variables, on the relative size of the ad- ministrative component. Of the studies reviewed, the earliest 50 appear was that conducted by Melman, who analyzed changes in the administrative overhead of American manufacturing firms 33 between the years 1899-1947.1 His investigation revealed an inverse relationship between size of firm, as measured by number of production personnel, and proportion of ad- ministrative personnel. Melman did not apply statistical analysis to his data, which are presented in a series of scatter diagrams. Thus, a precise statement on the strength Of the relationship between these two variables is not possible, but inspection of the scatter diagrams would appear to indicate that the magnitude of the corre- lation is very low.2 Using information drawn from German industrial experience between 1907 and 1933, Bendix showed that for concerns having six employees or more, the percentage of salaried administrative workers, in relation to all em- ployees, declines with increasing size of establishment, while the comparable percentage of salaried technicians rises. Since the former is the larger of the two groups, "the overall tendency remains for bureaucratization to be liighest in the smaller firms."3 ~ 1Seymour Melman, "The Rise of Administrative ()xrerhead in the Manufacturing Industries of the United £31tates, 1899-1947," Oxford Economic Pa ers, 3 (February, 1951), 62-112. 2Ibid., 100. 3Reinhard Bendix, Work and Authority in Industry (New York: John Wiley and Sons, 1956), p. 222. 34 In a longitudinal study of growth patterns in four industrial firms, Haire presents findings which suggest that the relative size of that part of the administrative component represented by supervisory and managerial per- sonnel is inversely related to size of organization, indicated by number of employees in the firm.4 He also presents some additional information regarding rates of increase for different types Of administrative personnel in relation to organization growth rates. These latter findings can be summarized in two points. 1. As the firm grows, the ratio of "staff" per- sonnel (i.e., employees who "provide specialized support, advice and help") to "line" personnel (i.e., those directly engaged in the making and selling of a product) increases rapidly in the earlier years and then tapers off to a relative stable figure illustrating parallel growth rates in the two groups.6 The relevance of this finding to the present study is indirect, in that Haire's discussion of Just what constitutes the "line" and "staff" components :in relation to the total administrative component is not (Blear. The instructive note for present research objec- txives would seem to be an indication in Haire's data that ¥ 4Mason Haire, "Biological Models and Empirical Histories of the Growth of Organizations," Modern Organis- 2=£1tion Theor , ed, Mason Haire (New York: John Wiley and SOns, 1959), pp. 296-297. 5Ipid., pp. 288-289. 35 size of organization is related both to the need for, and the resource capacity to support, specialized functions not directly involved in the production activities of the organization. 2. Haire concludes that the part of the adminis- trative component represented by clerical staff increases as company size increases at a generally proportionate rate, and it is only as a company ages that the tendency to acquire a larger percentage of clerks appears. He re- gards "the growth of clerks not as a function of size, but as a function of time"6--but, of course, both size of or- ganization and size of the clerical component are related to age. The implication of Haire's analysis is that the information processing needs of the organization--which are supplied primarily through clerical routines--increase proportionately with organization growth but tend to grow at a slightly accelerated rate as the organization ages. 6Ibid., Pp. 297-298. 7Inspection of the several line graphs presented 13y Haire (Ibid., pp. 298-301) to depict growth of the (:lerical indicate that, while the size of the clerical (momponent does indeed increase with both company size and Elgs, there is a somewhat erratic pattern to the lines, with many rises and dips. It is interesting to speculate on these variations. It may be that clerical personnel are 13<3th necessities and luxuries in any organization; that is, clerical tasks do exist in all organizations and must t><3nent rises above a level of some "minimum need" may be r«Ealated to the amount of "slack" resources (i.e., resources not immediately necessary for direct production activities) 1Itlat can be diverted to this area of indirect administration. “Ptlis is, of course, an untested speculation. 36 A recent attempt to deal with the organization size-administration relationship in industrial organizations is found in a cross-sectional study conducted by Rushing. Using data concerning the proportion of administrative personnel in relation to total number of production emr ployees in industrial firms, Rushing demonstrated that increases in average size of establishment are differen- tially related to increases in the relative sizes of dif— ferent administrative groups.8 In his study Rushing concentrated on the relative size of the administrative component, as represented by the average number of managerial and/or clerical personnel per firm in an industry. 1His findings can be summarized in three points. (1) No Significant relationship was ob- served between organization size (i.e., average firm size in an industry) and relative size of the total administra- tive component (i.e., total administrative component = average number of managerial personnel + average number of clerical personnel in an industry). (2) Total organizar tion size was inversely related at a significant level to the relative size of the managerial component. (3) Total organization size was positively related at a significant level to the relative size of the clerical component.9 8William A. Rushing, "Two Patterns of Industrial Administration," Human Or anization, 26 (Spring/Summer, 1967) , 32-39. 9Ibid., 33-36. 37 In discussing his findings Rushing notes that con- fusion often arises in research on organization size and administration because no distinction is made between different types of administrative personnel. For example, the negative correlation found between Size of organization and relative size of the managerial component is a IOgical one, if the assumption is made that increases in organizar tion size may affect the total administrative burden more in a quantitative rather than qualitative direction; that is, administrative tasks do increase as size of organiza- tion increases, but perhaps more in the direction of addi- tional routine, repetitive activities, than in decision- making or policy formulation demands. Managerial surveil- lance can thus be expanded without necessarily adding Significant numbers of managerial personnel, provided clerical staff are added to supply the informational needs of those making decisions and formulating policy. Addi- tion of clerical staff becomes, then, a less costly but functional alternative from the standpoint of operating efficiency.10 In a related study Rushing has offered a sugges- tion for further refinement in the methodology of studying the organization size-administration relationship by ad- ‘vancing the hypothesis that the relative Size of the ad- ininistrative component may be independently affected by 101bid., 36-38. 38 the degree of complexity within the administered unit(s). This hypothesis is based on an assumption that the primary function of administration is that of coordination; conse- quently, the overall administrative burden should be greater in organizations having a more complex division of labor.11 The findings of Rushing's second study have limited relevance to this review discussion, in that industry Size, measured by total number of production employees within an industry, rather than organization size is included as a key independent variable. In contrast to the first study, he provides no information concerning average firm Size in an industry. At the same time, his findings regarding the relationship between division of labor and the adminis- trative component are both interesting and instructive. The division of labor is measured by observing the uni- formity or diversity of occupational types among produc- tion personnel within each industry and is analyzed in relation to the relative size of the managerial, profes- sional, and clerical staff components. In each case the relationship between organizational complexity and rela- tive size of the staff component is shown to be positive, thus lending credence to the hypothesis as a general rela- tionship to be examined in studies concerned with the 11William A. Rushing, "The Effects of Industry Size and Division of Labor on Administration," Agh_nini§- tzwrtive Science anrterly, 12 (September, 1967), 273-295. 39 organization size-administration relationship.12 One of the earliest studies to examine the organi- zation size-administration relationship in a nonindustrial organizational setting was conducted by Terrien and Mills. In their study data were collected on school districts in California in order to test the relationship between total staff size and the proportion of the total staff in adminis- trative positions. Analysis of the data provided apparent support for the Parkinsonian thesis, in that the adminis- trative component contained a higher mean percentage of the total staff in large than in small school districts.13 Questions regarding the validity of Terrien and Mills' conclusion were later raised, however, on the basis of results obtained by Anderson and Warkov in their study of the relationship between total staff size and relative size of the administrative component in Veterans Adminis- tration hospitals.14 In outward appearance, their findings were directly contradictory to those obtained by Terrien and Mills, in that among the hospitals studied, the mean 121bid., 288-291. 13Frederic C. Terrien and Donald L. Mills, "The Effect of Changing Size upon the Internal Structure of Organizations," AmQrican Sociological Reviey, 20 (February, 1955), 11-13. 