1W" 5%" .. l‘;:u':4 33.1.1.7-‘2m‘éiw- 5 4"“ 1 L‘ ~27?» t 1‘ .5: ' W3 3“)“ . Tr \ruq'urqnfifi E07- 35".? l\:‘ 1". \ V hxité" - rflf \ x“ ' ifiifil ' $ 5 ...:A .‘r~5~‘.— a .J-. 3“ 9 - ? 3' :. -_-.: ‘ \ 11"" :fil 3.4.4. «‘lft‘ In” l'J' . 5‘5 '5“? 1"“. "3?... ‘ l .5 ;; 5 I553: 514:9. 5&5 _. .A so a o ., tf'.'-_.._z‘- . . '1’}. ('5 ’ Err-551...: -~..-'.'3 Uh-¢.’+6"r I 7‘1 E -i-."-°' "*5"”r"t.. . . " .a I 5.8».- '4 .13; '.'f This is to certify that the dissertation entitled THE EMPLOYEE UNIONIZATION DECISION: A PATH ANALYSIS OF A PROCESS MODEL presented by Steven Louis Premack has been accepted towards fulfillment of the requirements for Ph.D. degree in Social Science with a concentration in Labor and Industrial Relations . ., , .. 55/55/5555" i 3 1293 Major professor 7/19/85 Richard N. Black Date MS U it an Affirmative Action/Equal Opportunity Institution 0-12771 MSU RETURNING MATERIALS: Place in book drop to LnaaAmas remove this checkout from .‘—.—5. your record. FINES will be charged if‘book is returned after the date stamped below. 5’0JUN 0‘0 5999 .. :9 0‘15 59937 41555510132000 I 045415 9 I3 200 U .i 1“ ..- hp t 5:; e: 7 8 "5* APP- 0 5 5‘ ‘1 ‘v' 0 APR 1 1 2000 .. 5 96 THE EMPLOYEE UNIONIZATION DECISION: A PATH ANALYSIS OF A PROCESS MODEL BY Steven Louis Premack A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY School of Labor and Industrial Relations 1985 ABSTRACT THE EMPLOYEE UNIONIZATION DECISION: A PATH ANALYSIS OF A PROCESS MODEL BY Steven Louis Premack An important component of the employee unionization process is the prediction of the individual's decision to vote for or against union representation. A process model was presented that posits direct causal relationships between variables potentially influencing the individual's unionization decision. From the process model 7 variables were identified to be tested in the path model: wage level, perceived pay equity, extrinsic job satisfaction, satisfaction with administration, perceived union instrumentality, voting intentions, and report on vote. It was hypothesized in the path model that: (l) a low wage level will lead to perceived pay inequity; (2) perceived pay inequity will lead to extrinsic job dissatisfaction; (3) extrinsic job dissatisfaction will lead to dissatisfaction with the way the organization is administered; (4) dissatisfaction with administration will lead to the perceived instrumentality of union representation; (5) union instrumentality perceptions will lead to intent to vote for a union; and (6) intent to vote will lead to a vote for union representation. Data on the relations among this set of variables was derived through meta-analyses of the results of 14 individual-level unionization studies. A total of 5378 individuals participated in the 14 studies. Due to missing data, the pay equity variable had to be deleted from the analysis. The results of the path analysis testing the causal model largely confirmed the hypothesized sequential structure among the predictors. A path model was found that better fit the data, however. Two conclusions were made on the basis of this revised path model. First, the relationship between intent to vote for a union and report on vote is one of meassurement rather than causality. Second, many individuals do not go through the entire sequential decision process. Some employees decide for union representation based solely on their wage level, some also consider their extrinsic job satisfaction, others also take into account their satisfaction with administration, and others also consider their union instrumentality perceptions. ACKNOWLEDGEMENTS As I reach the end of my graduate school training I am able to reflect on how much I have developed as a person and a scholar. A number of individuals have had a profound influence on my development and direction. For my initial interest in graduate school in general, and industrial relations in particular, I am grateful to Bawa Singh and Brian Heshizer for their encouragement and guidance. Some graduate students have been fortunate enough to have had the opportunity to benefit from the mentor-protege relationship. I was fortunate enough to have not one but three mentors: Richard Block, John Hunter, and John Wanous. I convey my gratitude to these fine individuals and scholars whose assistance and knowledge allowed me to learn psychology and organizational behavior as well as industrial relations. This training enabled me to develop as an interdisciplinary industrial relations scholar. Also, the members of my dissertation committee made many helpful comments and suggestions on this project. My thanks to Michael Moore, John Wagner, and Daniel Kruger. Robert Heneman and David Pincus also participated in the mentoring process as fellow graduate students, by helping ii me to organize my knowledge as well as my priorities regarding my career objectives. And of perhaps greater importance they provided me with their friendship. Michigan State University provided me with the financial resources to support my graduate studies. I have also been blessed with the consistant encouragement and support of my mother and brother, Kim Stempien, and Brian Gray. Finally, I wish to thank Kim Bancroft for her technical expertise in preparing this dissertation. iii TABLE OF CONTENTS Page LIST OF TABLES........................................Vi LIST OF FIGURES.......................................Vii CHAPTER ONE: Introduction and Overview................ 1 Importance of the Topic.......................... Key Assumptions and Limitations.................. Contributions.................................... Description of Chapters in the Dissertation...... Literature Review............................. Hypotheses and a Model of the Individual Unionization Decision......................... Research Methodology.......................... ReSUItSOOOOOOO00.0.0000...OOOOOOOOCOOCOOOOOOOQ DiSCUSSionOO00......00000OOOOOOOOOOOOOOO00.... 0000qu mmmm CHAPTER TWO: Literature ReVieWOO00.000.00.00000000000019 How Individuals Become Union Members.............lfi Operationalizing Unionization....................12 Empirical Literature.............................l3 Wage Level....................................lS Perceptions of Equity.........................l9 Extrinsic Job Satisfaction....................24 Perceived Union Instrumentality...............28 Intent to Vote................................3l Summary and Conclusions..........................32 CHAPTER THREE: Hypotheses and a Model of the Individual Unionization Decision.......35 Previous Models..................................35 Farber and Saks, 1980.........................36 Kochan, 1980..................................37 Brief and Rude, l981..........................39 Youngblood, et al., l984......................40 A Model of the Unionization Decision and Specific Hypotheses..............................41 Summary..........................................Sl CHAPTER FOUR: Research Methodology....................53 overVieWOOOOOOOOOOOOOOCOOOOOOO0.00.00.00.0000000053 MethOd Of ReView. O O O O O O O O O O O O O O O O O O O O O O O O O O I O O C O O 53 StUdy CharaCteriStiCSocoooooooooooooooooooooooooo56 Procedure and Analysis...........................58 iv Table of Contents Continued CHAPTER FIVE: Resu1ts.00......OOOOOOOOOOOOOOOOOOOO0.0.64 Path Analysis of the Unionization Decision.......64 sumarYOOOOOOOOOOIOOOOOOO0..0.00.00.00.000000000075 CHAPTER SIX: DiSCUSSiODOO0......00.000.00.00000000000078 Modeling the Individual Unionization Decision....78 Implications.....................................82 Limitations......................................86 Future Research..................................87 APPENDICES A. Studies Used in the Meta-Analysis................9l B. Coding Sheets....................................93 C. Results of the Meta-Analyses Used to Derive the Correlation Matrix...........................112 REFERENCESOOOOOOOO.0...OOOOOOOOOOOOOOOOOO0......O0.00.115 LIST OF TABLES Page Characteristics of Studies Included in the Meta-AnaIYSiSOOOOOOOOOO00......0.0.0000000000057 Mean and Variance of Scale Reliabilities Used in Correcting for Attenuation.....................62 Intercorrelations of All Variables Free of the Effects of Error of Measurement and Sampling ErrorOOOOOOOOOOOOIOCOOOOOOOOO0.0.0.00000000000000065 Information on the Relationships Among Variables in the correlation MatriXOOOIOOOOOO0.0.0.00000000066 The Assessment of the Path Analysis for the MOdel in Figure 3.0.0.0000000000000000.0.0.000....72 The Assessment of the Path Analysis for the Model with Intent to Vote and Report on Vote combinedOOOOOOOOOOOOOOOOO00.0.00....0000000000000074 The Assessment of the Path Analysis for the ”Odel in Figure 500......0.0.0.000...0.0.0.000000077 vi LIST OF FIGURES Figure Page 1. 2. A Process Model of Individual-Level unionizatiODOOOOOOOOOO...00.0.00...00.0.00000000043 A Causal Model of Individual-Level Uflionization.’0.0.00.0...00......00.000.000.0000046 The Quantitative Causal Diagram Derived from the Original Model of Figure 2...................7l The Quantitative Causal Diagram for Intent to Vote and Report on Vote Combined.................73 The Quantitative Causal Model Which Fits the DataOOOOOOOOOO...00......OOOOOOOOOIOOOOOOOO0.0.0.76 vii CHAPTER ONE Introduction and Overview This chapter provides a brief introduction and overview of the entire dissertation including: the reasons why it is of importance to investigate the individual's unionization decision, the research approach to be followed, the limitations and contributions of this research, and an outline of the subsequent chapters in the dissertation. Importance g: the Topic The process of employee unionization has been the subject of considerable research in the area of personnel and industrial relations. There has been a decline in unionization as a proportion of the eligible employees organized from roughly a third of non-agricultural employees in 1950 to perhaps less than 26 percent today. Researchers and practitioners in the field have questioned the reasons for this decline, the extent of future decline, and the conditions under which unions may again expand their membership (Strauss, 1984). Research on unionization has primarily been focused at the macro level where the relationship has been demonstrated between the level of union growth and aggregate economic indicators, the political and social climate, and the legal environment (Block & Premack, 1983). This research, however, by failing to account for the impact of collectively negotiated union security provisions, does not distinguish between the decision to join a union and the decision to accept employment. Specifically, using aggregate changes in union membership as the dependent variable, this research does not consider the substantial proportion of unionized employees whose choice to join a union cannot be distinguished from the choice to take a job. These individuals accepted a job in an organization that has a union security provision, making union membership compulsory within a given time period. Because of the importance unions have placed on negotiating union security provisions, as overall employment increases so does the extent of union organization, and vice versa. Investigations of overall union growth using macro variables have waned in recent years. And because of the increased sophistication of employers and unions in influencing the outcome of the union representation election through campaign tactics such as legal delays, the unionization process is less influenced by economic conditions than the actions of the parties themselves. Less aggregated research has focused on the union representation election as the relevant unit of analysis. This research has investigated the relationship between election outcomes and such independent variables as the size of the election unit and the time lapse between the filing of the petition and the actual election (e.g. Heneman & Sandver, 1983). However, election research does not assess the antecedents of the individual's decision to vote for or against union representation. How important is an individual decision in a group determined process? Roomkin and Block (1981), for example, found that on the average union representation elections were decided by approximately eight votes. Employer won elections were determined by an even narrower margin of victory. Thus, there is considerable reason to focus on the individual's unionization decision, i.e. voting in a union representation election. In the last decade, there has been a proliferation of individual-level studies investigating why workers join unions. These studies have used a wide array of work-related attitudes and individual characteristics in attempting to predict the individual unionization decision, with job satisfaction the most frequently investigated predictor (Heneman & Sandver, 1983). However, this research has not been theoretically oriented and has not led to a process model of the individual unionization decision (Block & Premack, 1983; Fiorito & Greer, 1982; Heneman & Sandver, 1983). This dissertation will investigate the unionization process at the level of the individual's decision to vote for or against union representation. This research will present and test a process model of the individual unionization decision. This model builds upon prior. research and posits direct causal relationships between variables potentially influencing union voting behavior. The data to test the proposed model will be collected from meta-analyses of the results of individual-level unionization studies. These data will offer several distinct advantages over the data used in previous research. By quantitatively cumulating the results of research studies, meta-analysis produces a more accurate estimate of the true (population) relationship between two variables by correcting for the influence of statistical artifacts (e.g. Hunter & Schmidt, 1978; Hunter, Schmidt, & Jackson, 1982). Meta-analytic techniques have been widely used in the personnel selection area (e.g. Schmidt, Hunter, & Pearlman, 1981) and are beginning to be used systematically in the organizational behavior area (e.g. Fisher & Gitelson, 1983). While several techniques exist for statistically cumulating research findings across studies, the most accurate method involves a cumulation of effect sizes (correlations) across studies to produce a descriptive estimate of the magnitude for a given effect size (Hunter, et al., 1982). Through meta-analysis, a more accurate estimate of the relationship between variables will be possible than through any single set of data uncorrected for the influence of artifactual error variance. Finally, when these data are subjected to path analysis to test the hypothesized causal process model, it will yield a substantive degree of explanation and prediction of the individual unionization decision. Previous research has only indirectly assessed the relationship of the potential antecedents of employee unionization. Key Assumptions and Limitations The dissertation research on employee unionization is limited to an investigation of the causal antecedents of the individual unionization decision. A substantial proportion of individuals, however, become union members in any period of time through the impact of union security provisions contained in collective bargaining agreements (Block & Premack, 1983). Such individuals become union members by virtue of taking a job in an organized workplace without ever having made a decision for or against union representation. This research focuses on the process of the individual's unionization decision and not on union growth per se; it rests on the assumption that unionization is an individual choice behavior. This research may be circumscribed by the limited set of variables investigated. Because the data to be used for the dissertation are dependent upon the characteristics of the research literature, this research may be limited to a subset of the variables potentially impacting on the unionization decision; the estimation of parameters may suffer to the extent all relevant predictors of union vote are not included in the causal model. Macro variables that serve to influence the context within which employees make their decision to vote for or against unionization have been identified as potentially important in understanding the unionization process (Block & Premack, 1983). For example, an employee may fear employer retaliation for organizing efforts or an individual's immediate work group may exert