Journal of Occupational Psychology (1991), 64, 129-143 (g) 1991 The British Psychological Society Printed in Great Britain 129 Predictors and outcomes of different patterns of organizational and union loyalty* Clive Fullagarf Department of Psychology, Bluemont Hall, Kansas State University, Manhattan, KS 66506, USA Julian Barling Queen's University, Ontario This research identified four different patterns of loyalty to the organization and union in two Canadian unions (N = 426): (1) loyalty to both company and union (dual allegiance/loyalty); (2) loyalty to neither the union nor the company (dual nonallegiance); (3) loyalty to the union, but not the company (unilateral union loyalty); and (4) loyalty to the company, but not the union (unilateral company allegiance). The predictors and consequences of these different patterns of company and union loyalty were assessed in a longitudinal design across three six-month periods using a discriminant analysis and multivariate analysis of covariance, respectively. The results of the discriminant analysis confirmed the existence of unique predictors of organizational and union commitment. Only one common predictor of union and company commitment was found, namely union involvement. Similarly, the multivariate analyses of covariance suggested unique consequences for union and organizational allegiance. The implications ofthe findings for both unions and organizations are discussed. In particular, the possibility that involvement in either the union or the organization does not necessarily exert negative consequences for the other is emphasized. The purpose behind much ofthe earlier research on dual commitment^ to the organization and the union (e.g. Dean, 1954; England, I960; Purcell, 1954, I960; Stagner, 1954) was to ascertain whether membership in, and commitment to, labour organizations precluded commitment to the employing organization. This initial research * Portions of this paper were presented at the Annual Meeting of the Industrial Relations Research Association, 28-30 December 1988, and the Canadian Psychological Association Meeting, 9-10 June 1989. t Requests for reprints. X For the purposes of clarity, two terminological distinctions relevant to the present study should be emphasized. First, we will refer to (union) loyalty rather than (union) commitment. Union commitnient is a multidimensional construct (e.g. FuUagar, 1986; Fullagar & Barling, 1987; Gordon et al., 1980), of which union loyalty is only one component. Like previous research (e.g. FuUagar & Barling, 1989), we focus only on the loyalty component. So as not to obscure the meaning of our constructs, we refer to union loyalty (and dual loyalty) speciBcally. Second, although much ofthe previous literature refers to 'organizational' commitment (e.g. Mowday et al., 1982) when teferring to 'profit-oriented' companies, unions are also organizations. Our use ofthe term 'organization' throughout the present study is specific rather than generic, and refers to the type of organization studied in this research, namely universities as organizations. Consequently, we favour use of the more specific terms, union loyalty and organizational loyalty, as they refer to one of the two different entities. 130 Clive Fullagar and Julian Barling was guided by cognitive dissonance and role theory: implicitly accepting the inevitability of an adversarial relationship between companies and unions, researchers questioned whether individuals could be loyal to both organizations simultaneously. The results obtained were equivocal, since positive (e.g. Dean, 1954; Purcell, I960) and negative (e.g. England, I960) relationships between loyalty to the organization and loyalty to the union emerged. Recently, there has been a resurgence of interest in dual loyalty which stems in part from decreasing unionization levels in the United States and the development of new forms of collective bargaining which require more cooperative labour—management relationships (Ephlin, 1988; Kochan, Katz & McKersie, 1986). In addressing this issue, research has moved beyond the question of whether dual loyalty exists. Instead, to explain the discrepant relationships between union and organization commitment, researchers have resorted to 'moderator' hypotheses (Angle & Perry, 1986; Magenau & Martin, 1985; Martin, Magenau & Peterson, 1982). Within this framework, it is suggested that situational variables moderate the direction and size of the correlation between company and union loyalty. However, major problems exist with some ofthe research focusing on moderators ofthe dual loyalty phenomenon which seriously limit its utility, and there are sound practical reasons for going beyond an examination ofthe relationship between union and company commitment. First, whereas earlier research was based on cognitive dissonance theory, which suggested that dual loyalty to the organization and the union is not possible, a major trend within current labour relations practice is the cooperative relationship between the union and the company (Ephlin, 1988; Kochan et al., 1986). Even though there is some question as to whether this pattern represents a lasting trend or is merely a transitory phenomenon (Dunlop, 1987), there is little doubt that it is a common