14Theodore R. Anderson and Seymour Warkov, "Organi- zational Size and Functional Complexity: A Study of Ad- Ininistration in Hospitals," American SO 1010 ical Review, 26 (February, 1961), 23-28. 40 percentage of employees in administrative positions de- clined with each size category arranged on an ascending scale.15 In discussing the different results of theirs and the Terrien and Mills study, Anderson and Warkov suggested that the difference might be attributable to variations in the structural complexity of the two organizational set— tings. For example, while each of the hospital organiza- tions studied had a single location, school districts, especially larger ones, may include several school loca- tions. Thus, the possibility arises that, in the Terrien and Mills study, by not introducing the number of School locations as a control variable, the factor of complexity was allowed to vary freely, thus accounting for the discrep- ancy between the two studies. That this is a plausible interpretation is given support by the results of a later study of school systems conducted by Lindenfeld. His data Showed that the organi- zation size—administration relationship is an inverse one, when number of schools in the system is held constant.16 Lindenfeld also extended his analysis to examine the fur- ther relationship between organization size and the rela- tive sizes of upper and lower echelons in the administrative 15Ibid., 25-26. 16Frank Lindenfeld, "Does Administrative Staff Grow as; Fast as Organization?" School Life, 43 (May, 1961), 20-23. 41 component. Although the data are presented in cross- tabular form, Observation indicates that the relationship is an inverse one in both cases, with the negative effect probably most pronounced in the case of upper-level ad- ministrators.17 Before moving on in the review discussion, note Should be taken of three specific propositions advanced by Anderson and Warkov, which they feel might offer a basis for resolving the discrepancy between theirs and Terrien and Mills' findings, by taking account of organizational complexity. These propositions are: l. The relative size of the administrative component decreases as the number of persons per- forming identical tasks in the same place increases. 2. The relative size of the administrative component increases as the number of places at which work is performed increases. 3. The relative size of the administrative component increases as the number of tasks per- formed at the same place increases (or as roles become increasingly specialized and differen- tiated).18 These propositions are important for their indicar tions of two different possible dimensions of what can be generically termed organizational complexity. The distin- _)f”' 63‘ guishing element is whether the focus is on number of tasks ,5 Cfllfifl. i ’f ‘ "‘3-“ “ ~ Q .9 1 —'_'\-",.-¥ Performed or number of work places as the key companent of Ibid., 22. 18 original) Anderson and Warkov,,gp. cit., 27.(Italics in 42 the complexity criterion. The former relates to the nature of role specialization as the conceptual basis for under- standing complexity; the latter points to a structural form of complexity which takes the pattern of subsystem differentiation as the indicator. Hence, if Anderson and Warkov are correct, two conceptual approaches are identifiable--one based on compositional features of the task-performing subunits, most clearly expressed in terms of role specialization, and the other based on structural differentiation of the total organizational population into subsystems. To stipulate that one or the other of these ap- proaches is the correct one for conceptualizing organiza- tional complexity would appear to be inappropriate. In any empirical instance, both could be important, thus making it necessary to take account of the compositional features of the task-performing subunits and the pattern of structural differentiation within the total system. In a study of institutions of higher education, Hawley and associates attempted to shed light on the or- ganization size-complexity issue raised by Anderson and Warkov.19 Variables used in their study included size of teaching faculty as a measure of organization size, number 19Amos H. Hawley, Walter Boland and Margaret Boland, "Population Size and Administration in Institu- tions of Higher Education," American Sociological Review 30 (April, 1965), 252-255. 43 of full-time professional administrators as a measure of administrative staff size, and number of departments and nondepartmentalized schools as a measure of organizatiOnal complexity. AS can be seen, their conceptualization of organizational complexity focuses on the nature of struc- tural differentiation in the organization--in this in- stance, the results Of departmentalization in the univer- sities studied. It is difficult to determine the full implications of Hawley and associates' study for the relationship of organization size and complexity to the relative size of the administrative component, since, with the exception of one table presented, their measure of the administrative component is in terms of absolute numbers, rather than a relative size index. Thus, the comments which follow are primarily based on the findings presented in a single tab1e,20 in which the administrative ratios (i.e., number of full-time equivalent administrators per 100 total faculty) are presented for various cross-tabulated cate- gories of faculty Size and number of departments and schools. The data presented by Hawley and associates indi- cate that organization size and relative size of the adr [9 ministrative component are negatively related. Since the ; 20Table 2 in ibid., 253. 44 administrative ratios are presented for size categories only, there can be no confident assertion about the under- lying linearity of this relationship, but inspection of their table gives an impression that the negative effects of size may tend to level off as the larger size ranges are reached. Using number of departments and non-departmentalized schools as an indicator of complexity, Hawley and associates found the complexity-administration relationShip to be nega- tive. However, this operationalization of complexity may be questioned, since the number of organization parts is not necessarily a good indicator of organizational (or structural) complexity. The degree of complexity within the administered system may be better determined by a relar tive measure; for example, the number of parts in relation to the total population size of the organization. Using this alternative approach to the variable Of eomplexity, it can be shown in the tabular data presented by Hawley and associates that complexity and relative size of the administrative component are positively related among the indtitutions studied.21 21The problem of an appropriate operational ap- proach to the variable of organizational complexity is probably related to the way in which Anderson and Warkov's second proposition (see p. 41 above), which this study apparently attempts to test, is interpreted. The proposi- tion actually makes little sense unless prefaced by a qualifying phrase, "When organization size is controlled ," since complexity, in whatever form, must somehow relate to the way in which population elements are dis- tributed among organization parts (subsystems) or among 45 The study by Hawley and associates also included the total amount of resources available to the organizar tion, as indicated by total operating budget, as an analytical variable. Interestingly, this was the only study reviewed which included a resource measure, although on logical grounds it is reasonable to assume that an or- ganization's resource base would be an important deter- minant of its ability to develop the administrative apparatus. In the study total operating budget was found to have a strong positive relationship to both size Of faculty and size of administrative staff. This, of course, is not a surprising finding, but it does lend support to the sug- gestion that it may be profitable to include a resource capacity measure as a test variable in analyzing the organization Size-administration relationship. The only other study reviewed which has focused its investigation on a single class of organizations was conducted by Raphael, who analyzed data on local labor unions. Briefly summarized, her study Showed virtually no relationship between membership size of the local union and relative size of the administrative component but a fairly strong positive correlation between membership size different kinds of roles (specialization). The study by Hawley and associates does not take cognizance of the imr plied population dimension, resulting in an operationali- zation which, to the present writer, is questionable. 46 and organizational complexity, the latter indicated by the degree of occupational diversity in the local's membership. There is one methodological feature in Raphael's study that is a bit confusing, however, from the standpoint of typical usages of concepts in the study of the organi- zation size-administration relationship. This relates to the operations she uses to determine size of the adminis- trative component. The measurement is an index term based on whether day-to—day administrative decisions in the local union are made primarily by high-ranking officers or are shared between officers of both higher and lower rank, with the latter circumstance receiving the higher score. It would seem that Raphael's index is, in reality, an indication of the relative degree of "oligarchy" or "democracy" in the decision-making structure of the local union, and elements of administration as usually conceived are only indirectly incorporated into the index. It is possible that this is a function both of the relatively 22Edna E. Raphael, "The Anderson-Warkov Hypothesis in Local Unions: A Comparative Study," American Sociologi- cal Review, 32 (October, 1967), 768—776. Raphael‘s concept of organizational complexity is, of course, conceptually similar to that which takes the nature of role specializa- tion as its definitional base. It can also be noted that Raphael's findings also show a fairly strong negative correlation (r = -.377) between spatial dispersion of mem- bers' workplaces and size of the administrative component. This, however, probably reflects only that local unions which have members working at spatially dispersed points depend more on the Single, roving business agent for ser- vice functions than do local unions whose members are concentrated in factory or shop settings. 