pressure for or against voting union. Both of these factors may result in a vote that differs from an employee's true preference. Based on the characteristics of the available data, this research is unable to assess the impact that the behavior of the parties (the employer and the union) has on the individual's unionization decision. Although actions taken by the parties during the election process may influence individual voting behavior (Dickens, 1983), the range of organizations from which individuals were sampled from may moderate the behavior of the parties. For example, if no employer resistance occurred during a unionization attempt in one organization, and a moderate degree of resistance occurred in an organizing drive involving another organization, then the mean level of resistance may be relatively low across studies. Still, the actual level of resistance encountered across the organizations investigated in this research is unknown. Although a limited subset of all the potential antecedents of the individual's unionization decision will be investigated in this research, this caveat is balanced with the decay in prediction that occurs as the length of the causal chain increases (Hunter 8 Gerbing, 1982). Contributions This study lends itself to several outcomes that will serve to increase the understanding and prediction of the individual unionization decision for researchers and practitioners in the field of personnel and industrial relations. Unlike previous research in this area, a causal model of the unionization process is proposed. This model posits direct relationships among a set of variables potentially determining the individual's unionization decision. The use of path analysis will allow the assessment of the relative causal impact of all the variables in the model. Moreover, because the data used to test the causal model is derived from meta-analyses of research findings across studies, it will permit a superior estimation of the population relationships among variables by correcting for the impact of statistical artifacts such as sampling error. Path analysis will yield more accurate (parameter) estimates of causal influence when the impact of statistical artifacts is controlled for (Hunter & Gerbing, 1982). Description of Chapters in the Dissertation Presented below is a brief description of the material contained in the next five chapters of this dissertation. Literature Review This chapter presents a brief institutional overview of the process by which individuals become union members. Following this is a review of the empirical individual- level unionization literature. Within the empirical literature, emphasis is placed on the set of variables investigated as predictors of unionization. Hypotheses and a Model of the Individual Unionization Decision Previous attempts at modeling the individual's unionization decision are examined followed by a synthesis of the literature review and the development of path analytic hypotheses on a theoretical and empirical basis. In this chapter a process model of the individual unionization decision and a qualitative causal diagram will be presented. Research Methodology This chapter describes the research design, method of data collection, procedure and type of data analysis used in the dissertation. Results This chapter reports the quantitative causal diagram including the path coefficients between variables. The correlation matrix that these parameters were estimated from will also be reported. A description of the support or lack of support for the hypotheses is presented. Discussion In this final chapter a discussion is presented concerning the support or lack of support for the hypotheses, the theoretical and applied implications of this research, the limitations associated with this study, and the directions that future research might take in this area. CHAPTER TWO Literature Review The purpose of this chapter is to review the relevant empirical literature for each of the variables selected for analysis. At the outset, the process by which individuals may become union members is briefly reviewed. Following this is a discussion of the manner in which unionization as a criterion measure has been treated in the individual-level research literature. The five independent variables examined next (wage level, equity perceptions, extrinsic job satisfaction, union instrumentality, and intent to vote for a union) were incorporated in this study not only because of their importance in the empirical literature, but also because they represent the components of the path model of the individual's unionization decision to be tested. Each of these topics will be covered in turn. This review serves as the stimulus to the path model and hypotheses developed in the next chapter. How Individuals Become Union Members The National Labor Relations Act (NLRA) passed in 1935 and amended in 1947, 1959, and 1974 established the basic procedure for determining whether employees desire union representation. 11 The NLRA enables a union selected by a majority of employees in an appropriate unit to be the exclusive representative of all employees in the unit for the purpose of collective bargaining. Upon the filing of a petition by a labor organization claiming the allegiance of at least 30 percent of employees in the proposed bargaining unit, the NLRB will conduct a secret ballot representation election to determine whether employees wish representation for collective bargaining purposes by the petitioning labor organization. When the NLRB holds a secret-ballot election to determine whether employees want a given bargaining representative, a central objective of the NLRB in administering the election process has been to ensure that employees are able to exercise complete freedom of choice through the establishment of "laboratory conditions" for the election process. Any unfair labor practice that tends to interfere with free choice will result in an election being set aside; for example, threats of job loss or loss of benefits by the employer and/or union to influence votes or union activities of employees (Kochan, 1988; Mills, 1982). Although elections may be conducted within 30 days after the regional director of the NLRB directs that an election he held, delays do occur as the result of actions 12 by the parties. Empirical evidence indicates that these delays cost unions votes in representation elections. Roomkin and Block (1981) found that the longer the delay between petition and election the greater the decline in employees' participation in the election. Operationalizing Unionization The decision by an individual to join a union as a criterion measure has been operationalized in two different ways in the individual-level unionization literature. First, unionization has been examined as a behavioral intention to vote in a union representation election should one occur. For example, Allen and Keaveny (1981) in surveying faculty members, assessed unionization with a one-item intention measure, "Do you feel there is a need for a union to represent the interests of faculty members at the University of Wyoming?" (p.584). Second, unionization has been assessed as a report on actual vote for or against union representation (see DeCotiis & LeLouarn, 1981 for an example). Although intent to vote should be most strongly related to actual vote (Fishbein, 1967), this relationship may be less than perfect. Consequently, in the review to follow, the empirical literature assessing the predictors of the individual's unionization decision will be distinguished by the method of criterion measurement used (i.e., intention or report on vote). l3 Empirical Literature A number of individual-level empirical studies have been conducted to assess the relative predictive power of various components of the individual's unionization decision. As will be seen in the review of these studies to be presented here, the majority of them have focused on identifying work-related attitudes and to a lesser extent individual characteristics as statistically significant determinants of employee unionization. While a large proportion of studies have investigated the effect of attitudes, the specific choice of variables to be included has been more a function of advances made in the measurement properties of attitudes than the application of any theoretical framework. These studies have investigated the unionization process among a number of occupational groups, although the majority of studies have focused on university faculty. Studies will be reviewed according to the variables to be investigated in the path model. These variables include wage level, perceived equity, extrinsic job satisfaction, union instrumentality beliefs, and intent to vote for a union. All these predictors of employee unionization, with the exception of wage level, are work-related attitudes. An approach for better understanding how work-related attitudes influence an individual's unionization decision 14 is to examine the nature of attitudes. Most often, work-related attitudes are divided into three components: a cognitive component dealing with beliefs and ideas about the work environment; an affective component dealing with an individual's feelings toward the work environment; and a behavioral intent component dealing with the behavioral intentions an individual has toward the work environment as a result of affective responses (Triandis, 1971). More recently, other theorists have argued that the notion of an attitude is best understood simply as an affective response, with cognitive and behavioral intent components as antecedents and outcomes of the attitude itself (Fishbein & Ajzen, 1975). Still, attitude theorists use the same three basic components of attitudes, whether for measurement and/or heuristic purposes. In the context of employee unionization, beliefs represent the information an individual holds about the work environment. An individual may describe his or her pay as equitable, or a union as instrumental for obtaining work-related rewards. Attitudes or affective responses are thought to be influenced by beliefs. For instance, an individual who believes her pay is inequitable may develop a negative attitude toward the job and be dissatisfied. This affective dissatisfaction with the job or organization 15 may, in turn, lead the individual to choose a particular type of behavior; for example, the behavioral intention to vote for a union in a representation election if the employee believes a union can help. Wage Level The most frequently investigated job characteristic in the unionization literature has been an individual's wage level. In some instances, wage level may be a function of both individual characteristics and job characteristics; as for example, when differential pay is partially based on one's performance in that job. For most organizations, however, those working at similar jobs receive similar rates of pay. Getman, Goldberg, and Herman (1976) conducted one of the original studies in the unionization area that assessed the importance of an individual's wage level on his or her unionization decision. In this investigation, the largest scale study of individual voting behavior ever conducted, Getman, et a1. (1976) surveyed 1239 workers who voted in thirty-one NLRB representation elections occurring in the Midwest and Midsouth between February 1972 and September 1973. Individuals were surveyed twice, once before the election campaign began and again after the election. Both measures of unionization were used: report on actual vote and intent to vote for the union. 16 They found correlations of r = -.13 and -.12, respec— tively; the higher an individual's level of pay the less likely they are to vote, or intend to vote for unionization. More recently, Farber and Saks (1988) used the Getman, et al. (1976) data to test a utility - maximization model of employee voting behavior in NLRB elections. Their general hypothesis was that an individual will vote in favor of unionization if the expected utility of the job changing to a union job is greater than the expected utility of the job remaining nonunion. A major determinant of the utility function was hypothesized to be an individual's relative wage level. In other words, the individual's position in the wage distribution of the organization is considered to be a major determinant of voting behavior. In their reanalysis, Farber and Saks (1980) found that an individual's location in the organization's wage distribution, rather than his or her absolute level of earnings, was a better predictor of voting behavior. Those workers falling in the lower end of the wage distribution are more likely to vote for the union because the earnings advantage of unionization is inversely related to one's position in the organization's earnings distribution. This is because unions function to raise 17 the mean level of wages while reducing the variance of the wage distribution (Freeman & Medoff, 1984). Bigoness (1978) conducted a study of 222 faculty members at a New England land grant university. Salary level was found to be a strong correlate of intent to unionize among faculty (r = -.41). Similarly, in another study of university faculty, Allen and Keaveny (1981) surveyed 220 faculty members at the University of Wyoming. They found almost an identical relationship as Bigoness (1978) between salary level and intent to vote for'a union (r = -.37). Hammer and Berman (1981), however, did not find a strong relationship between salary level and unionization. Surveying 109 faculty members at a four—year private college in the Northwest one month after a runoff election in which a union was designated as the exclusive bargaining agent, salary level of faculty was found to be essentially unrelated to report on vote for union representation (r = -.05). In yet another study of university faculty members, Ahlburg (1984) examined relative wage level as a predictor of voting behavior, attempting to replicate the Farber and Saks (1980) findings. In a survey of 277 faculty members at the University of Minnesota, Ahlburg (1984) found, as did Farber and Saks (1980), that the higher an individual 18 is in the organization's earnings distribution, the less likely he or she is to vote for a union. Unfortunately, since Ahlburg (1984) and Farber and Saks (1980) used only multiple regression analysis, and did not report individual correlations assessing the magnitude of the relationship between wage level and report on union vote, their findings are not directly comparable with the other research in this area. The final study of the relationship between wage level and unionization was conducted by Youngblood, DiNisi, Molleston, and Mobley (1984). They interviewed members of a consumer panel who are regularly surveyed for their views on a number of topics. Their sample, drawn from two southeastern states, constituted a representative cross-section (according to the geographic distribution of the population) of individuals with a median household income of greater than $6,000 annually. Youngblood, et a1. (1984) found a correlation of r = -.17 between wage level and intent to vote for a union with a sample of 303 individuals. These studies, taken together, at first glance appear inconsistent. The range of reported findings varies from -.05 (Hammer & Berman, 1981) to an upper a low of r bound of r -.41 (Bigoness, 1978). The study with the largest sample size (and correspondingly the least 19 sampling error), Getman, et a1. (1976) N = 1239, reported correlations falling between these two values for report on vote and intent to vote (r = —.13, r = -.12), respec- tively. Perceptions of Equity Equity, as a predictor of employee unionization, has been treated two basic ways in the empirical literature. First, equity has been examined as the relative equity of one's pay in comparison to others in the workplace. Second, equity has been assessed as the degree to which individuals are satisfied with the way the organization is administered; this can be termed system administration equity (Goodman, 1974). In the individual-level research literature on unionization, equity has been operational- ized with both dimensions of this concept, reflecting the interdisciplinary nature of this research. Investigators in the behavioral sciences have tended to examine perceived pay equity, while those in industrial relations have emphasized satisfaction with organizational policies and procedures in their investigations into the individual's unionization