current feature of labour relations. Second, considerable variation in the conceptualization and measurement of both company and union commitment prevails, preventing easy comparison between studies (Angle & Perry, 1984). Third, and more important from the perspective of the present study, previous research on dual loyalty has ignored the full range of different patterns of loyalty. Specifically, a positive correlation between union and organizational loyalty can suggest either of two patterns of loyalty: (1) loyalty to both company and union (dual allegiance/loyalty); or (2) loyalty to neither the union nor the company (dual non-allegiance). Likewise, a negative correlation between union and organizational loyalty suggests two further patterns of unilateral (or divided) loyalty: (3) loyalty to the union but not the company (unilateral union allegiance); and (4) loyalty to the company but not the union (unilateral company allegiance). Consequently, a comprehensive assessment of simultaneous loyalty to both organizations must assess these four different patterns of loyalty. Given the existence of these four distinct patterns of loyalty, two more relevant questions are how individuals become associated with one pattern (i.e. a question of causal inference), and whether differential consequences accrue for individuals' behaviour in organizations as a function of their membership of one of these four groups. The selection of appropriate constructs to be tested as possible predictors and outcomes is critical for assessing the predictors and consequences of different patterns of company and union loyalty. We based our selection of predictors and outcomes on previous research and theorizing about company loyalty (e.g. DeCotiis & Summers, 1987; Mowday, Porter & Dual loyalty to organization and union 131 Sreers, 1982) and union loyalry (Fullagar & Barling, 1987, 1989; Gordon, Philpot, Burt, Thompson & Spiller, 1980). Ir must be noted that research which has investigated the relationships between job and union-related variables, and union and organizational commitment, has been correlational in nature and based almost entirely on cross-sectional research designs, which preclude a true assessment of cause-effect relationships. Consequently, the purpose of the present study was to adopt a longitudinal design in order to develop, in an exploratory fashion, a process model which identified antecedents and outcomes of different patterns of commitment from a set of variables which had been associated with both types of commitment in previous research. Because of the exploratory nature of the research, no attempt was made to distinguish theoretically between which variables should constitute the antecedents, and which the consequences, of the patterns of allegiance on ana priori basis. Indeed, the same variables were tested as predictors and outcomes in order statistically to explore the causal nature of the relationships between variables. It was our intention to use the data to build a theory of dual and unilateral allegiance, rather than to use previous correlational and cross-sectional data to justify a theoretically driven model. First, work experiences and attitudes were considered. Both intrinsic and extrinsic job satisfaction were studied, as research suggests consistently that the different facets of job satisfaction are associated with company loyalty (DeCotiis & Summers, 1987) and union loyalty (Decotiis & LeLouarn, 1981; Hamner & Smith, 1978; Schriesheim, 1978; Zalesny, 1985). Likewise, job involvement has been associated with company loyalty (Buchanan, 1974; Lodahl & Kejner, 1965; Mowday et al., 1982) and union loyalty (Fullagar & Barling, 1989). Research has also shown that work stress and role stress predict both company commitment (Fukami & Larson, 1984; Mowday et al., 1982) and pro-union attitudes and activities (DeCotiis & LeLouarn, 1981). Finally, supervisory support is associated with company loyalty (Mowday et al., 1982) and union voting intentions (DeCotiis & LeLouarn, 1981). Several union attitudes and experiences are associated with unionization in general, and union loyalty in particular. First, perceptions of the quality of union—management relations moderate the dual allegiance phenomenon (Angle & Perry, 1984). Second, general attitudes towards unions are central to the unionization process (Brett, 1980^; DeCotiis & LeLouarn, 1981) and they predict all four dimensions of union commitment (Gordon etal., 1980). Third, recent research has highlighted the central role of perceived union instrumentality in understanding union voting behaviour (Beutell & Biggs, 1984; Brett, 1980^; Premack & Hunter, 1988) and union loyalty (Fullagar & Barling, 1987, 1989). Fourth, because the two core behavioural elements of industrial relations are conflict and change, which themselves are central to the experience of negative stress (Bluen & Barling, 1988), and because work stress correlates significantly with organizational commitment (Fukami & Larson, 1984) and pro-union attitudes (DeCotiis & LeLouarn, 1981), the role of industrial relations stress in predicting different patterns of union and company commitment was investigated. Finally, because research has shown that behavioural participation in union activities