47 small membership size of many local unions and of their character as voluntary associations. On the average, local unions do not normally demand high levels of con- tinuous administrative service, and much of the work can be performed through voluntary service contributions. It is necessary that they reach a certain membership size before employment of full-time officers or appointed staff is economically feasible. It is possible, then, that the character of administrative processes in local labor unions differs according to whether administration is handled by full-time employees or by volunteer workers. In summary, Raphael's study probably reveals more about the administrative dilemmas facing relatively small- sized voluntary associations than about the organization size-administration relationship which, in terms of cur- rently available conceptualizations, may well be limited in study potential to organizations that are more formalized in structure and have continuous administrative needs. Two other research studies were reviewed, each of which included organizations of varying type in its study sample. In a study of 30 organizations, Haas and associates examined the relationship between relative size of the "supportive" administrative component (the dependent variable) and total organization size, age of organization, number of operating sites, and amount of diversity in key 48 activities performed within the organization.23 The sup- portive component was Operationally defined by distinguish- ing between administrative personnel engaged in direct administrative activities (e.g., executive and managerial) and those engaged in indirect, supportive administrative services (e.g., bookkeeping and other clerical, personnel administration, maintenance service). In testing each relationship it was hypothesized that increases in the value of the independent variable would be associated with a disproportionate increase in the dependent variable. None of the hypotheses was sup- ported. Except for age of organization, rank-order corre- lations were negative in each case, though none attained a level of significance. Age of organization showed a positive but weak relationship to Size of the supportive component. The authors do assert that inspection of scatter diagrams suggests that the relationship between organization size and size of the supportive component is Slightly curvilinear; that is, the relationship is slightly positive in the case of the smallest organizations and the largest ones.24 23Eugene Haas, Richard H. Hall, and Norman J. Johnson, "The Size of the Supportive Component in Organi- zations: A Multi-Organizational Analysis," Social Forces, 43 (October, 1963), 9-17. 24Ibid., 14-15. 49 The study by Haas and associates should be viewed with some caution. The 30 organizations included in the study represent a diversity of types, extending from a local labor union to a manufacturing firm to a bank. It is possible that differences in structural and environ- mental features within each type render reliable compari- sons difficult, given the central concerns Of the study. Indik, in a study of five different types of or- ganizations, examined the relationship between the rela- tive number of supervisors and employment or membership Size of the organization.25 Each organization was a local unit of a larger parent organization. Operationally, only persons in each organization whose functional role involved either decision making or direct interpersonal supervision of others' work were classified as supervisory personnel; all individuals serving mainly clerical functions were excluded. Relative Size of the supervisory component was based on the ratio of number of supervisors to total number of members in the organization. Correlation coefficients were negative for each set of organizations studied. These correlations were strongest for industrial labor union locals, followed in order of descending magnitude by nonpartisan political organization chapters, volunteer fire companies, automobile 25Bernard P. Indik, "The Relationship between 0r- ganization Size and Supervision Ratio," Administrative Science anrterly, 9 (December, 1964), 301-312. 50 . . . 26 sales dealerships and package delivery stations. Al- though the data in the two studies are not entirely comparable, Indik's findings show considerable corres- pondence with those of Rushing discussed previously.27 Conclusion The preceding discussion has concentrated mainly on depicting the empirical findings of various studies which have focused major attention on the organization size-administration relationship. Some comparisons have been made between studies, but for the most part each has been treated in a rather discrete manner. Further com- parisons of the findings in these studies are deferred until the next section, which is concerned with an identi- fication of key conceptual issues and their theoretical implications in the study of organization Size and ad- ministration. Conceptual and Theoretical Implications of Previous Research Studies In general, the review of research literature dis- cussed in the preceding section has shown the following: (1) The volume of research studies focally concerned with 261bid., 307. 27See above, PP. 36-37. 51 organization size and administration is still quite small, although increasing. (2) The analytical tone of the in- vestigations is largely exploratory, which is probably attributable in large measure to a lack of continuity in conceptual developments and investigative interests among most of the studies. (3) The level of theoretical development, particularly in terms of identifying gen- eral statements about relationships between variables, is relatively low, a condition which is partly a conse- quence of the combined influence of the two preceding points. In this section a number of theoretical considera- tions are advanced about the organization size—administra— tion relationship, basing the discussion on an interplay of information derivable from the several research studies and more general commentaries relative to the study of organization size and administrative structure. The dis- cussion is confined to those organizational characteristics which have already appeared in the literature under the general rubric of research interest in organization size and administration--namely, the administrative component, as a dependent variable, in relation to the independent variables of organization size, complexity, and resources The objective pursued in this section is a modest one, namely, to work toward a more systematic organization of certain key issues as a means of specifying the general 52 contours of a conceptual scheme that might be profitably used in examining the effects of organization Size, com- plexity, and resources on the administrative component. In an immediate sense, the discussion serves the purpose of providing a clearer analytical perspective for the empirical investigation of these relationships among national labor unions. Two major questions underlie the organization of this discussion. Why is organization size, or complexity, or resources, an important influence on the development of administrative structures in formal organizations? And, what are the underlying organizational dynamics which give particular colorations to the variable relationships between these organizational characteristics and the administrative component? These questions do not exhaust the logical dimensions of the problem under consideration, but, prac— tically speaking, they provide convenient handles with which to grasp the basic character of the theoretical problem. The Principle of Coordination and the Concept Of the Administrative Component Essential ingredients of an answer to the first of the above two questions are derivable from two assumptions that are commonly made regarding administration in formal organizations. (1) Every organization is faced with the 53 key administrative problem of coordinating the various activities which occur within it. Viewed in this way, coordination is primarily an internal, integrative func— tion, but it also has external relevance, in that the coordination achieved internally affects organizational output. (2) The coordinative problems in an organization are partly functions of its population size and internal complexity. To these two assumptions can be added a third, which, though not commonly found in the literature on or- ganization size and administration, has emerged in the present study as a potentially important factor, namely, the resource capacity of the organization. Presumably, the amount of resources available to an organization has implications for the type of administrative apparatus which is, or can be, develOped to deal with problems of coordination. None of these assumptions is unique or even new. They have conceptual linkages, for example, with the early work of Durkheim, Weber and Simmel. With reference to societal development, Durkheim associated increases in population Size with increased organizational complexity 28 and, consequently, with coordinative problems. In Weber's discussion of bureaucracy, the process of 28Emile Durkheim, The Division of Labor in Society, trans. George Simpson (New YOrk: The FFee“Press of Glencoe, 1964 edition), pp. 256-282. 