decision. In the review of research assessing the importance of perceived equity on the individual's unionization decision, a distinction will be made between pay equity and system administration equity. Several studies have 20 provided evidence that the degree to which an individual perceives inequity with regard to wages and/or the system administration they will favor unionization. In one study, Schriesheim (1978) surveyed 59 production workers and operationalized system administration equity as satisfaction with company policy. Schriesheim (1978) found a strong relationship between equity and report on union vote (r = -.55). In another study, Kochan (1979) analyzed data from the 1977 Quality of Employment Survey conducted by the University of Michigan‘s Survey Research Center. This study surveyed a representative sample of 804 individuals, including all major occupational groups in a wide variety of organizational settings. Kochan (1979) used a measure of pay equity that assessed the extent to which one perceived their wages to be inequitable relative to others doing the same kind of work. He found a correlation of r = -.21 between perceived equity and intent to vote for unionization. In a study involving 95 registered nurses employed by a private hospital in the Northeast, DeCotiis and LeLouarn (1981) examined the impact of perceived system 7 administration equity on both voting intentions and report on actual vote. The authors classified system administration equity, termed fairness, as a component of 21 the organizational climate. Perceived equity was found to be substantially related to both intent to unionize and report on actual vote (r = -.38, -.33), respectively. The remainder of the studies in this area examined the relationship between some operationalization of equity and unionization among university faculty. Allen and Keaveny (1981) examined perceived pay equity assessed by a three-item scale. The purpose of this scale was to ask faculty, "whether their raise was too small, about right, or too large, considering their own job performance relative to their co-workers" (p. 584). Low scores on the scale indicated feelings of undercompensation and high scores indicated feelings of being equitably compensated. Allen and Keaveny (1981) found a correlation of r = —.32 between perceived pay equity and intent to vote for a union. In another study, Hammer and Berman (1981) examined the importance of a measure of system administration equity on the individual's unionization decision. Operationalized as trust in decision-making, equity was measured with a three-item scale containing the following items: "To what extent do you trust the central administration (your college administration, your department administration) to make decisions that you consider appropriate?" (p. 417). This measure of system 22 administration equity was strongly related to report on union vote (r = -.53). In another study, Zalesny (1985) surveyed 455 faculty members at Michigan State University prior to, and immediately after a union representation election. She examined the impact of perceived system administration equity on both voting intentions and report on actual vote. System administration equity was, like Hammer and Berman (1981), operationalized as trust in administrative decision-making. Faculty were asked to indicate the extent to which they trusted the university administra- tion, and to what extent the administration dealt with faculty in an open and honest manner. Zalesny (1985) found system administration equity identically related to both intent to unionize and report on actual vote (r = -.47). Dworkin and Lee (1985) surveyed 510 faculty members at a large Midwest university in Spring 1984. They examined system administration equity through a measure of organizational fairness (i.e. satisfaction with the fairness of administration policy). Dworkin and Lee (1985) found that faculty members who were less satisfied with administration policy are substantially more likely to intend to vote for a union. Similarly, Ahlburg (1984) examined the importance of system administration equity in predicting the 23 individual's unionization decision. He used a three-item index measuring an individual's satisfaction with the manner in which university policies are administered within his or her department and college, and the way promotions and salary increases were given out prior to the election. Ahlburg (1984) found that faculty who felt the system administration was equitable were substantially less likely to vote for the union. Both Dworkin and Lee (1985) and Ahlburg (1984) reported only multiple regression analysis results; since correlations are not presented, results are not directly comparable. Finally, three surveys of faculty members at Michigan State University conducted in 1977, 1978, and 1981 are reported by Zonia (1983). She assessed a measure of system administration equity using a three-item index. Faculty were asked whether they felt procedures for reappointment of faculty were generally fair to the faculty involved; whether the university is dealing with its current budgeting problems in the fairest way possible; and if the current grievance procedures for faculty are adequate. In the 1977 study, a substantial relationship was found between perceived equity and intent to vote for the union with a sample of 318 faculty members (r = -.41). The 1978 study surveyed 263 faculty and found a strong relationship between system administration equity 24 and report on union vote (r = —.6l). Similarly, in the 1981 survey of 290 faculty members a correlation of r = -.58 was found between perceived equity and intent to vote for the union. In all three studies the same measure of system administration equity was used. In summary, the correlations between equity measures and unionization vary from r = —.21 (Kochan, 1979) to r = -.61 (Zonia, 1983). However, the lowest reported values: r = -.21 (Kochan, 1979) and r = -.32 (Allen & Keaveny, 1981) are for measures of pay equity. It could be argued that due to the focus of these measures (i.e. perceived equity with regard to pay in comparison with referent others in the work environment) they are measuring a different concept than does the system administration measure. Equity with regard to an individual's economic outcomes of the exchange process between employer and employee may be different from the more global satisfaction with the way the organization or workplace is administered. Extrinsic Job Satisfaction The concept of extrinsic job satisfaction is one of the most widely studied variables in the behavioral research literature on the individual's unionization decision. In a general sense, job satisfaction may be defined as "a pleasurable or positive emotional state 25 resulting from the appraisal of one's job or job experience" (Locke, 1976, p. 1300). It is thought to result from the perception that an individual's job actually provides what one values in the work environment. Satisfaction represents an affective response to the job or work environment. Based on work experiences, individuals receive outcomes or rewards from the job. These outcomes have been classified according to whether they are intrinsic or extrinsic in nature. Intrinsic outcomes have been most commonly defined as those "derived directly from or inherent in the task or job itself - associated with the content of the task or job"; extrinsic outcomes are those "derived from the environment surrounding the task or work — associated with the context of the task or job" (Dyer & Parker, 1975, p. 456). Extrinsic satisfaction is typically associated with rewards like pay and promotion. Intrinsic satisfaction is thought to be associated with outcomes like meaningful work and satisfying co-worker relationships. Extrinsic job satisfaction in the unionization literature, has been measured using well-developed (extensive normative data) ratings scales such as the Job Descriptive Index (JDI) and the Minnesota Satisfaction Questionnaire (MSQ) as well as a number of ad hoc scales assessing economic or "bread and butter" satisfaction. 26 Getman, et a1. (1976) conducted one of the original studies in this area. They used an ad hoc extrinsic measure composed of satisfaction with pay, benefits, promotional opportunities, and job security. Getman, et a1. (1976) found a correlation of r s -.36 between extrinsic satisfaction and report on union vote. In another study, Schreisheim (1978) assessed extrinsic job satisfaction using the MSQ with the following facet measures: job security, pay, and working conditions. He found a strong relationship with report on union vote (r = -.59). Bigoness (1978) examined the relationship between extrinsic job satisfaction and unionization with the JDI. He found a correlation of r = -.35 between satisfaction with pay and promotional opportunities, and intent to unionize. In one study, Kochan (1979) investigated the importance of satisfaction with economic or traditional bread and butter aspects of the job. He found a correlation of r = -.30 between an ad hoc measure of satisfaction with wages, fringes, and working conditions, and intent to vote for a union. In a study involving 182 librarians in a public library system in the Midwest, Maxey and Mohrman (1980) investigated the importance of economic satisfaction on 27 the individual's unionization decision. Economic satisfaction, measured by the MSQ, included satisfaction with benefits, amount of compensation, and compensation equity. They reported a correlation of r = -.33 between extrinsic satisfaction and intent to unionize. Hammer and Berman (1981) also examined the role of extrinsic job satisfaction. They used two items from the MSQ to measure satisfaction with job security and salary. These were combined into an economic satisfaction scale. Hammer and Berman (1981) reported a correlation of r = -.21 between extrinsic satisfaction and report on union vote. Another study (DeCotiis & LeLouarn, 1981) assessed economic satisfaction on an ad hoc scale containing the following extrinsic aspects: pay, pay increase, career opportunity, job security, pay equity, benefits, and conditions of work. DeCotiis and LeLouarn (1981) found a substantial relationship between extrinsic job satisfaction and both intent to vote and report on actual vote (r = -.40, -.38), respectively. Finally, Zalesny (1985) used a two-item ad hoc scale to assess the role of economic satisfaction in predicting unionization. She found an identical relationship between economic satisfaction and both intent to vote and report on actual vote (r = -.39). 28 The reported relationships between extrinsic job satisfaction and both intent to vote and report on actual vote appear consistent. Correlations range from r = -.21 (Hammer & Berman, 1981) to r = -.59 (Schreisheim, 1978). The majority of correlations fall between r =-.30 and r = -.40, however, In fact, the studies with more extreme values (Hammer & Berman, 1981; Schreisheim, 1978) together account for less than 4% (N = 182) of the total sample size cumulated across all studies (N = 4484) examining the relationship between extrinsic job satisfaction and both intent to vote and actual vote. This suggests that observed variation in correlations is due to sampling error. Perceived Union Instrumentality The instrumentality of union representation refers to the assumption, explicit in the expectancy theory of employee motivation (Vroom, 1964), that individuals perceive their actions in the work environment as leading (or not leading) to various outcomes. Galbraith and Cummings (1967) in their extention of expectancy theory, advanced the notion of first level or focal outcomes, which are perceived to be instrumental to the obtainment of second level outcomes. In the present context, unionization would represent a first level outcome that leads to certain second level outcomes; for example, increased pay and fringe benefits. 29 The perceived instrumentality of union representation for the obtainment of valued work-related outcomes has generally exhibited a strong relationship with unionization. In one study, Kochan (1979) examined the impact of the perceived instrumentality of unionization as a strategy for improving working conditions. He reported a correlation of r = .33 between perceived instrumentality and intent to vote for a union. DeCotiis and LeLouarn (1981) measured union instrumentality on an eight-item scale which assessed employee perceptions of the extent to which the presence of a union would result in better pay, benefits, working conditions, supervision, and fair treatment. They investigated the relationship between union instrumentality and both intent to vote and report on actual vote for or against union representation. A correlation of r = .67 was found for the relationship between union instrumentality and report on vote; the correlation with intent to vote was r = .76. Zonia (1983) reported the relationship between union instrumentality and unionization for three separate studies, conducted in 1977, 1978, and 1981. Union instrumentality measured the extent to which faculty perceived a union would improve their overall economic status; help aquire additional funds from the legislature; 30 give faculty greater involvement in decision-making; and provide greater job security. In the 1977 and 1981 studies, the relationship between perceived union instrumentality and intent to vote for a union was r = .72, .74, respectively. Zonia (1983) reported a correlation of r = .77 for the relationship between union instrumentality and report on actual vote for the 1978 study. In another study, Ahlburg (1984) investigated the importance of a five-factor measure of union instrumentality in predicting the individual's unionization decision. Union instrumentality included the following factors: "belief that the union will more effectively protect against deteriorating faculty employment conditions than the present decision-making process; be more effective for representing faculty economic interests than the present system; will not adversely affect promotional opportunities; will lead to higher salaries; and will not reduce the incentive for excellence in faculty performance" (Ahlburg, 1984, p. 16). Instrumentality was found to be a significant predictor of report on union vote using multiple regression analysis. Youngblood, et a1. (1984) assessed the impact of union instrumentality beliefs on unionization by asking individuals the extent to which they agreed or disagreed 31 (using five-point scales) that joining a union would result in a greater probability of occurrence of twenty outcomes: wages, improved benefits, protection from being discharged, creation of new jobs, substantial union dues, work disruptions, strikes, promotions, work stoppages, employer relocating, work hours, productivity, working conditions, fairness of treatment, employee-management relations, accidents, interesting work, better grievance handling, better health environment, and number of grievances. They found a strong relationship between perceived union instrumentality and intent to vote for a union (r = .73). Except for Kochan's (1979) findings, correlations between perceived instrumentality and employee unionization range from r = .67 (DeCotiis & LeLouarn, 1981) to r = .77 (Zonia, 1983). The perceived instrumentality of union representation appears as a consistently strong predictor of the individual's unionization decision. Intent to 2233 Few studies have examined the relationship between an individual's intent to vote for a union and their report on actual vote. In particular, many studies have used intent to vote as the sole criterion measure, presumably because investigators could not access a union 32 representation drive in progress to assess report on voting behavior. In one study, Getman, et a1. (1976) examined the degree to which behavioral intent to vote for a union actually predicted voting behavior. They reported a correlation of r = .73 between voting intentions and report on actual vote. DeCotiis and LeLouarn (1981) found a more moderate relationship between behavioral intentions and report on actual vote (r = .47). Zalesny (1985) reported a correlation of r = .82 between pre-vote intentions and report on voting behavior. The one moderate correlation found between voting intentions and report on vote (r = .47) was reported in the smallest sample study (DeCotiis & LeLouarn, 1981). Sampling error alone may account for this more moderate relationship. DeCotiis and LeLouarn's (1981) study constitutes only 6.1% (N = 95) of the total sample size (N = 1554) across all three studies examining the relationship between these two variables. Summary and Conclusions The individual's unionization decision can be operationalized as either a behavioral intention to vote for a union or as a report on actual vote for or against union representation. Regardless of how the individual's unionization decision is operationalized, the empirical 33 literature suggests that all the variables reviewed here are important predictors of employee unionization. However, the magnitude and consistency of reported relationships varies. The perceived instrumentality of union representation appears to be a consistently strong predictor of individual-level unionization. Extrinsic job satisfaction and perceived equity seem to be less strongly related to the individual's decision to unionize then union instrumentality. The relationship between equity perceptions and unionization, however, exhibits considerable variation across studies. The inconsistency in reported results may be due to sampling error, measurement error (all measures of perceived equity are ad hoc), or as previously mentioned, the type of equity measure may affect the magnitude of the relationship with unionization. The reported relationships between extrinsic job satisfaction and unionization are substantial and largely consistent. Wage level tends to correlate less strongly with individual-level unionization than the attitudinal predictors reviewed here. However, the magnitude of this relationship varies widely. It appears, though, that attitudes about one's pay are more important than the absolute level of wages. This is consistent with Farber and Saks (1980) finding that an 34 individual's location in the organization's wage distribution is more important than his or her absolute level of wages. Last, intent to vote for a union is a strong predictor of actual voting behavior. As can be seen from this summary, the research on the individual's unionization decision is relatively consistent. Although all the variables reviewed here are related to unionization, the direct nature of their interrelationship and how they influence the individual's decision to vote for or against union representation is still unclear. In particular, a model of the individual unionization decision, including the variables reviewed here, and direct tests of hypotheses deduced from such a model should be made. To accomplish this, more primary research need not be done; rather the existing empirical data reviewed here can be used for model testing. The construction of such a model will be the focus of the next chapter. CHAPTER THREE Hypotheses and a Model of the Individual Unionization Decision In this chapter, a model of the individual unionization decision and specific hypotheses concerning the relationship among variables in the model is described. In the next two chapters, a formal test of these propositions will be presented. The information in this chapter is organized in the following fashion. In the first section, a discussion of previous attempts at constructing a model of the individual's unionization decision is presented. The second section presents a synthesis of the literature review and the development of path analytic hypotheses on a theoretical and empirical basis. A process model of the individual unionization decision and a qualitative causal diagram is presented. Previous Models Much of the research on individual-level unionization has not been founded on explicit theoretical frameworks. Rather, the research process has been inductive, with researchers providing post hoc theoretical explanations for the empirical relationships they find. Although an inductive research strategy is useful in identifying relationships, there are limits to its usefulness in explaining why those relationships exist. 36 Attempts at advancing a theoretical framework to explain the individual's unionization decision are found in the economic literature (Farber & Saks, 1980), the industrial relations literature (Kochan, 1980), and the management literature (Brief & Rude, 1981; Youngblood, et al., 1984). These frameworks will be discussed in turn. Farber and Saks, 1980 This framework is based on the notion that if the expected utility of a job becoming a union job is higher than it not becoming a union job, then the individual will vote for a union. The major determinants of the utility function are hypothesized to be: the individual's relative wage level or position within the intrafirm wage distribution; the characteristics of the job as a union job and as a nonunion job which encompasses an individual's satisfaction with supervision and chances for promotion; and the subjective probability of the security of the job as a union job and as a nonunion job. Analyzing the Getman, et al. (1976) data, Farber and Saks (1980) found that one's location in the wage distribution is a strong predictor of voting behavior. Workers in the lower end of the distribution are more likely to vote for union representation. In addition, those individuals who felt that they could not replace their current job and felt dissatisfied with their job security were more likely 37 to vote for a union. Job characteristics also affected voting behavior. Those workers who were dissatisfied with their supervision and felt they had poor promotional opportunities were more likely to vote for union representation. Kochan, 1980 Kochan (1980) presented a model of the individual unionization decision that postulates three sets of determinants of voting behavior. He suggested that the initial stimulus to unionize arises out of dissatisfaction with the economic or bread and butter aspects of the job. That is, in contrast to Farber and Saks (1980), dissatis- faction with wages, fringes, and working conditions are thought to be more strongly associated with a desire to unionize than dissatisfaction with other aspects of the job, such as the supervisory relationship. Also, perceptions of inequities in the administration of the compensation system are viewed as an equally strong predictor of individual-level unionization as is dissatisfaction with the level of compensation. The second set of determinants influencing the unionization decision is the desire for participation or influence on the job according to Kochan (1980): While dissatisfaction with job conditions may provide the initial stimulus for unionization, not all workers are likely to turn immediately to unions as a way of coping with these problems. Some workers may 38 not desire greater participation on their job. Furthermore, among those who do desire greater participation or influence, only those who are unable to influence their work environment through more informal individualistic, or employer - initiated programs are likely to turn to unions as an alternative (p. 145). Also discussed is the degree of commitment to the job as an important factor influencing whether an individual will attempt to change their work environment or will find it simpler to leave the job and search for better alternatives elsewhere. The third set of factors thought to influence unionization is whether the benefits of unionization outweigh the expected costs of organizing. Costs and benefits include the general beliefs workers hold toward unions in general and the expected responses of their employers to unionization. Workers who hold negative images of unions as institutions would be reluctant to support unionization. Conversely, individuals who are ideologically predisposed toward unionization or who have favorable perceptions of the labor movement should be active advocates of unionization regardless of their work-related conditions. Another aspect of the costs and benefits of unionization is the beliefs that workers have about the instrumentality or success of unions in improving the working conditions of their members. Finally, in this framework, the most important factor influencing an individual's perceptions 39 of the instrumentality of union representation is whether one will be better off pursuing their concerns or goals on the job collectively rather than individually. Kochan (1980) draws support for this framework from his analysis of the 1977 Quality of Employment Survey data discussed in the previous chapter (Kochan, 1979). Brief and Rude, 1981 This model, essentially an extention of Triandis's (1971) theory of attitudes to the unionization context, contains two basic determinants of the individual unionization decision: worker attitudes toward unionization and general subjective norms toward voting for a union. Attitudes toward unionization are principally determined by the subjective probabilities that voting for a union will lead to better fringe benefits, more job security, and higher wages. General subjective norms toward voting for a union are a function of the worker's social environment, reflecting one's perceived expectations of salient referents and the motivation to comply with these referents. According to Brief and Rude (1981), salient referents include an individual's immediate supervisor and co-workers. Attitudes toward management are assumed to be captured in one's satisfaction with supervision. Subjective norms are assumed to be determined by individual characteristics, 40 although the specific characteristics that influence subjective norms is not stated. The authors provide no data to test their framework. Youngblood, 95 al., 1984 Similar to Brief and Rude (1981), Youngblood, et al. (1984) proposed a model of individual-level unionization (voting intention) that includes the following variables: reactions to the work environment, union instrumentality perceptions, labor union image, and subjective norms. For individuals to intend to vote for a union they must perceive low motivating potential in their work, be dissatisfied, perceive unions as instrumental, and not hold a negative image of labor unions in general. Three main effects are postulated: the trigger effect of the work environment resulting from the low motivating potential of one's work and job dissatisfaction, the augmenter effect of high union instrumentality beliefs, and the veto effect of an unfavorably perceived labor union image and subjective norms against unionization on individual voting intentions. Instrumentality beliefs, labor union image, and subjective norms were all found to be related to unionization. Youngblood, et al. (1984) did not find work environment variables, however, measured as the motivating potential of the job and overall job satisfaction, substantially related to voting intentions. 41 In summary, the theoretical frameworks discussed here look at similar variables: union instrumentality or union image perceptions, various facets of satisfaction with the work environment, some type of attitude regarding compensation, and the influence of referent others in the workplace. Still, the process the individual goes through in deciding to favor union representation remains unclear. There is no formal causal model or testable hypotheses that would lead researchers to determine the exact nature of the relationship among variables. For example, it is unclear in Kochan's (1980) model whether the three sets of determinants influencing the individual's unionization decision operate sequentially or simultaneously. In the following section a path model of the individual's unionization decision and testable hypotheses will be advanced. A Model of the Unionization Decision and Specific HyEEEhEsEE In this section of the dissertation, a process model of the individual unionization decision is presented, followed by a causal model and specific path analytic hypotheses concerning the relationship among variables. In Figure l, a process model of the individual unionization decision is presented. The process model begins with characteristics of the individual and the characteristics of their job. 42 Individual characteristics are conceptualized as relatively stable and unchanging differences between individuals; for example, sex and race. Job characteristics are those factors largely determined by the nature of the job such as one's level of pay. Individual and job characteristics determine an individual's perceptions of equity. The perceived degree of equity in turn determines an individual's level of satisfaction. If an individual is satisfied, then the unionization process will not proceed beyond this stage. However, to the extent an individual is dissatisfied then a search process begins in an effort to determine those alternatives available to remedy one's feelings of dissatisfaction. The most likely alternative may be the search for a better job, which may ultimately lead to turnover. The degree to which an individual perceives their alternatives as limited and having little expected utility or instrumentality, then the notion of union representation may become salient. The perceived instrumentality of unionization may be affected by such factors as an individual's age, tenure, work group, previous experience with unions, and fear of employer retaliation (Block & Premack, 1983). 43 Individual Job Characteristics Characteristics Iquity Porcoptions Instrumentality f O Unionization Intent -—“ ’IO’W‘OD to Unioniao Vote Figure 1. A Process Model Of Individual-Level Unionization 44 If an individual is young and has little tenure, dissatisfaction may lead to more searching behavior which in turn may lead the individual to find another job. The greater an individual's age and tenure, however, the less feasible the option to leave the current job, and the greater the instrumentality of union representation. To the degree a worker has had an unfavorable previous experience with unions, perceives there will be employer retaliation for favoring union representation, and belongs to a work group that exerts norms against unionization, then the individual will be unlikely to perceive union representation as instrumental for obtaining valued outcomes in the work environment. If an individual feels that a union will be useful for increasing work outcomes and reducing dissatisfaction then one will likely intend to vote for a union. Finally, the intention to vote for a union will lead to pro-union vote. Based on the review of the empirical individual-level unionization literature presented in Chapter 2, this process model is operationalized into a qualitative causal or path model. This path model shown in Figure 2 is to be tested in the dissertation research. Although reviews of the individual-level unionization literature (Block & Premack, 1983; Heneman & Sandver, 1983) have noted that the variables included in the model, 45 i.e. wage level, equity perceptions, extrinsic job satisfaction, satisfaction with administrative procedures,. perceived instrumentality of union representation, and intent to vote for a union, are all related to union vote, the relationship among these variables has never been examined. For example, while the relationship between an individual's wage level and union vote has been documented (Farber & Saks, 1980), the relationship between wage level and equity has not been examined in the unionization literature. Thus, support for these antecedents of the individual unionization decision has been indirect at best. The dissertation research will directly assess the causal linkages between these variables. Direct causal relationships among variables are proposed as hypotheses in the path model. This model draws in part upon three streams of research; equity theory as articulated by Adams (1965), expectancy theory as formulated by Vroom (1964), and Fishbein's theory of behavioral intentions (Fishbein, 1967). Equity theory, as advanced by Adams (1965) and Weick (1966) is the most widely known and fully articulated in a series of social comparison theories of motivation (Goodman, 1977). Social comparison theories all focus on individuals' perceptions of how fair they are being treated compared to others (Mowday, 1979). It is assumed 46 Wage Laval Pay Equity Parceptiona Extrinaic Satiafaction Batiafaction with Adminiatratlon Inatrulantallty of Unionization Intent to Unioniao 'Figure 2. A Causal Model Of Individual-Level Unionization 47 that individuals engage in a process of evaluating their social relationships much like they would evaluate financial transactions in the marketplace. Social relationships are viewed as an exchange process in which individuals make contributions or investments and expect certain