is associated with union loyalty (Fullagar & Barling, 1989), the present study also investigated whether the diverse patterns of company and union loyalty exerted a differential effect on union participation. To ascertain the predictors and outcomes of the four patterns of loyalty, the present study used a three-wave panel design. Such a longitudinal design addresses questions of 132 Clive FuUagar and Julian Barling causality and controls statistically for possible confounding variables to strengthen causal inferences. Consequently, the present study extends previous research by {a) exploring more comprehensively the different relationships that can emerge between union and organizational loyalty, and Q}) identifying the causes and consequences of four patterns of union and organizational loyalty. Method Subjects and setting All participants in the study were employed on a full-time basis by one of two large universities in Ontario. To enhance the generalizability ofthe results, two different unions were studied, one from each university: an academic staff union, and a support services union. Demographic profiles of subjects who responded at Time 1 suggest that there were marked differences in the demographic composition ofthe two unions. Respondents in the academic staff union were significantly older than their counterparts in the support services union (AI = 47.56 vs. 38.08 years), had received more formal education {M = 20.77 vs. 15.01 years), and had been employed by the university (M = 14.03 vs. 9.43 years) and a member ofthe union for longer periods of time (Al = 9.23 vs. 6.94 years). In addition, there were more males than females in the academic staff union (72 vs. 28 per cent) and significantly more females than males in the support services union (67 vs. 33 per cent; ;f^(l) = 55.11,p < .01). However, the proportion of union officials to rank-and-file members amongst the respondents did not differ across the two unions (17 vs. 83 per cent) in the academic staff union, and 12 vs. 88 per cent in the support services union; ;f^(l) = 1.49), and thete were no other significant differences between the two unions on the variables described below. Procedure Questionnaires and a letter from the union President indicating the support of their Executive Committee for the research were sent to every member of both unions (N = 1623). The questionnaires were numbered in order to match responses from later surveys. However, 120 questionnaires were returned as undeliverable for a variety of reasons, reducing the initial targeted population to approximately 1503 respondents. Subjects were asked to return the questionnaires in the self-addressed, reply-paid envelopes that were provided. To increase the likelihood of responding, reminder cards were sent to all subjects 10 days after they received the questionnaires. Ofthe 1503 questionnaires initially distributed, 464 were returned at Time 1, a response rate of 30.9 per cent. A second administration ofthe questionnaire was conducted six months later. It was felt that this time period was sufficient to allow for causal effects to emerge. Ofthe 1503 questionnaires distributed, 426 were returned, yielding a response rate of 28.3 per cent at Time 2. Because ofthe longitudinal nature of the present study, it would have made no sense to study respondents who completed data at one point in time only. Consequently, at the third and final data collection phase, questionnaires were sent only to those respondents who had replied at either Time 1 or Time 2 (N = 603). To maximize the tesponse rate, reminder postcards were again sent to all respondents two weeks after the questionnaires had been distributed. In total, 323 questionnaires were returned, providing a Time 3 response rate of 53.6 per cent. Given these response rates, which are typical for research focusing on union members (Fryxell & Gordon, 1989; FuUagar, 1986; FuUagar & Barling, 1989), the extent to which the respondents differ from non-respondents is of some concern. There were insufficient data available from the two universities to be able to compare respondents with non-respondents. However, individuals who responded only once did not differ significantly from their counterparts who responded on two or more occasions in terms of any of the demographic variables measured. Questionnaires The data gathered for the study were classified into three sections: (1) demographic data, (2) organizational/ work-related experiences and (3) union-related experiences. [Copies of the questionnaires are available from the fitst author.] Dual loyalty to organization and union 133 Demographic data. Research on both company and union loyalty suggests that a number of demographic variables predict loyalty (Fullagar & Barling, 1987; Mowday etal., 1982); subjects were therefore asked to provide relevant information (e.g. gender, company and union tenure, age). Companylorganizational experiences. Different questionnaires were used to assess organizational loyalty, job satisfaction, job involvement/alienation, role conflict and ambiguity, and peer and supervisory support. Each of the scales used is presented below with a brief rationale for its inclusion in this research. The means, standard deviations and reliability coefficients of