54 bureaucratization, leading to the formalization and elabora- tion of an administrative apparatus, is associated in part with the phenomenon of organization growth.29 Weber also noted that the stability of the resultant administrative machinery depends upon constancy of income 1-(a dependable resource Supply) to the organization.30 In discussing the formal properties of group structure, Simmel asserted that increases in the size of a group create the potential for a breakdown in group solidarity, thus jeopardizing the possibility of coordinated group activity. Conse- quently, the creation of intermediary positions, charged with coordinative and communications functions, are neces- sary responses to increases in group size.31 Though stated with varying degrees of explicitness, emphasis on the coordinative function of administration is also found in the research studies reviewed in the pre- ceding section. On the whole, the assumption is made that the primary function of administrative personnel is that of coordination, whether this is accomplished directly through the exercise of executive or supervisory functions or indirectly through the use of technical expertise and clerical routines, the last of which facilitates 29Weber,.pp. cit., pp. 209-216. 3OIbid., p. 208. 31Georg Simmel, The Sociology of Georg Simmel, trans. Kurt H. Wolff (New York: The Free Press of Glencoe, 1950 edition), pp. 87-116. 55 coordination through the processing of information and the handling of formal communications. In this study emphasis is also placed on coordina- tion as the major function of administrative personnel. It should be noted, however, that the principle of coor- dination does not by itself provide clues to the form of the organization size-administration relationship. On the one hand, it simply implies the existence of relationships between organization size, complexity and resources, and administrative structure, though predictive statements about these relationships must be decided on other groups. On the other hand, it serves an important definitional func— tion by indicating the same aggregate of positions in the organizational system has unit character because of common involvement in the administrative function of coordination and can be conceptualized as the administrative component. With reSpect to this latter point, it is also instructive to register an additional note relative to the manner in which the administrative component is treated conceptually and an analytical component is established to interpret relationships between it and other organiza- tional characteristics. Within the total administrative component, there will be Specialized functions contributing in different ways to the overall achievement of coordination. In the design of research studies on organization size and administration, or interpretations of others' studies, 56 careful attention must be paid to what is being defined as the administrative component and what form of differentia- tion, if any, is being made between different functional aSpects of administration in determining Operational measures. As a general principle, it can be stated that there is no singular best usage of the concept "adminis- trative component," and only when there is a clear delineation of a particular Operational usage can there be research continuity leading to generalizations of broad SOOpe. The review of research literature presented earlier has indicated some variations in Operational approaches to the concept of the administrative component. This general observation is not intended as a criticism of any individual study or group of studies, Since a Specific usage is partly a result of pragmatic considera- tions related to levels of refinement in available data and of a researcher's confinement of interest to a par- ticular functional aspect of the administrative apparatus in the organizations studied. It is not possible, however, to dismiss these differences as having simply a curiosity value, for the cumulative evidence of these studies indi- cates that different parts, or aSpects, of the administrative component may be differentially associated to any one of the major organizational characteristics of organization Size, complexity or resources. 57 A summary review of the research studies suggests that, at a general level, there are two major aspects of administration which provide a meaningful basis for analytical distinctions between administrative personnel. Each is related to a different form of participation in the overall administrative function of coordination. 1. An important part of the administrative com- ponent is represented by those personnel who are directly engaged in control and supervisory functions in relation to the selection and carrying out of organizational pro- grams and activities. In general terms, this aspect of the administrative apparatus can be labeled the "direct" administrative component.32 2. A different part of the administrative com- ponent is represented by those personnel who participate indirectly in the administrative function of coordination, primarily by means of providing supportive services to those in direct administrative positions. This aspect of 32The particular adjectival term applied to this aspect of the administrative component varies among the several studies incorporating a corresponding usage. How- ever, reference can be made to the following: the "managerial" component, as used by Rushing, "Two Patterns of Industrial Administration,",gp._git.; the "salaried administrators" component, as used by Bendix,.pp.,g;t.; the "line" component, as used by Haire,.pp._glt.; the "supervisory" component, as used by Indik,.pp..g;$.; the "professional administrators" component, as used by Hawley _et.alg,,_p.,git.; the "administrative Officials" component, as used by Lindenfeld,__p.,g;§. In each of these studies, reference is being made to a functional aspect similar to *what is termed here the "direct" administrative component. 58 the administrative apparatus can be generally termed the "supportive," or "indirect," administrative component.33 To a great extent, supportive administrative personnel are involved in informational processes, either in the sense of providing specialized forms of information (technical and professional staff) or by the handling of formal communications pertaining to organizational pro- grams and activities (clerical staff). A comparison of previous studies indicates that, in general terms, there have been two operational ap- proaches to the concept of the administrative component. In some instances, a broad definition has been used and the administrative component regarded as including all staff involved in the coordinative tasks of administration, both direct and indirect. In others, a narrower definition has been used, resulting in an implied or explicitly stated distinction between different types of administrative per- sonnel. As this study proceeds, it will become evident that cognizance of these differences in approach serves an important purpose as a caveat in working toward summary statements about research findings and tentative generali- zations regarding relationships among major variables in 33Four of the studies reviewed included Operational measures having some correspondence to the "Supportive" or "indirect" administrative component, as defined here. They are: the "supportive component," as used by Haas.gp.al., .22..g;$.; the "salaried technicians" component, as used by Bendix,._p..g;§.; the "clerical component, as used by Rushing, "Two Patterns of Industrial Administration,",gp. .g$£.; and the "clerical" and "staff" components, as used by Haire,.Qp.,§;1. 59 this area of organizational analysis. Organization Size, Complexity and Resources in Relation to the Administrative Component The research studies previously reviewed indicate that three organizational variables have been examined in terms of their relationship to the relative size of the administrative component--size, complexity and resources. In the present discussion, and throughout the remainder of the study, the major focus of attention will be on these organizational characteristics as independent variables. Although each of these organizational attributes has been included in one or more studies as an independent variable, there is considerable variation in the amount of empirical evidence regarding the effect each has on the relative size of the administrative component, or differen- tiated aspects of it. Organization size appears as an independent variable in each of the studies reviewed, organizational complexity in four and resources in but only one. The information that is available suggests, however, that each of these variables can be analyzed from two van- tage points: (1) as an independent variable relating in important ways to the relative size of the administrative component; and (2) as a control variable, with possible conditioning effects on the character of generalizable 6O conclusions that are formulated about the relationship between the other independent variables and the relative size of the administrative component. Inasmuch as organization size is the only one of the three variables for which the span of empirical evi- dence is relatively complete, it will be used in the sub- sequent discussion as the focal independent variable, and information regarding the characteristics of complexity and resources will be incorporated as the discussion proceeds. In Simplest terms, there are three possible forms for a general hypothesis regarding the relationship between organization size and relative size of the administrative component; namely, that the two variables are (l) positively 34 35 related, (2) unrelated, or (3) negatively related. It 34Without further specification of what "unrelated" means, this hypothesis is unclear with respect to general interest in the phenomenon of organizational growth and administrative development. The hypothesis of "no relar tionship" can have two meanings: first, that the propor- tionate rates of increase for organization size and for the administrative component are essentially random in relation to each other; or second, that organization size and the size of the administrative component tend to in- crease at approximately the same rates. Which is the better explanation of an observed relationship according with this hypothesis would depend on additional observar tions relative to the linearity of effects between the two variables. 