outcomes in return. March and Simon's (1958) inducements-contributions theory is one such example. Individuals are thought to have certain expectations about the outcomes they receive as a result of their contributions of time and effort. It is also assumed that individuals do not assess the equity of an exchange in a vacuum. Rather, they compare their own situation or exchange balance with others to determine the relative balance and the extent to which an exchange is satisfactory. The theory suggests that this social comparison process proceeds in the following fashion. First, individuals weigh their various inputs and outcomes according to their perceived importance. Inputs are defined as anything that contributes to the work setting such as prior work experience or level of effort. Outcomes are all factors one perceives as valued rewards received from the work environment; for example, positive feedback and financial rewards. Second, workers arrive at a ratio of their outcomes to inputs as compared to the 48 ratio of another individual's or groups outcomes and inputs. The comparison individual or group, termed the referent source, may be individuals or groups with whom one engages in direct exchanges, or individuals engaged in exchanges with a common third party. Potential referent sources are virtually unlimited and may also include: the profession, relatives, neighbors, a craft group, labor market patterns, and so on. Sources are assumed to be relevant referents only if they possess one or more attributes which are comparable to those of the individuals involved in the equity process. Adams (1965) is not more precise in specifying how the referent source may be identified, although he did assume that co-workers are commonly used. The referent source becomes the individuals point of comparison in determining their degree of equity. Finally, inequity is said to exist when the individual feels their ratio of inputs to outcomes are not what they should be in comparison to their referent source. In order for inputs and outcomes to be considered in evaluating exchange relationships, they must be both recognized and considered relevant by the participants. Shared concepts of what are fair relationships between outcomes and various inputs are learned as part of the overall socialization process. Not surprisingly, research 49 on equity theory has focused largely on pay as the relevant outcome variable. Adams (1965) also hypothesized that perceived inequity should lead to certain motivational consequences. An individual could reduce the inequity by: altering inputs, altering outcomes, cognitively distorting one's own inputs and outcomes, acting on the inputs or outcomes of the referent source, changing the referent source, or leaving the field. A number of theoretical questions have been raised regarding equity theory. For example, it is unclear whether equity predictions generalize to other than economic rewards (Mowday, 1979). Research on equity theory has established, however, that when individuals experience underreward inequity, they will be dissatisfied with the economic aspects of their employment (Goodman, 1974). In fact, job satisfaction research (Herzberg, Mausner, & Snyderman, 1959) has found that feelings of inequity or unfairness are the most frequently reported source of job dissatisfaction. Furthermore, this dissatisfaction will be directed at the controller of the reward structure, the employer (Prichard, 1969).1 Thus, it is reasonable to propose the following hypotheses: 1 One implication of maintaining pay secrecy is that it may reduce perceived inequity. 50 Hypothesis 1: The lower an individual's wage level the more likely he or she is to believe other individuals working at similar jobs have a higher wage level which in turn would lead them to feel their compensation was inequitable. Hypothesis 2: The greater the extent to which individuals perceive their wage level as inequitable the greater their dissatisfaction with the extrinsic aspects of their employment Hypothesis 3: The greater the degree of extrinsic dissatisfaction, the greater an individuals dis- satisfaction with administrative procedures. Research on equity theory has failed to resolve the issue of what equity reduction strategy will be chosen by an individual in a given situation (Mowday, 1979). The process of evaluating and selecting alternative strategies is predicted to follow Vroom's (1964) notion of instrumentality. Instrumentality is the perceived contingency an outcome has for the obtainment of other valued work outcomes. Thus, the instrumentality of unionization is the degree to which a union is seen as a means to obtain valued outcomes in the workplace; outcomes that the individual may not be able to obtain independently. The guiding philosophy of the American trade union has been business unionism, in which unions have functioned as pragmatic organizations, attempting to improve the short run goals of their membership. American unions have concentrated on improving economic or "bread 51 and butter" outcomes through collective action. Research on the individual unionization decision has found no evidence of any widespread anti-union sentiment among unorganized employees. Rather, workers turn to unions when they perceive them as instrumental to the improvement of their work-related goals (Block & Premack, 1983). The following hypotheses can be proposed: Hypothesis 4: If an individual feels dissatisfied with administrative procedures they will be likely to consider the instrumentality of unionization as a means to increase their satisfaction. Hypothesis 5: The greater the perceived instru- mentality of unionization for remedying dissatis- faction through the obtainment of economic outcomes the more likely an individual is to intend to vote for representation by a union. Fishbein's (1967) theory of behavioral intentions suggests that the most reliable attitudinal predictor of an individual's behavior is their intention to engage in that behavior. Hypothesis 6: To the extent an individual intends to vote for a union this in turn should lead one to vote for a union should a representation election occur. Summary A summary of the hypotheses presented in this chapter is listed below: 1. The lower an individual's wage level the more likely he or she is to believe other individuals working at similar jobs have a higher wage level which in turn would lead them to feel that their compensation is inequitable. 52 The greater the extent to which individuals perceive their wage level as inequitable the greater their dissatisfaction with the extrinsic aspects of their employment. The greater the degree of extrinsic dissatisfaction, the greater an individuals dissatisfaction with ad- ministrative procedures. If an individual feels dissatisfied with administra- tive procedures they will be likely to consider the instrumentality of unionization as a means to increase their satisfaction. The greater the perceived instrumentality of unionization for remedying dissatisfaction through the obtainment of economic outcomes the more likely an individual is to intend to vote for representation by a union. To the extent an individual intends to vote for a union this in turn should lead one to vote for a union should a union representation election occur. These hypotheses are advanced to provide a deductive rationale for the path model and directly examine the causal antecedents of the unionization process at the individual level of analysis. The research methodology used to test these hypotheses is presented in the next chapter. (CHAPTER FOUR Research Methodology Overview Six variables will be investigated to assess their relative causal impact on the individual's unionization decision: wage level, equity perceptions, extrinsic job satisfaction, satisfaction with administration, perceived instrumentality of union representation, and intent to vote for or against union representation. Data on the relations among this set of variables was derived through meta-analyses of the results of individual-level unionization studies. The following sections describe the method of review, study characteristics, and the procedure and type of analysis used. Method of Review In order to examine the causal relationships between variables in the path model, studies were collected for review using the following criteria: Pearson correlation coefficients or statistics capable of being transformed into point-biserial correlations were reported,1 sample size was reported, and a measure of unionization was treated as a criterion. 1 Formulae for the transformation of various statistics into point-biserial correlations is presented in Hunter, et a1. (1983). 54 Employing these criteria, an extensive review of the literature was conducted, using a manual search of the psychological, sociological, economic, industrial relations, and personnel journals, complete searches of references cited in relevant published reports, and unpublished studies known to the author. Since the task of the dissertation is to determine direct causal relations among the variables in the path model, the complete intercorrelation matrix for the set of variables investigated in a given study was required. Unfortunately, in several studies in this research literature, only the correlations between the predictors investigated and unionization was presented (Ahlburg, 1984; Dworkin & Lee, 1985; Getman, et al., 1976; Kochan, 1979; Maxey & Mohrman, 1980; Schriesheim, 1978; Zalesny, 1985). These results will only permit the establishment of one column in the correlation matrix to be derived from the data. Because the reporting of the correlation matrix is essential to the data collection process, where the correlation matrix was not reported for a given study, the principal investigators were contacted by telephone in an effort to obtain this data. There were, however, two exceptions to this data collection procedure in which investigators were not contacted. One study reported a 55 very low sample size (N = 59) in comparison to the other empirical studies (Schriesheim, 1978). Another study examined only one predictor of unionization included in the path model (extrinsic job satisfaction), so that the correlation matrix would not be required (Maxey & Mohrman, 1980). A total of six investigators were contacted. In two cases the correlation matrices for the particular studies were sent to the author (Dworkin & Lee, 1985; Zalesny, 1985). For the two largest data sets (Getman, et al., 1976; Kochan, 1979) the author was able to obtain the original data tapes and generate the correlation matrices from the raw data. In both of these cases, attitude scales were constructed and reliabilities computed. Only one author failed to respond after a follow up contact (Ahlburg, 1984). Once the correlation matrices were obtained, studies were coded according to the following characteristics: sample size, correlation coefficients, type of measure used, reliability coefficients, and respondent's age, and tenure. It is worthwhile to note that the causal linkages among variables presented in the path model have been investigated in other research literatures. The relationship between wage level and perceived equity, for example, has been the subject of considerable research in the equity theory literature (e.g., Mowday, 1979). The 56 use of other research literatures, however, may yield a different range of variation than is found in the unionization literature; differences in the context in which the research was conducted may impact on means and standard deviations. Although correlations tend to remain robust, they may be moderated to an unknown degree by contextual differences. The majority of equity theory research, for example, has been conducted in a laboratory setting (Mowday, 1979). Study Characteristics Ten published and four unpublished studies were found which assessed the relationship among one or more of the predictors of unionization relevant to the path model. The use of unpublished studies minimizes any potential "file drawer problem" (Rosenthal, 1979). The file drawer problem refers to a potential selection bias that may occur in published studies, where significant or expected findings have a better chance of being published than correlations that are of low magnitude or not in the predicted direction. Table 1 shows the characteristics of the studies used to assess the relative causal impact of the predictors investigated. These studies are listed by number to conserve space. Full citations for the numbered studies are found in Appendix A. Sample sizes ranged from 59 to 1239. A total of 5378 individuals participated in 57 .coauoaoo mourning no: anon» “nonnoauo> mo amass: o co mosao> woman‘s 0» mac uoHHoEn noawuoeon mum: nooaaocououue goouoo on» now can: ensue nausea n .UNH can tnwa .oNH ”nouosun :omuon«coucs auouomoo noun» vouuoaou «A upsum o mhmm w A." h w a N o Anuoa Nun x x x ON." noN x x x ANA hvm x x x QNH wmm x x x x AH sav x x x 5H mm x x x a de x x a «an x x x x x h mad x x x x o mmwa x x x x x m sum x x v mm x x x x x n NNN x x x N aNN x x x H wwwm ouo> co wumc0wc: auwunuc089uuocu couumuunficwfipc couuuounfiuom auasvm Ao>oq Hanan: nwamsnm uuommm cu ucmucH come: cu“: coauoounwuow non ouncmuuxm and «on: napsum nanafioc 0» acmucH sa.>.m.m mass. mm. sm.umn. e suwamucmssuumcH :0wca Ha.w.m uses. an. os.5oo. m cofiumuumflewscc saws cofiuonumwunm Ha.h.m.m mmaa. an. mn.5eo. e eofluommmoumm non omewuuxm H 5 me. u H sunswm and manEsz mmnunfiwnmname sunaanmwdmm mmcmm mucmwonuumoo «Snow. hpsum mo wocmwum> com: aufiawanHmm we nonspz c0wunscmuu¢ now mewuowuuoo cw poms mmwuwafinmwamm wamom mo oucmwum> com com: N manna 63 Path analysis or "structural model estimation" is used to test the proposed path model. Path analysis is a procedure for systematically combining the use of partial and multiple correlation to study the causal relationships between a set of variables. In path analysis any variable may be both an independent and dependent variable; a variable is used to explain the relationship with another variable and in turn is explained itself in terms of other variables to which it is related (Hunter & Gerbing, 1982). Since the causal model proposed here is unidirectional, the path analysis is recursive. And because wage level is not explained in terms of its causal antecedents it is an exogenous or "starter" variable. All other variables are endogenous; their causal antecedents are specified within the system. The system refers to the set of all variables in the model. The procedure for path analysis proceeds in three stages: (1) the construction of the model; exogenous variables are inscribed in a box while endogenous variables are denoted by a circle, (2) estimate the path coefficients from the correlation matrix and apply them to each arrow in the model in order to explain the strength of the causal impact, and (3) test the fit of the path model to the data. CHAPTER FIVE Results In this chapter the results for the tests of the hypotheses are presented. First, the correlation matrix and information on the meta-analyses used to derive the correlations is reported. This is followed by the results of the path analysis testing the model of the individual unionization decision presented in Figure 2. In the latter half of this chapter a largely exploratory path analysis is conducted to determine if an alternative model will better fit the data. A description of the support or lack of support for the hypotheses is presented. Path Analysis of the Unionization Decision Table 3 presents the basic correlational data essentially free of the effects of sampling error and error of measurement. Even when a meta-analysis was not possible because multiple study correlations were not available, correlations were corrected for attenuation. This was the case for the following relationships: wage level - pay equity, pay equity - extrinsic job satisfaction, pay equity - union instrumentality, and wage level - satisfaction with administration. Table 4 presents information on the relationship among variables in the correlation matrix. Included in Table 4 is the total sample size corresponding to each correlation, the 95 percent confidence interval around which the 65 Table 3 Intercorrelations of All Variables Free of the Effects of Error of Measurement and Sampling Error Variable l 2 3 4 5 6 l. Wage Level 2. Pay Equity .05 3. Extrinsic Job Satisfaction .11 .42 4. Satisfaction with Administration .12 __ .69 S. Union Instrumen- 6. Intent to Vote -.16 -.25 -.44 -.60 .60 