the scales at each time period are presented in Table 1. With the exception of the supervisor support scale, where the internal reliability was modest, the internal reliability and test-retest reliability over six months (Time 1 to Time 2) of the other scales were satisfactory. Commitment to the organization was measured by Porter & Smith's (1970) nine-item Organizational Commitment Scale. This particular scale was chosen because of its role in the development of an operational definition of union loyalty, which consists of essentially the same factors (Fullagar & Barling, 1989; Gordon et al., 1980). Warr, Cook & Wall's (1979) l6-item Overall Job Satisfaction Scale was used because it assesses both intrinsic and extrinsic job satisfaction. Kanungo's (1982) 10-item scale was administered to assess Job Involvement. Whereas other scales (e.g. Jans, 1982) operationalize job involvement as comprising three dimensions (viz. general job involvement, specialization involvement and a performance/self-esteem contingency), job involvement is measured by Kanungo along a single bipolar continuum (involvement-alienation). Role conflict (six items) and ambiguity (eight items) were assessed in the present research using Rizzo, House & Lirtzman's (1970) questionnaires. Finally, the nine-item supervisory support dimension from Moos' (1981) Work Environment Scale was administered. Union experiences. There is a plethora of psychometrically acceptable questionnaires assessing work experiences (Cook etal., 1981). Consistent with the fact that industrial and organizational psychologists have focused so little attention on union and union experiences over the past 3 0 ^ 0 years, there are very few psychometrically acceptable questionnaires on union-related experiences. Union commitment was assessed with Gordon et al. 's (1980) 28-item questionnaire. The construct validity of this scale has been replicated on white-collar (Gordon etal., 1984; Ladd, Gordon, Beauvais & Morgan, 1982) and blue-collar (Fullagar, 1986) workers. Nevertheless, an iterated principal factor analysis with varimax rotation was computed on the scores of union commitment measured at Time 2, i.e. that time period at which union commitment served both as an outcome of prior experiences and predictor of subsequent attitudes and behaviours. Five factors were extracted using the criterion that eigenvalues should be greater than unity. However, the first factor accounted for 46 per cent of the total variance and 80 per cent of the common variance. Thirteen of the 28 items loaded on this factor (loading >.35) and it closely resembled the first factor in other studies on the construct validity of union commitment (Fullagar, 1986; Gordon et al., 1980; Klandermans, 1989; Ladd et al., 1982). Consequently, this factor was labelled union loyalty and used in the present study as an indicator of union commitment. [Full details of the results of the factor analysis conducted at the second time period are available from the first author on request.] To assess the general quality of union-management relations, Biasatti & Martin's (1979) 11-item questionnaire was administered. Scores derived from this unidimensional scale correlate predictably with specific measures of job satisfaction and perceived organizational climate (Barling & McLennan, 1985). To assess the negative perception of industrial relations events, Bluen & Barling's (1987) 19-item short form of the Industrial Relations Event Scale (IRES) was administered. There is support for the test-retest reliability and validity of the negative industrial relations stress index (Bluen & Barling, 1987). In addition, data from a longitudinal study showed that negative perceptions of industrial relations events predicted psychological distress as long as six months after a strike had ended (Barling & MiUigan, 1987). Brett's (1980i') eight-item questionnaire was administered to assess attitudes towards unions in general. Union instrumentality was assessed using Chacko's (1985) seven-item perceived union instrumentality scale. Chacko's (1985) scale was chosen above other available scales because it provides a unidimensional measure of perceived union instrumentality. Finally, a seven-item measure similar to that used by Fullagar & Barling (1989) was used to assess the extent of individual involvement in union activities (i.e. union participation) at the first testing phase. For the second and third time phases, however, three additional items (namely, 'been a delegate at a national union convention', 'been a candidate in a union election' and 'voted on a union issue') were included to increase the range of union participation behaviours assessed. 