35Although the characteristics of size, complexity and resources reflect a different organizational attribute, they share a common operational dimension in that each can be arranged on an ascending (or descending) scale of values, with the values reflecting larger (or smaller) increments in the measurement units of the particular variable. Thus, 61 should be recognized that the applicability of any one of these hypotheses may be affected by the manner in which the concept "administrative component" is dealt with operationally; that is, organization size may be differ- entially related to the relative sizes of different aSpects, or parts, of the administrative component. Several studies have investigated the relationship between organization size and the relative size of what can be regarded as the total administrative component-—that is, the entire complement of officials and staff employees engaged in both direct and indirect administrative func— tions. The cumulative evidence of these studies does not suggest unvarying support for the applicability of any one of the three aforementioned general hypotheses to this re- lationship. The only summary statement that can be con- fidently offered about the organization size-total adminis- trative component relationship is one of negative stipula- tion; namely, that the relationship is not positive in form.36 On balance, the studies reviewed range from those showing virtually no relationship between organization size an examination of the relationship between complexity or resources and relative size of the administrative component would be, in effect, a test of the plausibility of these same three general hypotheses. 36Only the study by Terrien and Mills offered data showing a positive relationship, but, as noted earlier, a careful review of their methodology suggests that this re- lationship may have been spurious, since a factor of organ- izational complexity was left uncontrolled in the analysis. 62 and relative size Of the total administrative component to those indicating an inverse relationship, though usually of modest strength.37 There are, of course, good reasons for the inability to generalize with specificity about the organization size- total administrative component relationship from one study to the next, particularly if the studies being compared differ as to the type of organizations being analyzed. Essentially, this is attributable to the fact that the relative size of the total administrative component is it- self a product of the way in which personnel are distributed among its subcomponents and the relationship which each of these subcomponents has to organization size.38 Organizar tions in different types of task—environments may differ 37In actuality, few of the studies reviewed used correlational analysis; hence, this observation is based largely on inspection of data presented in contingency tables. 38Although implicit in several of the studies re- viewed, in only one is this point clearly explicated. Bendix compared the relative numbers of administrative employees and salaried technicians in German industrial firms of different Sizes. He found the former to be in- versely related and the latter to be directly relatedr-a difference he attributes generally to processes of func- tional specialization associated with organization growth. However, the relationship between total administrative staff size (i.e., administrators + technicians) and firm size he found to be inverse, a re31lt which he explains in the following manner: "Because administrative employees are always more numerous than technicians . . . the over- all tendency remains for bureaucratization to be highest in the smaller firms." Bendix,._p.,gi§., pp. 222-223. 63 in their needs for a particular balance of direct and indirect administrative positions. The cumulative evidence of the studies reviewed is much clearer with respect to the relationship between or- ganization size and relative size of the direct adminis- trative component. The findings in several studies sug- gest that this relationship is an inverse one and of fairly strong magnitude. Information regarding the relationship between organization Size and relative size of the indirect, or supportive, administrative component is somewhat unclear. It will be recalled that Rushing found average firm size and relative size of the clerical component, which can be viewed as part of the indirect component, to be posi- 39 Haire's findings on the clerical com- ponent can be similarly interpreted.4O On the other hand, tively related. Haas and associates found organization size and relative size of the supportive administrative component, which would include clerical personnel as a major part, to be unrelated.41 In sum, predictive statements regarding the indirect component of administration would probably depend on the level of refinement in the data and the particular method of operationalization. What does seem clear is the 39Rushing, "Two Patterns of Industrial Adminis— tration,"|2p. cit. 4OHaire, pp. cit. 41Haas gt_§l,, pp. cit. 64 general implication of these studies that the direct and indirect administrative components display differential rates of increase in relation to increases in organization size. It is likely that the particular form of any of the three relationships discussed above is influenced by the manner in which organizational complexity is impli- cated in the general pattern of increasing organization size. This contention depends for its viability on a particular conceptual approach to the definition of or- ganizational complexity, and the discussion which follows is concerned with explicating such an approach. One approach to the concept of organizational complexity is based on mathematical specifications which indicate that arithmetic rates of increase in the population size of an organization are accompanied by geometric rates of increase in the number of possible relationships between the population elements.42 The tenability of this approach must be examined from the vantage point of an implicit assumption underlying its formulation, namely, that the potential relationships between the population elements are all compelling influences on administrative processes of coordination. 42Theodore Caplow, "Organizational Size," Adminis- ygative Science Qparterly, 1 (March, 1957), 484-505; William M. Kephart, "A Quantitative Analysis of Intragroup Relationships," Am rican Journal of Sociolo , 55 (1950), 544-549. 65 This assumption has doubtful validity since it ignores an important facet of administration in formal organizations. The key units through which administrative coordination is achieved are typically different types of subsystems into which individual members are aggregated as a matter of administrative convenience; for example, de- partments in a business concern, local unions in a national labor union, congregations in religious denominations, and the like. Consequently, the degree of organizational com- plexity in an organization is more correctly determined by observing the number and relative sizes Of these adminis- tered subsystems than by placing reliance on index charac- teristics based on potential numbers of interactions within groups of varying size. This latter approach to organizational complexity stems from a particular conceptualization of an organiza- tion as a task-performing unit. In a recent discussion of differentiation and integration in complex organizations, Lawrence and Lorsch have noted that many students of large organizations "tend to start with the individual as the basic unit of analysis and build toward the larger organi- zation. . . ."43 They suggest that an equally effective approach can start with "larger sociological entities--the 43Paul R. Lawrence and Jay W. Lorsch, "Differen- tiation and Integration in Complex Organizations," Administrative Science Qparterly, 12 (June, 1967), 3. 66 entire organization and its larger subsystems."44 Viewed in this way, the concept of organization comes to have a particular definitional form. To quote Lawrence and Lorsch: An organization is defined as a system of inter- related behaviors of people who are performing a task that has been differentiated into several distinct subsystems, each subsystem performing a portion of the task, and the efforts of each subsystem] being integrated to achieve effec- tive performance of the system. It can be easily seen that this definitional ap- proach is quite similar to the concept of "complex social system," used by Ramsoy "to refer to the problematic relar tionship between an inclusive system and its subsystems."46 The usefulness of Lawrence and Lorsch's and Ramsoy's ap- proaches is based primarily on the attention they direct to the nature of structural differentiation in the organi- zation. To understand more clearly how organizational com- plexity might operate as a contingent condition affecting the organization size-administration relationship, it is necessary to consider (1) the relationship between com? plexity and the relative size of the administrative com- ponent, as indicated in previous studies, and (2) certain 44Ibid. 4%;Q1g. (Italics in original) 46Ramsoyhpp. ci ., p. 11. 67 other conditions that may influence the way in which structural differentiation is associated with increasing organization size. Organizational complexity has been included as an independent variable in only three of the studies reviewed. In each case the dependent variable related to organiza- tional complexity can be regarded as the direct adminis- trative component.47 In each instance the relationship was found to be positive, which is not unexpected on the basis of the assumption that coordination becomes more difficult under conditions of increased complexity. It is also lOgical to assume that the relationship between organizar tional complexity and the relative sizes of the other two aspects of the administrative component--total and suppor- tive--are probably Of the same general form, though perhaps varying in magnitude. How, then, might organizational complexity be im- plicated as a conditional influence on the organization size-administration relationship? There are three general conditions that may affect the character of its contin- gency effects. 47The three studies which included organizational complexity as an independent variable were those conducted by Rushing, "The Effects of Industry Size and Division of Labor on Administration,".gp..nit.; Anderson and Warkov, ,Qp._§;§.; and Hawley,_§3.