7. Union Vote -.12 ;_, -.45 -.60 .60 .79 Note. 109 < N < 2884 66 Table 4 Information on the Relationships Among Variables in the Correlation Matrix Total 95% Sample Number of Confidence Study Relationship Size Correlations Interval Numbers WL—UV 1226 2 -.12 < -.12 ( -.12 5,6 SA-UV 981 S -059 < -060 < -061 3,6,9, 11,12b 9,11 IV—UV 1713 3 .77 < .79 < .80 3,5,11 WL-PE 1011 l .00 < .05 < .10 7 WL-UI 1826 3 -.07 < -.08 < -.09 5,7,10 7,10 PE-UI 896 l .00 < -.03 < -.06 7 PE-IV 1024 2 -.25 < -.25 < -.26 1,7 ES-IV 2650 6 '044 < -044 < -044 2,3,5, 7,8,11 UI-IV 2391 6 .53 < .60 < .68 3,5,7, 10,12a,12c SA’IV 1668 5 -060 < -060 < -960 3,4,11, 123,12c WL-SA 109 1 ' .00 < .12 < .24 6 SA-ES 659 3 .69 < .69 < .69 3,6,11 SA-UI 966 4 -.55 < -.59 ( -.62 3,123,12b,12C WL=Wage Level; PE=Pay Equity; ES=Extrinsic Satisfaction; SA=Satisfaction with Administration; UI=Union Instrumentality; IV=Intent to Vote; UVIUnion Vote. 67 correlation can be found, and the studies the correlations are obtained from. Results for each correlation derived from the meta-analysis tend to exhibit little variance across studies. This reflects the fact that these studies have a much higher average sample size than is found in many other research literatures; a total of 5378 individ- uals across 14 studies or an average sample size of 384. This is large in comparison to the average sample size of 68 found for validity studies in the personnel selection literature (Lent, Aurbach, & Levin, 1977), where the Hunter, et al. (1982) meta-analytic procedure was first applied. Confidence intervals are computed rather than significance levels because the Type I error rate (e.g. 5 percent of less) associated with significance testing is guaranteed only if the null hypothesis that the true population correlation is zero is true. If the null hypothesis is false, then the error rate becomes a Type II error rate that may be as high as 95 percent (l-.05). Confidence intervals are, in general, superior to significance tests for at least three reasons. First, the confidence interval is correctly centered about the observed or mean correlation rather than the hypothetical value of the null hypothesis. Second, the confidence interval shows the level of uncertainty surrounding the 68 sample statistic; the width of the confidence interval. In contrast, significance tests treat all values above the minimum cutoff needed to achieve statistical significance as the same. It may be, for example, that while one correlation is found to be significant and another not significant, the confidence intervals overlap and the nonsignificant correlation may be the result of Type II error. Statistical significance is based on both effect size and sample size. Finally, the confidence interval gives the researcher an accurate picture of the degree of uncertainty in small samples. Appendix C presents the complete results of the meta-analyses used to derive the correlation matrix. Included for each meta-analysis are the studies from which correlations were obtained, the total sample size used in each meta-analysis, the number of correlations meta- analyzed, the mean correlation corrected for statistical artifacts, the observed variance, the variance corrected for statistical artifacts, the percent of variance due to statistical artifacts (sampling error and measurement error), and the results of a chi square test which assesses whether the corrected variance is greater than that expected by chance alone. An insignificant amount of corrected variance shows that the observed variance is due exclusively to artifactual error. 69 Since pay equity and satisfaction with administration are broken out separately, missing values remain for the relationship between pay equity - satisfaction with administration and pay equity - report on vote (see Table 3). This meant in order to conduct the path analysis, pay equity was deleted from the model of the individual's unionization decision. The ordinary least squares estimates (OLS) of the path coefficients for this model are presented in Figure 3. Table 5 presents the information needed for assessing the path analysis which includes the observed correlations, the reproduced correlations (predicted values), and the error matrix. This model of the unionization decision fails the data dramatically. The largest error expected by chance alone (p =.05) is d* =.08 with an average sample size of 1534. Seven out of ten errors are individually significant at or beyond the .05 level. A chi square test used to assess the overall fit of the model to the data is also significant, x2 (10) =219.74, p < .001. One disconfirmed assumption of the original model is that an individual's intent to vote for a union and report on vote are different phenomena. According to the product rule, correlations between intent to vote and report on vote with the other variables in the path model should be 21 percent lower, instead they are virtually identical. 70 The data in Table 5 shows no substantial dropoff in the correlation from intent to vote and report on vote. The obtained correlation of .79 can be interpreted as a test-retest reliability; both measures represent random variation in voting intentions over time. Since both intent to vote and vote proxy the same underlying variable, a more stable measure of the unionization decision that eliminates random error over time is achieved by combining both measures. The OLS estimates of the path coefficients for the model with intent to vote and report on vote combined are presented in Figure 4. Table 6 presents the observed and reproduced correlations, and the error matrix. In this model d* = .08 at the .05 level with an average sample size of 1476. As can be seen from the error matrix, three out of six errors are significant at or beyond the .05 level. Also, the overall goodness of fit of the model to the data is still rough, X2 (6) = 85.04, p < .001. These results show that although the hypotheses advanced in Chapter 3 were largely supported, the original path model does not fit the data. At this point an effort was made to determine post hoc the model which best fits the data. Figure 5 presents this model. In this model the OLS estimates of the path coefficients consist of both zero-order correlations 71 Wage Level Extrinsic Satisfaction Satisfaction with Administration Instrumentality of Unionization .79 Figure 3. The Quantitative Causal Diagram Derived From The ' Original Model of Figure 2 72 Table 5 The Assessment of the Path Analysis for the Model in Figure 3 Observed Correlations Wage Level 1.00 Extrinsic Job Satisfaction .11 1.00 Satisfaction with Administration .12 .59 1.00 Union Instrumen- Intent to Vote -.16 -.44 -.60 .60 1.00 Reproduced Correlations Wage Level 1.00 Extrinsic Job Satisfaction .11 1.00 Satisfaction with Administration .06 .59 1.00 Union Instrumen- Union Vote —.02 -.16 -.28 .47 .79 Errors (Observed-Predicted) Wage Level Extrinsic Job Satisfaction Satisfaction with Administration .06 Union Instrumen- tality -.04 .01 Union Vote -.10 -.29 -.32 .13 Note. Average N = 1534 73 Wage Level Extrinsic Satisfaction Satisfaction with Administration Instrumentality of Unionization Unionization Decision Figure 4. The Quantitative Causal Diagram For Intent To Vote And Report On Vote Combined. 74 Table 6 The Assessment of the Path Analysis for the Model with Intent to Vote and Report on Vote Combined Observed Correlations Wage Level 1.00 Extrinsic Job Satisfaction .11 1.00 Satisfaction with Administration .12 .59 1.00 Union Instrumen- tality -008 -034 -o59 1.00 Unionization -.15 -.47 -.63 .63 1.00 Reproduced Correlations Wage Level 1.00 Extrinsic Job Satisfaction .11 1.00 Satisfaction with Administration .06 .59 1.00 Union Instrumen- tAIIty -00‘ -035 -059 1.00 Unionization -.02 -.22 -.37 .63 1.00 Errors (Observed-Predicted) Wage Level Extrinsic Job Satisfaction Satisfaction with Administration .06 Union Instrumen- tality -004 .01 Unionization -.13 -.25 -.26 Note. Average N - 1476 75 between the predictors in the model and the beta weights (standardized regression coefficients) for the relation- ship between the predictors and unionization. Table 7 presents the information needed for the evaluation of the final path model. For this causal model, d* = .10 at the .05 level with an average sample size of 1075. The largest error in the matrix presented in Table 7 is .06 which is not significant at the .05 level. The overall chi square for goodness of fit statistic was also not significant, X2 (3) = 2.04, Rafi; Summary In summary, the six hypotheses were generally supported by the path analysis. The path analysis does indicate, however, that there is a model that fits the data better than the original model. These results and their implications will be discussed in the next chapter. 76 «one one muss noses Hana: Hanson «sauouauenao one cauusuwcouca mo auuasucossuuucu mm.l codauuon coausNLcouca caduceuswcuaos sud: eoquuauuauom mm. cauuususqu¢m cascauuxm .m ouomam HA. Hosea was: 77 Table 7 The Assessment of the Path Analysis for the Model in Figure 5 Observed Correlations Wage Level 1.00 Extrinsic Job Satisfaction .11 1.00 Satisfaction with Administration .12 .59 1.00 Union Instrumen- tality “.08 -034 -059 1.00 Unionization -.15 -.47 -.63 .63 1.00 Reproduced Correlations Wage Level 1.00 Extrinsic Job Satisfaction .11 1.00 Satisfaction with Administration .06 .59 1.00 Union Instrumen- tality -004 -035 -059 1.00 Unionization -.12 -.47 -.63 .63 1.00 Errors (Observed—Predicted) Wage Level Enxtrinsic Job Satisfaction Satisfaction with Administration .06 Union Instrumen- tality -004 001 Unionization -.03 0 0 Note. Average N = 1075 CHAPTER S IX Discussion In this final chapter a discussion is presented concerning the degree of support for the original hypotheses, the possible reasons why an alternative model fit the data, the theoretical and applied implications of this research, the limitations associated with this study, and the directions that future research in this area might take. Modeling the Individual Unionization Decision The basic model of the individual's decision to vote for or against union representation originally proposed (see Figure 3) has two parts in (1) the sequential progression hypothesized among the predictors in the model and (2) the relationship between the predictors and unionization. The basic sequential progression among the predictors was supported, although the relationship between wage level and extrinsic job satisfaction is weak (r = .11). The relationship between the predictors and unionization, however, indicates that while some employees go through each step of the sequential decision process before deciding to unionize, other individuals decide to unionize solely as a result of their wage level, others also consider their extrinsic job satisfaction, their satisfaction with administration, or their beliefs about union instrumentality. 79 The final hypothesis presented in Chapter 3, that intent to vote for a union predicts pro-union vote, was the hypothesis that had the most a priori support from other research literatures. For example, in the organizational behavior research on employee turnover, intent to leave has been found to lead to turnover (Mobley, 1982). Although intent to turnover on the average is the best predictor of turnover, many who intend to leave do not because of a number of constraints; for example, lack of alternative employment opportunities and restricted mobility. In the research presented here, the correlations between report on vote and the predictors in the basic path model presented in Figure 3 were not lower than those for intent to vote. Rather, the correlations between intent to vote and report on vote with the predictors in the model were perfectly parallel (Hunter & Gerbing, 1982), instead of 21 percent lower as predicted by the product rule (see Table 3). This indicates that the relationship between intent to vote and vote represents one of measurement rather than causality. Thus, both variables measure (or proxy) the same underlying construct but at different points in time. The obtained correlation of .79 between intention and vote represents transient fluctuations in the individual's decision over time. The 80 decision to join a union in the research studies cumulated here is measured at two different time periods; intent to vote for a union before the election is held and a report on vote after the election has taken place. Measuring the same construct at two different time periods and correlating the measures represents a test-retest reliability in measurement theory. The correlation of .79 reflects the stability of unionization intentions over time. The path model which fits the data (see Figure 5) shows that all four predictors of unionization (wage level, extrinsic job satisfaction, satisfaction with administration, and union instrumentality) predict an individual's unionization decision even when considered jointly. For some individuals, a low wage level will be enough to cause them to vote for a union. Other employees, however, must be dissatisfied with extrinsic working conditions before they will favor union representation. Still others must be dissatisfied with the way the organization is administered before they will decide to vote for a union. Finally, some individuals must perceive a union as instrumental for achieving valued work-related outcomes. One interpretation of this finding is that different workers have different criteria for when they endorse the union. This model suggests different 81 path models for different individuals. Thus, one model does not explain how all employees decided to unionize. And since so few individuals decide the outcome of a representation election, the final model presented here may have limited usefulness in predicting the outcome of an election. One explanation for multiple paths can be obtained by theorizing that individuals differ in cognitive complexity. Individuals with high cognitive complexity may need to think through issues more thoroughly before they are willing to reach a decision. A complementary explanation for this threshold of voting behavior considers an individual's attitudes towards unions before entering the work organization. For an employee holding strong pro-union attitudes before entering the organization, a low wage level may be enough to lead to the decision to favor union representation. Those who hold less favorable attitudes toward unions may have to evaluate their level of extrinsic job satisfaction before deciding to vote for a union. Individuals with still less favorable attitudes toward unions may need to determine if the employer or administration is to blame for unsatisfactory working conditions. For those with the least favorable attitudes toward unions, there must be the belief that a union will be instrumental for the obtainment of work-related rewards. 