134 Clipe Fullagar and Julian Barling OS O VD 00 r ^ 00 ^^^ O O SO CJN r ^ O m o Q 00 SO ON r-* ON -H ON OS 0 u ^ 0 ^ ^ ^ 0 , u \ < f O O O O Ost^00O00000iy-\O\00Os0000 Q T—r o 00 rA O 00 (N00v/-\OO»/Nr~^S0v/-N '^cOrTi'^rriT—< rnro, I Q SO t/3 cCi T—* o ^ 1—I '—I !;7\ M^i O '—' '—I •" .2 III a, " c 3 C g § S J3 "^ 1 •€ I I O w S 3 ^5 135 Dual loyalty to organization and union Data analysis A longitudinal design was used in which data were gathered at three points in time, each separated by a six-month time period. Although longitudinal designs do not in themselves provide proof of causality, causal inferences are strengthened through their use, especially if confounding variables are controlled experimentally or statistically. Subjects were divided into the four loyalty patterns outlined in the introduction on the basis of a median split of scores of union loyalty (median = 37) and organizational loyalty (median = 44) obtained at Time 2. The means, standard deviations and univariate f statistics for the four groups are presented in Table 2. Table 2. Means, standard deviations and univariate F tests tor the four loyalty patterns Organizational loyalty Pattern Dual allegiance Unilateral organizational allegiance Unilateral union allegiance Dual non-allegiance Univariate P statistic Mean 50.93 50.66 32.96 34.09 P = 226.98, p SD 4.79 4.80 8.32 7.75 < .001 Union loyalty Mean 46.65 26.85 47.01 28.00 F = 244.32, p SD G.69 I.IG 7.17 6.73 > .001 The first major issue addressed in the present study was whether prior organizational and union experiences predict current membership in one of the four groups. To assess this, a multiple discriminant function analysis was computed. The organizational and union experiences assessed at Time 1 were treated as the predictor variables, and group membership at Time 2 was the outcome variable. The second phase of the analysis aimed to determine whether different patterns of loyalty exert different organizational and union consequences. The four groups constructed on the basis of scores of organizational loyalty and union loyalty at Time 2 were treated as independent variables in a 2 X 2 multivariate analysis of covariance (MANCOVA). The dependent (outcome) variables consisted of organizational and union experiences measured at the third and final time period. The covariates in this analysis were age, and union and organizational tenure, all three of which have been associated with union and organizational attitudes and behaviours. Results Zero-order correlations between the study variables both within and across the different time waves are presented in Table 3. We will present separately the results of the discriminant analysis used to predict membership of one of the four groups and the multivariate analysis of covariance to isolate outcomes of membership of these groups. Prior to conducting the analyses presented, the possibility that the predictors and outcomes of different patterns of union loyalty and company commitment would differ as a function of the nature of the union was tested because of the diverse membership profiles of the two unions (see Table 1). As no substantial differences emerged across the two unions in terms of the predictors or outcomes of the different patterns of commitment, the data from both unions were pooled for all subsequent analyses. Predicting group membership: Discriminant function analyses A direct discriminant function analysis was computed using personal, job and unionrelated variables as predictors of membership in the four commitment groups. Three 136 Clive FuUagar and Julian Barling i §-^ CO ^ o I o I o I o I c -o ^ C MN <U U o o I CN I CN « "^ G CN t^ CN I I I CN 00 O I I I CN NO '-^ CN * I * # \O O " ^ O m CTs O O O O (N O N C C N s C S n I C r N n O O N C T N C o N C o N w I C c N C c N o . — o l O 00 n O C N (U (N 2 o rN ^ 00 G^ - ^ • ^ o o o o o1 o o 1 1 * * * * * * * * « o o CO 00 S i T ** # r I I I I I I I I i) o .9 o o o -= a. S .a • w v/N o o o 1 o o I o I CN I 1 ^ c % " * -00 C N * * * * * C O I .-^ o o t O CM l ^ ' — ^ ^ C N - H I / - \ 0 0 C N C N O P O C N > ^ O O O r N | . - . , O O ^ - - O O O O ^ o o o I I I I -13* r^ U .2 S | -02 O M g ss e -21* NO 1 1I I1 1 # * o CN o o 1 1 < ^ ' c ^ O N M ^ i y ^ —^ O O O O O O CO CN I I II > NO r— r<-i r.^ -.< rt O O O CN -^ T-- I N ^ ^ r<^ f N m O O O-I - H I I 1 OO i OO 1 11 * * * * C o I a X 00 o CN r ^ I .Sai CO Nio CN "^ ""^ "^ T T T T o I II 1 o .S 5i .2 S 3 4-» IB o S -2 .2 c e h -c < i ° c c §> ' I 3 « .2 3 * < ;> > 5 Dual loyalty to organization and union 137 discriminant functions were derived, but of these only two were regarded as significant. The first two canonical functions explained 66.43 and 27.86 per cent, respectively, ofthe between-groups variability. Measuring the residual discrimination by Wilks' lambda also supported the use of only two functions. Not enough residual discrimination remained to justify the derivation of the third function (Wilks' lambda = 0.91 after the first two functions had been derived: / ( 1 3 ) = 13.48, p > .05). Table 4. Results of discriminant function analysis Correlations of predictor variables with discriminant functions T Inivflrtfifp Predictor variable Extrinsic job satisfaction Intrinsic job satisfaction Job involvement Role conflict Role ambiguity Supervisor support Company tenure Union attitudes Union—management relations Union instrumentality Union involvement Industrial relations stress Union tenure Age Canonical R Eigenvalue 2 (3, 150) -.02 -.06 .09 .75 .01 -.30 -.35 .37 .03 13.39** 15.00** 1.74 2.19 5.10** 3.51* 1.55 40.12** 6.29** 8.81** 18.33** 1.27 3.89* 1.75 1 .13 -.04 .13 .83 .50 .39 -.30 -.08 .26 .04 .73 1.15 .79 .18 .12 .21 .10 .42 .17 -.04 -.07 .57 0.48 n * p < .05; ** p < .01. The loading matrix of correlations between predictors and discriminant functions (see Table 4) indicated that the variables which contributed the most to function 1 were general union attitudes (.83), perceptions ofthe quality of union—management relations (.50), union instrumentality (.39) and union involvement (—. 