§l.,._p..git. Complexity was in- cluded in a fourth study (Lindenfeld,,pp.,gi$.) but only as a control variable affecting the organization size- administration relationship. 68 1. From the standpoint of administrative ration-- ality, it would be expected that organizations would at- tempt tO achieve the maximum size in their administered subsystems that is feasible in terms of technological and environmental constraints. 2. No organization is in complete control of dif- ferentiation processes determining the size of subsystems within its operating structure. Differences in production technologies and environmental constraints will influence the size characteristics of subsystems that are possible for integrated organizational activity leading to effective task performance.48 3. In many instances organization size and the relative sizes of the administered subsystems may stand in a direct relationship to each other. This is based on an assumption that population elements in an organization are themselves a type of resource to be distributed among the subsystems, and larger organizations by their very nature have greater amounts of this resource. As noted in the preceding point, however, organizations will vary in the degree of control they have over this allocation process. 48For example, the size of production units in a manufacturing or extractive industry might be affected by the nature of materials or resource concentrations; the geographical concentration or dispersion of certain types of populations may affect the sizes Of congregations in a religious denomination; the structure of bargaining units in an industry can affect the relative sizes of local unions in a national labor union, and so forth. 69 By building on the foregoing discussion, it is possible to specify more particularly how organizational complexity may be involved in the process of organizational growth and thus influence the form of the organization size- administration relationship. Two general statements can be made. 1. As an organization grows, if the level of com- plexity remains relatively constant, the relationship be- tween organization size and relative size of the adminis- trative component will remain generally proportionate from one size level to the next. This statement in effect posits an equilibrium principle which assumes that for each incre- ment in organization size, the level of organizational complexity increases at a similar rate, so that over a series of comparisons organization size and size of the administrative component vary directly. 2. As an organization grows, if the level of or- ganizational complexity decreases--which would result from an increase in the physical density characteristics of the administered subsystems--over a series of comparisons, the relationship between organization size and relative Size of the administrative component would move in a negative direction. This statement posits an economy-of-Scale factor as an explanatory principle, thus assuming that increasing organization size, though resulting in an ab- solute increase'in the total administrative burden, can 70 have the effect of reducing the relative degree of organi- zational complexity.49 The foregoing points, of course, project a partial model of organizational growth, and to make the statements amendable to study designs utilizing a cross-sectional, single-point-in-time approach, it is necessary to suggest certain transformations. Essentially, this is achievable by incorporating organizational complexity--in terms of physical density characteristics of administered subsystems—- as a test variable and controlling for its effects on the organization size-administration relationship. Oppositely, organization size itself may be used as a test variable in analyzing the organizational complexity-administration rela- tionship, since size characteristics may have an important influence on the general form of this relationship. 49This is, of course, closely related to the inter- pretations applied to findings about spans of supervisory control in general administrative theory. In general, there are two facets of the Span-of-control argument that size may be inversely related to the relative size of the administra- tive component. (1) A part of the managerial function of administration involves the making of decisions and design- ing of programs for policy implementation to guide activities of the task-performing subunits. Thus, as the size of an organization increases, it is not necessary that the mana- gerial component increase at a similar rate, but that the content of decisions and programs be communicated to a wider range of personnel. (2) The number of personnel that can be directly supervised by a single administrator is not inelastic. The studies of Haire and Indik, which are in- cluded in the present research review, indicate that, as organization size increases, the supervisory spans of control can be increased with no apparent impairment of organizational efficiency. 71 The third independent variable to be considered in this discussion is that of organizational resources. The study by Hawley and associates would indicate the probar bility that the amount of resources an organization commands is directly related to the size of its administrative come ponent.50 Although this is the only study reviewed which included such a variable, on logical grounds there is con- siderable justification for its inclusion in an analytical schema focusing on organization size and administration. There are essentially two forms of reasoning which lend credence to this contention. 1. Adding to the administrative staff of an or- ganization is greatly influenced by two conditions--the level of demand for a particular administrative service, which cannot be met efficiently by redistributing respon- sibilities to existing staff, and a sufficient and depen- dable resource-flow to support the new position. However, the level of demand and resource capability may not be coincident, so that the organization may simply have to "make do" with a smaller administrative staff than ideally desirable. 50Hawley,.§p.gl.,,gp. cit. The dependent variable analyzed in relation to organizational resources was the "professional administrators" component, which is a part of what has been termed in the present discussion, "direct administration." At the same time, there is no reason to assume that the relationship between resources and other aspects of the administrative component would be other than positive. 72 2. Organizations may also be in a position of having "slack resources"r-that is, resources not critically required for current production functions and programs. March and Simon note that organizations with this type of resource capability are able to engage in more intensive specialization of functions than are others lacking them.51 Regardless of which form of reasoning might apply in a particular instance, the general suggestion can be made that the amount of resources an organization commands is importantly related to the degree of elaboration in its overall operations and, consequently, positively related to the size characteristics of its administrative component. However, it can be further suggested that the usefulness of organizational resources as an analytical variable helping to explain the relative size of the administrative component may be enhanced if its operational form is determined by a standardized measurement which permits comparisons among organizations of widely varying population sizes. Viewed in this way, it is not the absolute amount of resources available to the organization which is the critical analy- 5 tical factor but the resource capacity of the organization relative to that of another (or many others). Little can be said at this point about the potential effects of a resource capacity measure as a control variable 51James G. March and Herbert A. Simon, Or anizar tions (New York: John Wiley and Sons, 1958), p. 187. 73 in relation to the organization size-administration relar tionship. In the study by Hawley and associates, the total operating budget of the organization was found to have a strong positive relationship to both size of faculty and size of administrative staff; thus, the partialling effects of total operating budget on the relationship between the latter two variables was considerable.52 However, on logical grounds, there is no reason to assume that resource capacity, as a relative measure, stands necessarily in a positive relationship to organization size. In short, the single study by Hawley and associates provides little ad- vance information about the conditioning effects of resource capacity characteristics on the relationship between or- ganization'size and the relative size of the administrative component. Conclusion The preceding discussion has not attempted an ex- haustive examination of theoretical dimensions involved in the study of organization size and administration. Rather, its key purpose has been to provide a clearer focus on major conceptual issues as a means of establishing an analytical perspective on the relationships between organi- zation size, complexity and resources, and relative size of the administrative component. In an important sense it 52Hawley, 9131., pp. cit., 254. 