82 The final model which fits the data supports the independent role that Farber and Saks (1980) give to wages in their model of the individual unionization decision. An individual's wage level was found to have an effect on their unionization decision in this research, although it is by far the smallest effect size of all the predictors examined (B = —.07). The third strongest predictor of unionization, extrinsic job satisfaction, has over twice as large an influence on unionization (B = -.15) as does wage level. However, an individual's location in the intrafirm wage distribution, indicated by Farber and Saks (1980) to be the relevant way to assess the impact of wages on individual-level unionization was not directly investigated in this research. Finally, this model lends support to Kochan's (1980) assertion that individuals who are ideologically predisposed towards unions in general should be active advocates of unionization regardless of their work-related conditions. Implications There are a number of potential implications for the research findings. There are implications for practice, public policy, and research in labor and industrial relations. In terms of practice, this research has implications for organized labor. Unions should target organizing 83 campaigns to where perceptions of the instrumentality of union representation are high and satisfaction with ’ administration, extrinsic job satisfaction, and wage levels are low. Although not directly tested in the" research presented here, it may be that organizing campaigns would be more effective in influencing employee satisfaction with administration and perceived instrumentality of union representation (and subsequent voting behavior) by focusing organizing efforts where cognitive complexity is high. These individuals may be more likely to attend to and evaluate information related to organizing efforts. Cognitive complexity could be proxied by educational status; for example, university faculty would presumably have a high degree of cognitive complexity. For individuals with a lower degree of cognitive complexity, organizers should focus on those with lower wages and negative attitudes toward working conditions. Policy implications also exist regarding the conduct of the election campaign. Getman, et a1. (1976) concluded that the conduct of the parties (the union and employer) during the organizing campaign had little impact on the way individuals voted in representation elections. This finding may not apply to all groups of employees. The behavior of the parties during the election campaign may 84 have a substantial impact on those individuals high in cognitive complexity who will evaluate the instrumentality of union representation and who do not already hold strong pro-union attitudes. These employees will be likely to assess the information made available during the election campaign in deciding whether the union can provide work-related outcomes that they value. The average representation election is decided by approximately 8 votes and employer won victories are decided by an even narrower margin (Roomkin & Block, 1981). Thus, the undecided voters may be critical to the outcome of an election. For public policy, the "laboratory conditions" requirement of the NLRB discussed in Chapter 2 may be a useful policy in insuring that this group of individuals has access to accurate information from both parties. The path model that was validated in this research has an implication for managements who wish to remain union free. Contrary to conventional wisdom, simply keeping employees satisfied is not enough to prevent them from seeking union representation. Finally, the implications of this study for conducting research on the individual unionization decision are substantial. Because the relationship between intent to vote and vote is one of measurement rather than causality, it is appropriate to treat intent 85 to vote as a valid criterion measure. This finding conflicts with the traditional notion in the individual-level research literature that intent to vote is an "imperfect proxy" for report on vote, which represents a hard (accurate) criterion measure. This notion reflects two basic ideas. First, that an attitude cannot be measured as adequately as a behavior, or in the present research a report on a behavior. Second, that an attitude or intention to perform some behavior does not predict the behavior as accurately as observing the behavior itself; you cannot trust what people say; only what you see them do. In some research literatures the latter concern may be warranted. For example, for individual-level turnover research, situational constraints (e.g., labor market factors) may intervene between intent to leave and actual turnover. However, given the secret ballot, employee intentions directly translate into voting behavior. Even the role of fear of employer retaliation for organizing efforts or other actions by the parties in the election process, while potentially affecting one's instrumentality perceptions, do not affect voting behavior once voting intention has been stated. So entrenched is the notion that intention is a weak proxy for report on vote that Heneman and Sandver (1983), in their review of the individual-level 86 unionization literature, excluded all research that used intent to vote as a criterion measure. Dworkin and Lee (1985) also assumed a separate model for report on vote and intention to vote. Finally, in contrast to current thinking on the unionization decision (e.g., Block & Premack, 1983), wage level is a poor proxy for work-related attitudes. As evident from the final path model in Figure 5, wage level does not correlate well with the other variables in the model and it is not important in the process of deciding to join a union. Limitations This research is not without its limitations. First, the scope of this study has limited generalizability for union growth per se; individual-level unionization decisions may or may not translate into changes in union membership. Representation elections are decided by a majority of employees in the proposed bargaining unit. In addition, a substantial number of employees become union members due to the impact of union security provisions contained in collective bargaining agreements. Second, the variables examined in this study are a function of the characteristics of the research literature. A limited set of the potential predictors of the individual's decision to vote for union representation 87 is investigated in this literature. The scope of a path model is increased to the extent all relevant causal antecedents are investigated. And third, macro variables have been identified as potentially important in understanding the unionization process (Block & Premack, 1983; Heneman & Sandver, 1983). This research, however, is unable to directly examine their relative influence on the individual unionization decision. Future Research There are several directions for future research on the individual unionization decision. One such area is the investigation of additional predictors that may be important in the process of individual-level unionization. Two variables are suggested for further research: (1) cognitive complexity and (2) attitudes toward labor unions at the time of entry into the workplace. Different predictors may be important for the unionization decision based both on attitudes towards unions and cognitive complexity. Consequently, four types of people may need to be considered with a differing causal model and correlation matrix appropriate for each. Further research would be required to test this hypothesis directly. Support for the hypothesized role of cognitive complexity and attitudes toward unions would explain why the path (model presented in Figure 5 fits the data. 88 Another variable potentially affecting the unionization decision may be the degree of confidence one has in management. To the extent one's confidence in management is high they may be little influenced by the predictors investigated here. Confidence in management, however, has never been investigated in the individual-level unionization literature. Also, as a consequence of the unionization process, if a union wins the representation election the work environment may change substantially. However, what if an individual does not feel that change is beneficial? Generalized beliefs that employees have about the desirability of change may influence voting behavior. A union represents a change in the work environment, while the employer may represent stability or at least a slower rate of change. Those employees who in general hold negative beliefs about the desirability of change may be more likely to vote against the union. Another area for consideration in future research is the nature of the election itself. The type of election being held implies different conditions surrounding the election process that may serve to influence individual voting behavior. For example, the role of employer resistance is no doubt greater in a regional director or NLRB ordered election than in a consent election. Single 89 versus multi-union elections factor the role of interunion comparisons into the election process. Initial elections may differ from rerun and repeat elections. Unfortu- nately, none of these distinctions have been well reported in the individual-level research literature. The degree of variation in election type was not able to be determined in this research. Future research should report the type of election so that meta-analysis might be performed to determine if the unionization decision differs based on the type of election conducted. An additional future research area entails the re-examination of the paradigm surrounding individual— level unionization research. By focusing exclusively on why workers join unions, researchers have ignored why workers fail to vote for union representation or fail to vote at all in a representation election. There are three choices in a representation election: yes, no, and abstain. Research on the unionization decision has only investigated one choice; the decision to vote for a union. A logical extention of the research conducted for this dissertation is the exploration of the factors that shape the predictors investigated here; predictors that in turn affect the unionization decision. Specifically, what characteristics of the individual and the work environment determine extrinsic job satisfaction, satisfaction with 90 administration, and union instrumentality perceptions? What characteristics of the individual, the organization, and the labor market determine one's wage level? Under- standing the role of factors that cause the predictors of the unionization decision to function as they do will increase our understanding of the context within which individuals determine a need for union representation. APPENDICES Study 1 Appendix A Studies Used in the Meta-Analysis Allen, R.E. & Keaveny, T. J. (1981). Correlates of university faculty interest in unionization: A reli- cation and extension. Journal of Applied Psy- chology, 66, 582-588. Bigoness, W. J. (1978). Correlates of faculty attitudes toward collective bargaining. Journal of Applied Psychology, 63, 228-233. DeCotiis, T. A. & LeLouarn, J. Y. (1981). A predic- tive study of voting behavior in a representation election using union instrumentality and work perceptions. Organizational Behavior and Human Performance, 27, 103-118. _—— Dworkin, J. B. & Lee, D. H. (1985). Faculty inten- tions to unionize: A new look. Unpublished manuscEipt, Purdue University, Krannert Graduate School of Management, West Lafayette, Indiana. Getman, J. G., Goldberg, S. B., & Herman, J. B. (1976). Union representation elections: Law and reality. New York: Russell Sage Foundation. Hammer, T. H. & Berman, M. (1981). The role of non- economic factors in faculty union voting. Journal of Applied Psychology, 66, 415-421. Kochan, T. A. (1979). How American workers view labor unions. Monthly Labor Review! 1011 23-31. Maxey, C. & Mohrman, S. A. (1980). Worker attitudes toward unions: A study integrating industrial rela- tions and organizational behavior perspectives. Preceedings of the 33rd Annual Convention of the Industrial Relations Research Associatian, 33L 326-333. Schriesheim, C. A. (1978). Job satisfaction, attitudes toward unions, and voting in a union rep- resentation election. Journal g£_Applied Psy- chology, 63L 548-552. 91 10 11 12 92 Youngblood, S. A., DeNisi, A. S., Molleston, J. L., Mobley, W. H. (1984). The impact of work environ- ment, instrumentality beliefs, perceived labor union image, and subjective expected norms on union voting intentions. Academy of Management Journal, 27, Zalesny, M. D. (1985). A comparison of economic and noneconomic factors in predicting faculty vote preference in a union representation election. Journal of Applied Psychology, 70, 243-255. Zonia, S. C. (1983). The dilemma of union- ization: A longitudinal case study-Ef facuIty at Michigan State University. Unpublished doctoral dissertation, Michigan State University, East Lansing, Michigan. & 93 .mumwm c« mom came mmuocmp % m an pmonHom mm

wcs man .m xwpcommd c“ concomoum mc0wumuwum .555 ANS mm ms.5 $5.5m msH a: S III .III mm ma.5 mm.mm mafia z m ouscme comm wusmmmz mom: ucoflowmmoou mumm mnflm ncofiummsooo umoEsz mo sewuoamuuou oncommmm mHmEmm whpoum Dumas wuo> coflco pom Hm>wq mmoz cowzumm mfinmcowumaom may now mcofiuoaouuou pmuomuuooca can mowumfiumuomumno mosum H magma mummsm mcaooo m xwpcommm 94 .nuso» a“ ouocou some on» monocoo a s an ouso-0u ouscoam .numo» ca was some monocoo a s an oosoaaou moan .mousooooum m>fiusuuowcmsco cums comuoouo«usm a 1m “woos“ unocuusu 1 mo .muoxuos scouoooOum a m «names: a z nauaooou aufinuosuco a man .c wmocommc cu vouconoum occuuouAUs III III .lll cm Hm.5 aa.om new no own III III mm. mm hv.5 wa.wn mmv m: Ha III. III III am mm.5 a~.~m an m a 555 Am. as. am mm.5 ua.am and me a sao.e sam as. m nm.5 ou.mm mm 2 m couscoa coma auwawnsfiaom. magnum: ucoaoauuooo ouom mafia anewuomoooo nonssz noquOOum auwmwmeuo cogunaouuoo oncomnom wamsom omcnum ouo> news: can cofiuouunacfisod saws cofiuosunmusm coosuom mwzncOLusHom on» now nowufiawnowaom nouofiouud can .mcomuoaouuoo oouoouuooca .moaunuuouoousnu aboum ~ manna Acoscuucoov ouoonm mcmoou 95 .nuoo» ca dunno» some any amuocoo a s an oosoaaou woodman .ousoa cw was some nouocoo a s an vasoHHOu ammo .auwusoon nofi .aa& a 0m: «no: on a emu .muoxuos ceavosooum u m uofimson coxfis u : .nonusc a z «anaconu auamuo>uco a man .¢ xuocuddt cg coucoooum unawusufios .III .III an. ac mm.5 pa.¢m mmv m: AH III. III .III om: mm.5 «~.~m mm m m .555 Awe so. am: H~.5 ua.em mad as S III .III as. o¢ mm.5 am.mm umm z m use.. sun 555. cc on.5 «u.um ma . z m monocoe comm >umafinofiamm masons: ucofiufiuuooo mun“ mafia accuuomsooo Hosea: acuufipmud comuoounmuom cofiuomouuoo uncommon sadism mapsum oomma wuo> coon: can cofiuoounuuom non ofimc«uuxm cuosuom dwnncowuoaom on» new mofiufiflfinowaom nououooum can .ncouuonouuou oouoouuooca .nogumfiuouoouonu apnea n «Home Aposcuucouv nuoonm mcfivou 96 .mumoa ow wuscwu come may monocwp a m an pmzoaaow monomep .muomm ow mom come mmuocmp a m an pmsoaaom mmmo .meEnm ooxfle u : .momuoc u z “muasomm huwwuo>fics 0 man .< xfiocmmmd c“ noncommum mCOMuoufioo up. wa.mm mom mo nNH mm. ow. wm.mm saw 2 m ssG.a smm ea. so. we.mm mm 2 m couscoe comm auflawnowaom ucmfiofiwmmoo wuom muwm anewummsooo HooEsz wouowpmum COLuMkuuou oncommmm mHmEmm mwpsum muo> coflcs can wufiaoucweouumcH news: cwmsuom mwnmcofiuoamm may you mmwufiawnowawm nouofipoum 0cm .mcofluoamuuoo pmuomuuoocs .mowumwumuomumnu upsum a magma prscflucouv mumwnm mewpou 97 .mumma cw muses» some may monocmo a n ma posoaaom wuscmep .muooh ow can come monocmo a m an omsoaaow mm¢o .onEnm pmxflfi.n 2 .mwmusc u z «wanoom auwmum>wcs 0 man .¢ xwocommm cw cmuconoum chMuouwom mm. wa.vm mmv mo HH me. mm.mm mam z m sho.v smm 555. he. we.om ma 2 m owuscma oomd unwawnmflamm ucmwofimumou mama ouwm ncofiummsooo umnEsz nouowooum newuoamuuou uncommmm onEmm chosum ouo> cOAC: can news: s wow ouo> ou unsung cwmsumm masocOMuoamm ms» now mowufiafiomwamm nouowomum can .mcowuoawuuou omuowuuooca .mowumfiumuoouono ansum m manna «nonsmucouv mummnm mcwcoo 98 .mumm» a“ mum come mmuocmp a m an pmsoHHON mmcm .hufisvm and 0 amp .qudmuuuHmm 0 man .mHmEmm omst u to .c xfipcmmmm aw omucmmmum unawumuwom 555 s~.mm 5555 mm mm ma. wv.mn Hand 2 A muscma mmmd mafiawnowamm musmmm: musmmmz acmfloamumoo mumm mem anewummomoo amass: cowumufiuu xuwsom mum: commoamuuoo mmcommmm deEmm mapsum no mo ommaa ummaa mcofiudmmumm auwovm can Hm>ma mmoz cmmsumm mmsmcowumHmm may now nmmuwawoowamm cowumufiuo can .mc0wuoamuuou omuomuuomca .nmwumfiumuomumnu aosum w wanna AomsCMucoov mummnm mewoou £99 .muoma cu man came mmuocmo a s 5n posedaou mocu .xufiusvmm ooh .xod a 0m: “00: on u as «and .mcoquEOHQ u unnm .qudmuuuamm 0 amp .mcowuofimuuoo maoumosm ucs>mamu uo mmoum>o cc muocmo xmwumumo co xn omzo-0u mcofiucamuuouu .mumeon pmxma u t “aufisonu xummum>acs 0 man .4 amocmdd< cw omucmmmud meowuouLUm 555 >~.mm as. 94 em as. av.ms men 2 A 555 saw no. em: mm as. we.am ass L: S III III as. o< mm as. um.mm H~5H z m III Ame 555 San mm .mm. as.~v NNN L: N muscms ummc xufldwoswamm musmmmz musmcmz ucmwmwuumoo mama mwfim ncowumasooo ambasz cowumuwuu comumcmmmuom moc3 mcowucamuuou mmc0dnmm mHQEom oxpoum 55s. is. :owumcuwfiuow non mwmcfiuuxm pcm ~m>mq mom: :mmsumm mmzmcofluofimm mzu uOu mmaumdwonmfimm COLLmuLuU can .mcowumdmuuou cmummuuomc: .umwumwumuooumcu xpsum A manna .pmscmucou. mummnm mcwpoo 100 .mummw aw muscmu come mnu mmuocmp x m xn pmonHom muscmem .mummh CA mmm come mmuocmp a m an omzoaaom mmmo .uuommuimamm u mmm .mamEmm pmxAE 0 2n .a xfipcmmm< CA pmucmmmum mcofiumuwom Amm.ma 555 mm. mm mm.5 we.~m mam 2 as 555 A~.mm GA. am He.5 AS.MA mAm z A III III mm. mm aa.5 mm.mm mew z m mmuscme pmm< auflafinmflamm musmmmz ucmwmflmmmou mumm muflm ncoAummsOmo umnEsz cowumufiuu mom: cofiumflmuuou mmcommmm mHmEmm mapsum 55mm. xuwamucmEDHDmcH coflca 0cm Hm>mq mmmz cmmsumm mfinmcoflumamm may now mmwuwafiomwamm coflumufiuu pom .mcoAumHmuuou pmuumuuomc: .mmwumflumuomumno upsum m magma Apmscwucouv mummnw mcfipoo 101 .muomx CA muscmu some mnu mmuocmp a Acmuscmu ucmuumm may mmuocmo m m an omsoaaom muocmem .mummm a“ mom some mmuocmp a m an omzoaaou mmmo .uuommunuamm u mmo .mamsom omxAE u z Axuaoomw aummum>wc= 0 man .1 xwpcmmmm CA omucmmmum chMumuAUm Amm.ma 555 am. am AH.5 wa.