30). This function therefore was interpreted as a union function because union attitudes and experiences were its main contributors. The second function correlated with extrinsic (.75) and intrinsic (.79) job satisfaction, supervisor support (.37), role ambiguity (—.35) and conflict (—.30), and union involvement (.42). This function was therefore interpreted as an organizational function. An examination of the plots of the discriminant scores along the above two discriminant functions indicated considerable overlap. In other words, the predictor variables did not appear to be too successful in discriminating between the four patterns of commitment. Analysis ofa plot ofthe group centroids supported the above interpretation of the discriminant functions. For example, the union function (function 1) discriminated between high union allegiance groups (union allegiance and dual allegiance) and low union allegiance groups (organizational allegiance and dual non-allegiance), but could not distinguish between union allegiance and dual allegiance, or dual non-allegiance and 138 Clive FuUagar and Julian Barling organizational allegiance. A similar finding was evident with the organization function. Using the discriminant variables in the above two functions, a classification analysis was undertaken to ascertain the extent to which the discriminant functions could successfully predict membership ofthe four commitment patterns. The overall percentage of grouped' cases correctly classified was 62.94 per cent. With four commitment patterns, the expected accuracy is only 25 per cent. Consequently, 62.94 per cent correct prediction is a substantial improvement. The discriminant variables were most successful in predicting dual non-allegiance (71.4 per cent), unilateral organizational (61.5 per cent) and union allegiance (71.7 per cent), and less successful in predicting dual allegiance (46.5 per cent). Outcomes of union and company loyalty: Multivariate analyses of covariance A significant multivariate effect on the dependent variables was generated using the Pillai-Bartlett F approximation for both the overall organizational commitment effect (F(10, 204) = 3.55, p < .Q>\) and the overall union commitment effect (f(10, 204) = 8.98, p < .01), after controlling for age, organizational and union tenure. The multivariate interaction effect was not significant. Consequently, separate univariate ANCOVAs were computed for each dependent variable, with age, union tenure and organizational tenure as the covariates. Prior to the computation of the univariate ANCOVAs, Cochran's C statistic was calculated for each dependent variable to assess whether the underlying assumption of Table 5. 2 X 2 ANCOVA and Wilson's distribution-free ANOVA results Organizational commitmenit(A) High Dependent variable Industrial relations stress General union attitudes Extrinsic job satisfaction Intrinsic job satisfaction Job involvement Role conflict Role ambiguity Supervisor support M 7.58 38.51 41.99 38.86 41.63 29.23 13.84 8.44 Low M Union commitment (B) High F M 8.69 0.40 8.68 37.19 2.58 40.83 36.96 24.66** 38.63 33.71 22.85** 35.43 36.52 8.16** 38.66 32.55 5.26* 30.18 16.35 8.27** 15.25 7.18 10.44** 7.76 A XB interaction Low M F F 7.66 33.44 40.28 37.10 38.31 31.45 14.41 7.66 0.33 62.75** 2.49 2.24 0.13 0.55 0.82 0.06 0.12 0.25 0.02 0.08 0.00 0.04 0.18 0.20 Wilson's ANOVA (x^) Union-management relations Perceived union instrumentality Union involvement * p < .05; **p < .01. 0.45 1.79 0.20 2.94 0.24 0.54 84.05** 0.67 3.16 Dual loyalty to organization and union 139 homogeneity of variance was violated. In situations where this assumption was not satisfied, parametric tests of significance would not be justified (Rogan & Kesselman, 1977). In these cases a non-parametric test would be used. Wilsoti (1956) has provided a factorial ANOVA that is not based on the assumption of homogeneity of variance. Cochran's C statistic revealed that with the exception of union-management relations (C = . 5 1 , ^ < .05), union involvement (C = .65, p < .01) and union instrumentality (C = .62, p < .01), all the dependent variables satisfied the assumption of homogeneity of variance. The results ofthe parametric ANCOVAs and Wilson's (1956) distributionfree factorial ANOVA are presented in Table 5. These results indicate a pattern in the findings. Organizational commitment at Time 2 predicted subsequent organizational outcomes, but had no significant effect on unionrelated variables. Similarly, loyalty to the union at Time 2 predicted union behaviours and attitudes at Time 3, but not organizational experiences. An examination ofthe adjusted means indicated that the direction of t:hese rnain effects were as hypothesized in all instances (see Table 4). In other words, high organizational commitment facilitated greater intrinsic and extrinsic job satisfaction, and job involvement, and less role ambiguity and conflict. High organizational loyalty did not result in lower levels of union functioning. Likewise, high union loyalty did not detract from organizational participation. Discussion