74 serve as a prelude to the next chapter, in which the par— ticular forms of variables and relationships to be examined in this study of national labor unions are explicated. CHAPTER III METHODOLOGY OF THE STUDY AND STATEMENT OF RELATIONSHIPS TO BE ANALYZED Methodological procedures used in the study were greatly influenced by three conditions. First, the neces- sity of assembling a single body of data from several sources of information imposed some constraints on the study design. For example, a determination of the number of unions to be studied was necessarily deferred until the extent of available information on national labor unions was known in some detail. Second, the use of existing documents and reports as data sources tended to accentuate issues of data re- liability. Consequently, considerable attention was paid to this aspect of methodological procedure. For the most part, technical questions of measurement reliability are discussed in a separate appendix, rather than the main body of this chapter. Third, the study was intended to be cross-sectional, and in order to ensure compatibility of data on the several unions studied, it was necessary to specify a focal time period for data collection. The year 1964 was selected as the target period, although it was recognized that differences 75 76 among unions in fiscal reporting periods and national con- vention dates would make this somewhat proximate. This choice was not accidental, however, in that the most comr prehensive body of relevant data available to the author fell within this time range. Sources of Data Data for this analysis were obtained from existing reports and documents pertaining to national labor unions in the United States. A general description of the four basic data sources used is presented below. The Bureau of Labor Statistics The Bureau of Labor Statistics (BLS) of the U. S. Department of Labor was used as a source for both published and unpublished information about national labor unions. The Dizegtoyy digtional and International Labor Unions in the United States, issued biennially by the BLS,1 provides one of the most comprehensive published summaries of the current status and developmental trends in unionism in the United States. The 1965 issue of the BLS Directory was used as a primary data source in this study. 1Actually, the BLS Diregyogy has been issued on a biennial basis only since 1953. Earlier editions appeared in 1950, 1948, and 1947, the last-named being the first year of its publication. There were, however, four handbooks of national labor unions antedating the first issuance of the .Dlzggtgry--prepared by BLS personnel in 1944, 1936, 1929 and 1926. 77 Each edition of the BLS Directory contains informar tion for the preceding year as well as comparative summaries regarding trends in union growth and development over longer periods of time. Most of the current information in each Directory is Obtained by means of questionnaires sent to central offices of national labor unions headquartered in the United States.2 In cases where replies are not re- ceived or partially completed questionnaires are returned, the BLS may report information from other sources for the particular union(s), if these alternative sources are adjudged by the BLS to be reasonably accurate. In the published Directory, notations are placed alongside items taken from sources other than the self-reporting question- naire. Unpublished data regarding the number of full-time officers and staff employees of national labor unions in- cluded in the 1964 BLS survey were obtained directly from BLS personnel. These data were originally compiled by means of the questionnaire used in preparing the 1965 Directory and were made available for use in this study 2The inclusion criteria used by the BLS in its survey are presented in the Preface to the 1965 Directory: "The basic requirement for inclusion in the,¥1zegtg;y list- ing of national and international unions [is affiliation with the AFL-CIO or, for unaffiliated unions, the existence of collective bargaining agreements with different employ- ers in more than one State (except for national unions of Government employees and those in the Federal service meet- ing the recognition requirements under Executive Order 10988. Every effort [is] made to include all unions that [meet] this standard." 78 on condition that the anonymity of the responding national unions be preserved in the analysis. Office of Labor-Management and Welfare-Pension Reports The Office of Labor-Management and Welfare-Pension Reports (OLMWP) of the U. S. Department of Labor is the key agency given governance over labor organization reporting requirements set forth by the Labor-Management Reporting and Disclosure Act of 1959 (LMRDA).3 The LMRDA requires labor organizations in the United States engaged in "in- dustries affecting commerce" to file various informational and fiscal reports with the OLMWP on an annual basis. Essentially, the LMRDA definition of "labor organization" applies to federated bodies and to national, intermediate and local union organizations whose members work for em- ployers in the private sector of the economy. Labor organizations whose members work exclusively for units of government are exempted from the reporting requirements of the Act. In the study two documentary sources available from the OLMWP were used to obtain data: (1) the OLMWP Register of Reporting Organizations (1964); and (2) the annual finan- cial reports submitted by national labor unions to the OLMWP. Each of these is described below. 3The Labor-Management Reporting and Disclosure Act of 1959 is commonly referred to as the "Landruerriffin Act." 79 l. The Register lists all labor organizations having reports on file with the OLMWP as of January, 1964. Each reporting organization is identified by type of organization (e.g., national, local, intermediate), State where central office is located, and national union affiliation, if applicable. 2. All labor organizations with central Office 10- cations in the United States or its territories which re- ceive and dispense funds, other than those which deal exclusively with units of government, are required to file detailed financial reports with the OLMWP on an annual basis. Each reporting organization is required to follow stan- dardized procedures in recording information about fiscal operations in the preceding year. The reporting period is based on the individual organization's beginning and ending fiscal year dates. There is no uniform reporting period for all organizations filing with the OLMWP. The only requirement is for each annual report to be submitted within a 90—day period following the close of the organi- zation's fiscal year. The annual financial reports submitted to the OLMWP are public documents and are available for research and other purposes. They are of two types: The longer LM-2 form is required of all labor organizations having annual receipts of $30,000 or more. The LM-3 form is a condensed version of the LM-2 and is required of labor organizations having annual receipts of less than $30,000. 80 Only LM-2 reports were used as data sources in this study. It was decided that the reporting categories used on the LM-3 form were much too broad to provide usable in- formation regarding administrative or fiscal activities. Each LM—2 form requires an itemized listing of receipts and disbursements as well as a statement on assets and liabilities during the annual reporting period. For certain of the aggregate expenditure categories, additional scheduled sections of the LM-2 form require more detailed information. Only selected portions of the LM-2 reports were used to obtain data, and information about these is presented below in the appropriate sections dealing with operationalizations of variables. It should be noted that each LM-2 (or LM-3) report received by the OLMWP is audited for omissions or reporting irregularities. In some cases, supplemental or corrected reports are filed by a reporting organization. In the study, supplemental reports, Where found, were treated as the primary documents for the particular national labor union(s) involved. Union Constitutions National labor union constitutions were used as sources of information about formal characteristics of union government and administration. Copies of constitu- tions for the majority of unions studied were obtained by 81 written requests to the national headquarters of each union. Use was also made of files available in the OLMWP and the industrial relations libraries at Michigan State University and The University of Michigan. Union constitutions are admittedly limited guides to the dynamics of union government and administration. In this study information derived from constitutional provisions was used in two ways: first, to provide indi- cators of formal characteristics of union structure and procedures; and second, to provide information regarding per capita tax rates and structures. Union Convention Proceedings, Officers' Reports and Union Periodicals National convention proceedings, Officers' reports, and union periodicals were used in different ways to obtain data. They were particularly helpful in supplying informa- tion on the number of full-time field staff employed by particular unions. They also served two further ancillary functions: first, as a cross-check on the reliability of information obtained from one or another of the other data sources; and second, to supply information unobtainable from one of the other sources. In these latter instances, their usage was selective, primarily because the number of unions for which either type of problem arose was relatively minor. 82 The industrial relations library collections at Michigan State University and The University of Michigan were used to obtain copies of proceedings, Officers' re- ports and periodicals for examination. Selection of Unipns to be Studied The selection of national labor unions to be studied was carried out in a two-step process: (1) General selection criteria, mandated by certain analytical objectives of the study, were applied initially to constitute a potential study group. (2) The number of national unions in this first group was reduced by a series of eliminative decisions resulting from missing data or dubiety regarding the reli- ability of data on certain organizational characteristics. Each of these procedural steps is discussed below. Selection of a Potential Study Group Five inclusion-exclusion criteria were used in se- lecting a potential study group. They were as follows. 1. Only national labor unions listed in both the 1965 BLS Directory and the 1964 OLMWP Register were con- sidered as potential units for study. The availability of uniform information about receipts and expenditures on an annual basis for each union, which is reliably obtainable from reports filed with the OLMWP, gave initial preference 83 to the listing in the OLMWP Register in establishing selec- tion criteria. Also, the reporting requirements of the LMRDA apply chiefly to unions whose members work for pri— vate employers, and it was felt that dependence on the OLMWP listing would assist in providing a group of national labor unions with general functional similarities. Further, since both the OLMWP Register and the BLS Directory were to be used as data sources, the inclusion of each union in both listings was a minimum requirement. 