~o man 2 as 555 A~.mm 555 mm As.5 A4.MA GAA z A III .III .III mm NH.5 wm.mm mmHH : m .555 Ame mm. mm as.5 ma.~¢ ANA . A: a dA..o 555 555 mm Am.5 ao.mm AAA A: H mmuscme pmmm huflamnmwamm magnum: ucmwofiuwmou mumm muwm ncoAuomomoo umnEsz coAumuqu mum: coAuonuuou mmcommmm mHmEmm mucoum mwm>a mNACOAcD on ucmucu poo Hm>mq moo: cmmsumm mwsmcoAuonm mnu new mmwufiafinofiamm cowumuwuo pom .ncoAuonuuou cmuomuuomc: .mmfiumwumuomuono aboum a manna Apmscwucoov mummnm chooo 102 .muoma cu moo some mmuocmp » m an vmsoAHOu mmcm .00: on u can .aumsvm won a mom .mHmEom vmxwe a 2n .4 xapcmdmc cw omucmmmum moomuouAUm 555 A~.mm AA. ea we muscme mmmc aumawnmfiamm musmomz musmom: cowumufiuo cowumounwuom amazon no mo cmdxs omdaa mm. av.MA Ham 2 A ucmwmfiuumoo muom mnAm anewuomomoo amass: cofiumamuuoo monounmm mHmEom ohmoum cowuooumwucm non mamcuuuxm poo scamudmmumm auwavm :mmsumm mwnmcoAuonm mnu now mmwufiamnmAHmm cofiumufiuo can mcoqaoflmuuou pmuomuuomca .nmwunmumuooumzo upsum 5H manna Ammocmucouv mummcm mcuooo 103 .mummm CA mmm come mmuocmp a m an pmsoaaou moan .huwsvm and n mam .mHmEmm pmst 0 2n .4 xwpcmmm< CA omucmmmum mcofiumuflum ll A~.mm GA. ma 8.5 SSA saw 2 A muscme pmm< hufiafiomfiamm musmmmz ucmwmfimmmou mumm mnflm ocoflummsumo umnEDZ cofiumuwuo mufisvm cofiumHmuuoo mmcommmm mHmEmm whosum mo mmmaa hufiamucmEsuumcH cogs: can mcofiummmumm xuwsvm cmmsumm mwnmcoflumamm mnu mom mmwufiawnmflamm newumuwuo can mCOAumHmuuoo pmummuuomcs .mmwumwumummumno hpsum Ha manna AomscAucouv mummnm mcfioou 104 .omuscmu ucmmumm may mmuocmp m n ma pmsoHH0m muscmem .mumm» a“ mom come mmuocmp » m up pmsoHHOm mmmp .mHmEmm omxHE u z Aqusmmw huwmum>fics .huwsvm mmfimu u m Axuwovm and mam man .d xwpcmmmd CH pmucmmmum mcoHumuHUm I £43 I. me Hi £12 :5 z A 3.3 II mm. m Ami $.95. AAA .3 H mmuscme pmmd muHHAanHmm musmmmz ucmwowmmmou mumm mNHm ncoAummsmoo umnEDZ Houofipmud quswm cofiumHmuuoo mmcommmm mHmEmm mwpsum mo mmmxe coHca o How muo> ou ucmucH pom mcofiummoumd wuwsvm cmmzumm mfismcoHumHmm mnu How mmAuHHHboAHmm Houmfipmud can .mcofiumHmuuou nmummuuomcs .mmfiumfiumummumzo upsum NH manna HpmscHucouv mummcm mcfipou 105 .muoma cm muscmu come ms» mmuocmo a a an omsoHHou muscmam .nuom> cm mom came mmuocmp x a up omsoHH0u mode .00: on a ocu .mHmEoo pmxAE u : Hummus: a 2n .< xHocmdd< cu omucmmmud mcowumuHUn In 3.2” SA. A. 3 E5 3.: 8A 2 A III III om. mA. 04 Hm.5 am.mm mmm z m AAA... AAA Am. 555 2 3.5 3.3 mm 2 H mmuscms pmme quHanHHmm quHHosHHmm musmomz ucmmowuumou muom mnHw ncowumdsooo amass: cowumuwuu acuowcmua cofiumoumwuom cowuonuuou macadmmm mHmEsm mapsum 05%. quHmucmE=uumcH cows: pen coHumcumHuow non mwmcwuuxm cmmsumm manncofiuonm mnu Low mmHuLHHnoHHmm comumumuu can noumfipmud pcs .mcomuonuuoo vmummuuouc: .nmwunwumuonumno >psum ma mHQMB Apmscwucouv mummnm mcwoou .nuom> cH muacmu some mnu nmuocmp a a >n omsoHHOu muacmeu .nunma cw mom came mmuocmo a s an pmsoHHOu mmtm .Aumuammm new .hsm u out «mo: on a at Axon .mcoAuOEOHQ a Hahn .mcoHuonuuou mHmmnnsn ucs>mHmu no monum>s cs muocmo xnwumums so an pmsoHHOu ncoHuonuuoom .mumxuos cofiumzooud n m «deEom cmxHE a z “omens: u z uqu90su auwmum>mcs a was .< umpcmmmt cu vmucmmmum mcomumuMUo 106 555 555 555 .555 me an.5 au.-n mm. A: HH 555 555 555 555 om: Hm.5 «u.AA ASH z m .555 A~.mn 555 AA. as an.5 a-.nA as» x A 555 555 .555 AA. o4 54.5 sm.am «as 2 m AAo.q ANA 555 555. cc ev.5 au.um mm 2 m 555 Am. mm. .555 How .mm.5 «o.H. «AA A: A .922. .9. shaman“. mmmwm. sham”... .uuwwwww 9mm”... hm. .5298: “an...” ommma cowc: m new muo> Cu ucmucH can cofiummwmmuom non uHmcHuuxm cmmsumm aficmcomuonm mzu new mmwumHHomemm cowumuwuu can Leumwpmud pco .mcomuonuuou omummuuomc: .mumummumummumnu xpsum vu manmh .ccscHucoos mucosa mcooou 107 .mucmx an muacmu come mnn nmnocmo a s an omsoHHou muocmao .nuomh ca mom came nmnocmo a s an omaoHHOu mono .mHmEom omst a z «nmnua: a z «auHamou annuum>mca a man .c unocmddt ca omucmmmud ncoHusuHUs III. III III III «A. aa.¢v cam ma UNH 555 555 555. 555 NA. as.mm AHA A: . «AH Amm.mH .555 mm. AA. «A. as.~o men 2 AH I 5.3. 555 A. mm. 3.2 :5 z A III III .III we. we. am.am Ham 2 m AAS.. Ann .555 Am. GA. «n.0m ma 2 n amuscma mmmc quHHnoHHmm auHHHnoHHmm acmnmmuumou muom munm ncownomsooo umnaaz conumuwuu nouoaomum conuonuuou mucoammm mHanm sacsum coHca a you mno> on ncmucu was hunHoucm55uuncn coHca cmmsumm annnconnonmm mnn you nmauwnnnomHmm conumnnuu poo noummomud oco .nconnonuuou omnomuuomca .nmuumnumuuouonu hooum ma manna .omseaucoo. mucosa meaoou 108 .musma ca muacmu some mnu nmuocmo a nomuacmu unmoumd mnu nmuocmo a s an vmsoHHOu muocmem .nuoma an mus cams nmuocmo a s an omsoHHou moan .mmuoomooum moansunoacmeos nuHs conuosunuuso a {a «among unmounsu a mu .nmuuoc a z «auHsosu auanum>Hco a man .¢ uHocmdmc cu omucmomud ocoHusuHos III III III 4m am.l aa.vv saw no UNH III III III cm Hv.5 as.mm aHm ma aNH III III am. an Av.5 aa.vn mmw ma HH da.ao AA..¢ 555 am A¢.5 .«o.nn AHA as q AAo.. AAA SA. A on.5 .555 mm 2 A mmuocma ommc AnHHHanHmm musmsmz ucmHoHuumoo muom mnHw nconusmsooo nonsuz nouoaomua coHnoouansm conunHmuuou mocomnmm deEsm swoonm ommme conca o How muo> on ucmncu one coHusuuchHEpn nuns connmsumwusm cmmsumm aunchHHonm may now mmHuHHHnoHHmm nonmaomud ass .nconusHmuuoo amuumuuomca .nmuunmumumousno amouw SH anos HomscHucooa uummnm mcHuoo 109 .mummx ca moo some mmuocmo a s an omsoHHou mmcm .mmuoomOOHQ m>Husuuchneoo nan coHuooumHumm a two .qudmuuuHmm u mmo .anHauou auanum>wca a man .< uncommmc cu pmucmmmum acoaunnnos I A3 3 . «m em AH . S .3. 3H .5 o umuacmk mmoc >nHHnnmaHmm musmomz muammmz ucmnowuumou mucm mumm nconusaamoo umnEaz couumnauo conumoumnnom mom: cowuonuuou mchQnmm mHnEow o>paum no no pmmae mmaxa coauouumwcmsat nun: comummumnnow was Hm>ma mac: cmmsumm annmcomnnHmm mnu new mmaunnanmnnmm comumuwuu pcn .mcoauonmuuou vmummuuomca .mmaumnumummumno apauw ha manna Homacnucoua mummnm manpou 1Ll0 .muom> ca muacmu come mnu mmuocmo a o >n pmsoHHOu muocmeu .mumm» on man coma mmuocmo a s an pmsoHHOu mmcm .auwusmmn nan .xom u Om: noon as a 0:0 .mmuaamUOHQ m>auounmncmeco nun: conumoumauon a (w uxmpca nnmcuuou a mo .nmmua: u z «huHaoou aunmum>mca a ban .1 unpcmdac cm omncmmmua mconusuAUo 555 555 AA. so. as am A.. sa.vn mmv A: HH 555 AH. so. so. am: am we. Ha.am ASH A: s AAo.v AAA 555 SA. me A an. pa.om ma 2 H umuacme mmm< Aunnnnoaamm AuHHHnoHHmm muamomz muamomz ncmamauumou mnom mNHm ncomuodsouo umneaz comumumnu unummpmua onncauuxm coanmoummuom cowuonuuou mucoanmm mHQEnm sxpauw 5s. 5s. coauoouwnumm non Oamcnunxm aco connouumncnfip< nun: conumsumnumm cmmsumm annmcowumamm mnu uOu mmwunnnnonnmm comumumuu can acuumpmud pco .mcoauonunou pmuomuuomca .nmwumwumn00uonu xpaum AH mHnae 1cm=caucooa masons season 111 .muom> an muscmu come on» mmuocmc a s an pmsoHHOu muacmem .mucm> on mom came mmuocmo a m an pmonH0u mm4p .mmusommoud m>aucunmacnsoo nun: conumoumnuon a 4m uxmocn mmmcHHou a mo .nmmuoc u z Aquounu Aunnum>mcs I man .4 xaocmdd4 ca pmucmmmum mcoauounom III III III III 4m am.5 ”5.44 amm ma Own III III III III 4w mm.5 aa.wm mww ma n~H III .III III III 4m an.l va.mm mam ma own A3... A2 AA. SA. A 3.5 seam mm 2 H mmuscme omm4 >uHHanHHmm huHHHanHmm musmcmz ucmaoauumoo mums mNHw ncoHundzouo amassz coaumunuu acnmnamud connoouwnuom :oHuonuuou musedmmm deEom mxmaum 0...“. xuaaoucmsauumcH coHca pco connmunmncaec4 can: conumoumaunm cmmsumm mwnmcownmnmm mnu new mmmuaaanoanmm conumuwuu can Lenomamud can .mconumHmuuou pmnmmuuomca .mmmumnumnmmuono xpaum ma mans? Apmacancoua mummnm ocwpou 1J12 ..mm.AH om. Hm.~m omen. Nose. HNHH. . HAHN A.o.m.~ mm5az ..A..o. AA. a~.o Hose. smug. HsoA. m AHAH HH.m.m >a5>H ..Hm.ev ms.m H..A mmHa. mmHa. Anus. H mmm an.m.n >s5H: we.» .m.a ~a.~o ease. ones. we...5 m meoH HH.o.w.m.m >=5mm .Nm.HH em.~ AA.A¢ mmea. Amen. «sao.5 m Has EAH.HH.A.S >=5355: muoaqm anu coaumscmuu4 acuum mconuonuuou mcomumHmuuou conumHmuuou ocomuonuuou munm mumnEaz annmconuonm 0» mac mmcowus> mo mmmucmmumd oanmEow ca muconuo> cm mocnauo> com: no umnEaz mHmEsm apsuw 0» mac cmuomuuou pmummuuoo mmcmHum> no mmoucmOumd xnuuoz cowuonuuou mnu m>auma 0n amma mmma~oc45numz mnn uo muHammm U napcmma4 113 ONH.MNH aaa~.mvH ww.H ma.v hams. smma. mwoo. o ammn .a~.h.m.n >~IHD aw.m mH.aH mm.mm a anus. smvv.5 a amom Hn.m.n.m.m.~ >~5mm camH.Nm an. mN.n Nvma. mama. vam.5 m Hana h.m.n Halwm vm.~ a mm.mm vaaa. swan. mva.l N vNuH h.H >ulmm caso.vm a mv.vH HHHa. mHHa. NmmH.5 m vamw 5H.A.m.~.H >H543 acam.- ma. wa.o~ Nmaa. Noon. omwu.l n mwma aa.h.m main: mumavm “no conumacmuu4 neuum mcoauonmuuou mcoaumnmuuou coaumHmuuoo ncomuonmuuou munm mumnEDZ Qanmcomuonm on mac mcHHQEnm ca mmcmauo> cw mmcmmuo> com: no umnsaz deEmw >panw mocoHum> Lo by man pmnumuuoo pmnmmuuoo mmoucmmumd mocoauo> Lo mmmucmOHmd Homacwncoua U xapcmmd4 114 Axuaamucm62uumcH conca 0 Ha “cowumuummcmEp4 nuns coanuoumaumm u .muo> comca u >a as. v Qaa ma. v a. “muo> 0» ucmucH u >H 4m Anonymoumnumm mnmcnunxm u we “anaaom and n ma nnm>mn mom: a n3 .4 xapcmdd4 cm omucmmmud mcomusnnu .muoz 1.3.: As... MASH «AHA. oan. 3.3.5 c 9.5 625262.... S55,“ mm. a 5.2: a 33. 33. m om... HHS; $55 UNH.n~H Am.m am.a NA.mm swan. Nmas. Aaao.5 m mooH .HH.v.m >H54m mumsqm mnu cowusacmun4 Lanna mcoanonuuoo mconumHmLuou conusHmuuou mcoHnsHmuuou munm mumnEsz mwnmconumHmm on mac mcdeEmm ca mononuc> cw mucomun> com: uo umnEaz mHmEom >pauw mocnnum> no 0» mac pmuumuuou cmummuuou mmoucmOumd mmconum> no mmcncmOumd Homacmncoua U xmpcmmd4 REFERENCES References Adams. J. S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.), Advances 13 experimental social psychology, Vol. 24_ New York: Academic Press. Allen, R.E. & Keaveny, T. J. (1981). Correlates of uni- versity faculty interest in unionization: A replica- tion and extension. Journal pf Applied Psychol- 09yz 66, 582-588. Bigoness, W. J. (1978). Correlates of faculty attitudes toward collective bargaining. Journal pf Applied Psychology, 63, 228-233. Block, R. N. & Premack, S. L. (1983). The unionization process: A review of the literature. In D. B. Lipsky 8 J. M. Douglas (Eds.), Advances ip_industrial and labor relations, Vol. l;_ Greenwich, Conn: JAI Press. DeCotiis, T. A. & LeLouarn, J. Y. (1981). A predictive study of voting behavior in a representation election using union instrumentality and work perceptions. Organizational Behavior and Human Performance, 21, 103-118. Dickens, W. T. (1983). The effect of company campaigns on certification elections: Law and reality once again. Ipdustrial and Labor Relations ngigy, 36L 560-575. Duncan, 0- D- (1975). Introduction to structural 23233123 models. New York: AcadEmiE Press. Dworkin, J. B. & Lee, D. H. (1985). Faculty intentions £3 unionize: A new look. Unpublis e manu- script, Purdue Universify, Krannert Graduate School of Management, West Lafayette, Indiana. Dyer, L. & Parker, D. F. (1975). Classifying outcomes in work motivation research: An examination of the intrinsic-extrinsic dichotomy. Journal pprpplied Psychology, 60, 455-458. Farber, H. S. & Saks, D. H. (1980). Why workers want unions: The role of relative wages and job charac- teristics. Journal pf Political Economy, 88, 349-369. 115 116 Feuille, P. & Blandin, J. (1974). Faculty job satisfac- tion and bargaining sentiments: A case study. Academy pf Management Journal, 17, 678-692. Fiorito, J. & Greer, C. R. (1982). Determinants of U.S. unionism: Past research and future needs. Industrial Relations, 21, 1-32. Fishbein, M. (1967). Attitude and the prediction of behavior. In M. Fishbein (Ed.), Readings i2 attitude theory and measurment. New York: Wiley. Fishbein, M. & Ajzen, I. (1975). Belief, attitude, intention and behavior: Ag introduction 53 theory 33g research. Reading, MA: Addison- Wesley. Fisher, C. D. & Gitelson, R. (1983). A meta-analysis of the correlates of role conflict and ambiguity. Journal pf Applied Psychology, 68, 320-333. Freeman, R. B. & Medoff, J. L. (1984). What unions do? New York: Basic Books. Galbraith, J. & Cummings, L. L. (1967). An empirical investigation of the motivational determinants of task performance: Interactive effects between instrumentality-valence and motivation-ability. Organizational Behavior and Human Performance, 21, 237-257. Getman, J. G., Goldberg, S. B., & Herman, J. B. (1976). Union representation elections: Law and reality. New York: Russell Sage Foundation. Goodman, P. S. (1974). An examination of referents used in the evaluation of pay. Or anizational Behavior and Human Performance, 12: 70-195. Goodman, P. S. (1977). Social comparison processes in organizations. In B. M. Staw & G. R. Salancik (Eds.), New directions 12 organizational behavior (pp. 97- 131). Chicago: St. Clair. Hammer, T. H. & Berman, M. (1981). The role of non- economic factors in faculty union voting. Journal g£_Applied Psychology, 66I 415-421. 117 Heneman, H. G. & Sandver, M. H. (1983). Predicting the outcome of union certification elections: A review of the literature. Industrial and Labor Relations Review, 36, 537-559. Herzberg, F., Mausner, B., & Snyderman, B. S. (1959). The motivation £2 work. New York: Wiley. Hunter, J. E. & Gerbing, D. W. (1982). Unidimensional measurement, second order factor analysis and causal models. In B. M. Staw & L. L. Cummings (Eds.), Research 13 organizational behavior, Vol. 4, Greenw1ch, Conn.: JAI Press. Hunter, J. E. & Schmidt, F. L. (1978). Differential and single group validity of employment tests by race: A critical analysis of three recent studies. Journal pf Applied Psychology, 63, 1-11. Hunter, J. E., Schmidt, F. L., & Jackson, G. B. (1982). Meta-analysis: Cumulating research findings across studies. Beverly Hills, CA: Sage. Kochan, T. A. (1980). Collective bargaining and industrial relations. Homewood, IL: Irwin. Kochan, T. A. (1979). How American workers view labor unions. Monthly Labor Review, 101z 23-31. Lent, R. H., Aurbach, H. A., & Levin, L. S. (1971). Predictors, criteria, and significant results. Personnel Psychology, 24, 519-533. Locke, E. A. (1976). The nature and causes of job satisfaction. In M. D. Dunnette (Ed.), Handbook pf industrial and organizational psychology. Chicago: Rand McNally. Locke, E. A. (1969). What is job satisfaction? Organizational Behavior and Human Performance, ‘4, 309-336. March, J. G. & Simon, H. A. (1958). Organizations. New York: Wiley. 118 Maxey, C. & Mohrman, S. A. (1980). Worker attitudes toward unions: A study integrating industrial rela- tions and organizational behavior perspectives. Proceedings pf the 33rd Annual Convention of the Industrial Relations Research Associatign, 33, 326-333. Mills, D. Q. (1982). Labor-management relations, (2nd ed.). New York: McGraw-Hill. Mobley, W. H. (1982). Employee turnover: Causes, consequences, and control. Menlo Park, CA: Addison-Wesley. Mowday, R. T. (1979). Equity theory predictions of behavior in organization. In R. M. Steers & L. W. Porter (Eds.), Motivation and work behaviog, 329.29; New York: McGraw-Hill. Prichard, R. D. (1969). Equity theory: A review and critique. Organizational Behavior and Human Performance, 4, 176-211. Roomkin, M. 5 Block, R. N. (1981). Case processing time and the outcome of representation elections: Some empirical evidence. University pf Illinois Law Review, I, 75-97. Rosenthal, R. (1979). The "file drawer problem" and tolerance for null results. Psychological Schmidt, F. L., Hunter, J. E. & Pearlman, K. (1981). Task differences as moderators of aptitude test validity in selection: A red herring. Journal pf Applied Psychology, 66, 166-185. Schriesheim, C. A. (1978). Job satisfaction, attitudes toward unions, and voting in a union representation election. Journal pf Applied Psychology, 63, 548-552. Strauss, G. (1984). Industrial relations: Time of change. Industrial Relations, 23, 1-15. Triandis, H. C. (1971). Attitude and attitude change. New York: Wiley. 119 Vroom, V. H. (1964). Work and motivation. New York: Wiley. Weick, K. E. (1966). The concept of equity in the perception of pay. Administrative Science Quarterly, 11, 414-439. Youngblood, S. A., DeNisi, A. S., Molleston, J. L., & Mobley, W. H. (1984). The impact of work environ- ment, instrumentality beliefs, perceived labor union image, and subjective expected norms on union voting intentions. Academy of Management Journal, 27, 576-590. ' Zalesny, M. D. (1985). A comparison of economic and non- economic factors in predicting faculty vote preference in a union representation election. Journal of Applied Psychology, 70, 243-255. Zonia, S. C. (1983). The dilemma pf unionization: A longitudinal case study of facult at Michigan State University. UnpubTiShed acctEEal disserta- tion, Michigan State University, East Lansing, Michigan.