The results of the present study confirm the existence of divergent models of organizational and union commitment. Specifically, the analyses support previous findings that union and company commitment tend not to share common predictors, but are determined by distinct factors (Barling, Wade & Fullagar, 1990; Fukami & Larson, 1984; Magenau, Martin & Peterson, 1988; Martin, Magenau & Peterson, 1986). This runs contrary to Stagner's (1954) speculation that dual allegiance is attributable to employees' perceptions of their employer and union as a unit, and that union and company commitment can be predicted by common variables. Furthermore, the inability of demographic factors to discriminate between the four patterns of commitment again indicates the non-existence of a union or organization personality type (Decotiis & LeLouarn, 1981; Gordon «? rf/., 1980). However, even though no personal, job or work-related variables predicted both union and company commitment, one union-related variable did: involvement in union activities was significantly correlated with both the organization and union discriminant functions. Both Stagner & Rosen (1965) and Magenau etal. (1988) have suggested that involvement in union activities facilitates greater union commitment. The present findings go further in that they suggest that involvement in union-related activities predicts union loyalty and company commitment. Consequently, increasing membership involvement may not detract from company commitment. One practical implication of such a finding is that management should not view a union's attempts to increase their members' involvement as necessarily detrimental to the company's health. A comparison of this finding with that of Fullagar & Barling (1989) suggests that the causal direction of the relationship between involvement in union activities and union loyalty may be moderated by such demographic and situational characteristics as 140 Clive FuUagar and Julian Barling occupational/professional status, race, nature ofthe job and type of union. However, in the present instance the relationship between involvement in union activities and union loyalty was negative (r = - . 3 0 ) , * whereas that between union involvement and organizational commitment was positive {r = .42). It would appear from this that involvement in union activities does not necessarily bind the member to the union and facilitate consonant attitudes. Indeed, involvement in union actions and behaviours appeared to trigger a sense of disillusionment with the union, which caused a decrease in members' loyalty to the union and strengthened attitudes of commitment to the employing organization. The implication for labour organizations is that, unless involvement in union activities is perceived as rewarding, positive attitudes of attachment to the union are not necessarily forthcoming, and union members may focus their loyalty on the organization. As Magenau et al. (1988) have suggested, the nature ofthe association between union involvement and union commitment appears to be moderated by the degree of satisfaction with the exchange relationship between the individual and the union. The results ofthe discriminant analysis diverged from previous research on union and organization commitment in one other important respect. Prior research has indicated that extrinsic and intrinsic job dissatisfaction predict attitudes of union loyalty (FuUagar & Barling, 1989). In the present study, extrinsic and intrinsic job satisfaction discriminated between high and low organization commitment, but not between different levels of union commitment. The difference can probably be attributed to two factors. First, the two studies were conducted on two widely differing labour organizations. Whereas' the prior research investigated South African, blue-collar workers who were members of a 'social movement' union (FuUagar & Barling, 1989), the present sample comprised white-collar members of academic staff and support services unions in two universities in Canada. Dissatisfaction with various job characteristics becomes a stronger predictor of union loyalty amongst more alienated, lower occupational status workers compared to more occupationally privileged workers (FuUagar & Barling, 1987). A separate sample of Canadian community college teachers also showed no link between job dissatisfaction and union loyalty (Barling etal., 1990). Secondly, the relationship between job dissatisfaction and attitudes of union loyalty is moderated by perceived union instrumentality (FuUagar & Barling, 1989). In other words, whether job dissatisfaction predicts loyalty depends on whether the individual perceives the union as being instrumental for achieving extrinsic and intrinsic outcomes. To summarize, the results of the discriminant analysis suggest that patterns of commitment are determined by unique organizational and union experiences and attitudes. Consequently, labour and employing organizations should develop attitudes of loyalty and commitment by concentrating on organizationally appropriate experiences. In the organizational context, commitment to the university could be enhanced by developing both intrinsic and extrinsic job conditions and focusing on reducing role ambiguity. To develop