2. All national labor unions which filed LM-3 reports were excluded from the study. As explained earlier, the reporting categories on the LM-3 form are too general to provide much useful information about fiscal or adminis— trative Operations. 3. All national labor unions having fewer than 5,000 members, as reported in the 1965 BLS Directory, were not included as potential units for study. As might be expected, this criterion partly overlapped the preceding one, because many of these smaller unions were also those having total annual receipts of less than $30,000. But the choice of this membership figure as a cutoff point was also a result of two other considerations: (a) Several of these unions did not have an elected official or appointed staff employee as a full-time administrator. Thus, across the range of cases having fewer than 5,000 members, there was considerable unevenness in comparative administrative 84 development. (b) This decision was also based partly on conversations with several individuals having broad knowledge of the labor movement, among whom there was general agreement that unions in this Size category do not typically display stable patterns of administrative development. 4. National union organizations which have inde- pendently chartered national organizations, rather than local unions, as member units were excluded, though listed in both the 1965 BLS Directory and the 1964 OLMWP Register. It was felt that the structural and administrative features of these organizations would differ enough to make them noncomparable to national labor unions which are parent organizations of affiliated local units. 5. Those national labor unions not having local levels of organization, in the "conventional" sense of independently chartered and administratively separate local unions, were not included in the potential study group. In these types of unions, members affiliate directly with the national organization, although in some there may be re- gional branch offices as executive extensions of the national organization. It was decided that this basic dissimilarity to the more conventional organizational pattern of national and local units would have introduced noncomparable units into the analysis. 85 Table 1 shows the operative results of the applica- tion of these initial selection criteria. Table 1 RESULTS OF APPLYING INCLUSION-EXCLUSION CRITERIA IN SELECTING A POTENTIAL STUDY GROUP Inclusion/Exclusion Category Number Unions appearing in both BLS Directory and OLMWP Register listings 161 Unions excluded because fewer than 5,000 members reported in BLS Directory and/or LM-3 reports filed 38 Unions excluded because of federated form of organization with other national organizations as member units 2 Unions excluded because of no independently chartered local organizations, although having more than 5,000 members 4 Total number of unions excluded for all reasons 44 Total number of unions included in potential study group 117 Selection of the Final Study Group Application of the general selection criteria re- sulted in a potential study group of 117 national labor unions. Preliminary examination of data on these unions necessitated further eliminations on other grounds, as noted in the following points. 86 1. For several of the national labor unions re- garded as potential units for study, a reliable membership figure was judged not to be obtainable. In two instances this was due to an inability to obtain access to‘a copy of the union's national constitution. In six others, member- ship figures were initially calculated but later adjudged probably not to be reliable because of wide discrepancies between the calculated figure and alternative membership figures used as reliability checks. One other union, which resulted from a midrl964 merger of two former national unions, was omitted because of incomplete information about membership size following the merger. 2. Another reason for additional exclusions was a lack of information resulting in varying amounts of missing data on major organizational characteristics other than membership size. In ten cases information on one or more of the major analytical variables in the study was not obtainable, at least in a reliable form, from any of the data sources used. 3. Three other unions were excluded because the values of one or more of the analytical variables were in each instance so extreme as to make these unions noncom- parable to others selected for study. Whether or not these extreme values were the results of misapplications of data collection techniques by the author or of special conditions operating within these particular unions cannot be stated 87 with certainty, but attempts to examine the former possi- bility Showed negative results. In any event, their re- tention in the final study group would have undoubtedly had serious disturbing effects on the relationships between independent and dependent variables included in the analysis.4 A summary presentation of the reasons for, and fre- quency of, these additional exclusions is provided by Table 2. TABLE 2 REASONS FOR EXCLUDING OTHER UNIONS FROM THE POTENTIAL STUDY GROUP Reason for Exclusion Number Reliable membership figure not obtainable 6 Information on other major analytical characteristic(s) not obtainable from data sources used 10 Value of major analytical variable(s) too extreme for reliable comparisons to other unions studied 5 Total number of unions excluded for all reasons 21 Total number of unions included in the final study group 96 4This observation is based on inspection of scatter diagrams showing the relationship between each independent and dependent variable included in the analysis--before and after excluding these three cases. 88 Key Characteristics and Limitations of the Final Study Group In the end, 96 national labor unions were selected for study. In summary form they have the following key 5 each union has characteristics: (1) With one exception, more than 5,000 members. (2) With two exceptions, they are national labor unions in which the great majority of mem— bers work for private employers. Thus, considered in the aggregate, they are national unions displaying a relatively complete range of activities commonly associated with labor unions, including collective bargaining and strike actions. (3) They Share the basic structural similarity of being parent organizations of independently chartered and adminis- tratively separate locally based affiliates. There are two aSpects of the selection processes used in the study which place some limitations on the generalizability of results in the analysis. 1. The national labor unions selected for study do not constitute a representative sample of any known universe of organizations. The general selection criteria were purposefully established to assure that (a) each union studied would be of sufficient membership size to make 5Because of a coding error, one of the smaller unions, having slightly less than 5,000 members, was in- cluded in the final study group. Because this error was not detected until midway through the final data analysis, and in view of time and financial considerations, this case was retained in the analysis. 89 comparative analysis of the organization size-administration relationship a meaningful undertaking, and that (B) the unions studied would Share certain structural and functional similarities beyond those implied by the common label, "national labor union." The intended effect of the general selection criteria was to define, for analytical purposes, a study population having particular size and structural characteristics. 2. The possible effects on the analysis of addi- tional exclusions from the final study group is a question of a different order. It will be dealt with here in two ways: (1) Since organization size is a major independent variable in the study, and relatively complete and reli- able information about size characteristics was available for both included and excluded cases, an examination of the way in which different membership size categories are affected by these exclusions provides one basis for comment about their possible effects. (2) The industry type clas- sification proposed as a major control variable in the study6 offers a second basis for discussion of the possible effects of these exclusions, although it requires some assumptions regarding the appropriate classifications of the excluded unions. In Table 3 the included and excluded 6Industry type classifications used in the study are based on predominant product market characteristics in the industry (or industries) where a union exercises juris- diction for purposes of organizing and collective bargain- ing. These are discussed in detail in a later section of this chapter. 9O .smfiaofiaa so moaampcossoo Hmaosow as OHanHe>m cofipeeaomcfi eonm oo>finoo we .mofipmaaouoeneso oossmonm so comes one mcoflaa UOUSHONO pom msofipeoflmfimmeao Oahu “upmauoa n .OHQeSHsHowoo maomwooan poo stOSHHe .ooo.m seep noueonw on on ONHm mflsmnonsos coaosm monswfiw psooeosocse scene mom omosu one :oocHEHOPOU #02: mm cosmfimmeao meoHea m HOPOOSHQ mam area one scum scrap monsmwm ouwm assmaonsoze H «H OH as c an s Hence 0 o o o o o m ooanaoeoo eoz o H H n o oH o aoeo one ooo.ocm o m m o m HH 6 ome.eemuooo.cn H m h cH m m H oeo.¢nnoco.cm o m N nH m m H oeo.smaooo.c UOOSHONH OOOSHOSH oooaaoxm OOUSHOSH copsaoxm cocoaosn UOSHEHOpoQ manomOpdo aHso mcoflab UOOHHHam came hnpmaoqa morass HnnoHedz n 902 came ONHm Qfismnonsoz oqaa anpmaosm hnpmsosH posed: Hmooq . ZOHBo pen ooo.onm m.mm e m.eH n s.mm HH ooo.mem op ooo.me o.nm m m.mm mH m.em m mom.en on ooe.nm n.He n m.Hm ma e.Hm m mmm.em on ooo.n a .oz a .oz a .02 maso meoHss paonafism msoflss cachaflmh waspsaosfl no: .Oamu shamans“ uoxhxs Hanoprz Oahu shumscsfl nomads Hmooq ONHm sofiumufiswwno HQVB NMBWDQZH Qz< mNHm ZOHB