union loyalty, academic staff unions should enhance members' general attitudes to unions and perceptions of the labour-management relationship. The one exception to the support for divergent predictors was union involvement, which was found to be a common predictor of both union loyalty and organizational commitment. * Here, r refers to the correlation between job involvement and the canonical discriminant functions (see Table 4). Dual loyalty to organization and union 141 Overall, the organizational and union variables hypothesized to be predictors ofthe four patterns of commitment were capable of discriminating between these patterns 63 per cent of the time. However, they were least successful in predicting dual allegiance. To determine whether the differing patterns of dual commitment had different union and organizational consequences, multivariate analyses of covariance were computed. The results of the ANCOVAs suggested unique consequences for both union loyalty and organizational commitment, and there were no interactions between union and organizational allegiance. These findings again support the concept of divergent models of union and organizational allegiance. High loyalty to the union only exerted an impact on union-related consequences, in that it facilitated subsequent positive attitudes to unionism, greater involvement in union activities, and more positive perceptions of the union—management relationship. Siinilarly, high organizational commitment predicted greater intrinsic and extrinsic job satisfaction, increasing job involvement, reducing both role ambiguity and conflict, and fostering stronger perceptions of co-worker and Supervisor support. It would appear therefore that the consequences of union and organizational allegiance differ. The results of the multivariate analyses of covariance have important implications for both unions and universities a? organizations. Specifically, high commitment to the union does not appear to have any adverse effect on job and organizational attitudes. Indeed, no effect was found. Likewise, attitudes of commitment to the employing organization have no negative impact on attitudes to unionism, involvement in union activities, or perceptions of the union—management relationship. It would appear that union and organizational commitment are separate concepts amongst white-collar, academic staff union members in that they have unique and distinct predictors and outcomes. On a more pragmatic level, concerns and fears expressed by unionists and management alike that high commitment to either will necessarily exert negative impacts on the other received no empirical support from the data collected during this study. It would be interesting to test whether this 'distinctness' generalizes to blue-collar workers and other types of unions where the interdependency between management and labour is greater. Acknowledgements This research was supported by a grant from the Social Sciences and Humanities Research Council of Canada (grant no. 410-85-1139) to the second author. The authors express their appreciation to W. H. Cooper, P. Knight, K. E. MacEwen and F. Saal for valuable advice concerning earlier versions of this manuscript, and to Michele Laliberte for her invaluable assistance during the initial phases of this project. References Angle, H. J. & Perry, J. L. (1984). Dual conimitmeat and labor-management relationship climates. Af<?<ife»;i of Management Journal, 29, 31—50. Barling, J. & McLennan, J. (1983). Correlates of commitment to company and union. Unpublished manuscript. Department of Psychology, Queen's University, Kingston, Ontario. Barling, J. & Milligan, J. (1987). Some psychological consequences of striking: A six month longitudinal study. Journal of Occupational Behaviour, 8, 127—138. Batling, J., Wade, B. & Fullagar, C. (1990). Predicting employee commitment to company and union: Divergent modtli. Journal of Occupational Psychology, 6 3 , 49—61. l42 Clive Fullagar and Julian Barling Beutell, N. J. & Biggs, D. L. (1984). Behavioral intentions to join a union: Instrumentality, valence, locus of control, and strike attitudes. Psychological Reports, 55, 215—222. Biasatti, L. L. & Martin, J. E. (1979). Ameasureofthequalityofunion—management relationships._/oara(?/o/ Applied Psychology, 64, 387-390. Bluen, S. D. & Barling, J. (1987). Stress and the industrial relations process: Development of the Industrial Relations Events Scale. South African Journal of Psychology, 17, 150—159. Bluen, S. D. & Barling, J. (1988). Psychological stressors associated with industrial relations. In C. L. Cooper & R. Payne (Eds), Causes, Consequences and Coping with Work'Stress, pp. 175—205. London: Wiley. Brett, J. M. (1980«). Behavioral research on unions and union management systems. In B. M. Staw & L. L. Cummings (Eds), Research in Organizational Behavior, vol. 2. Greenwich, CT: JAI Press. Brett, J. M. (1980^). Why employees want unions. Organizational Dynamics, Spring, 47-59. Buchanan, B. (1974). Building organizational commitment: The socialization of managers in work org&nizitions. Administrative Science Quarterly, 19, 533-546. Chacko, T. I. (1985). Member participation in union activities: Perceptions of union priorities, performance Bind sdLtisfaction. Journal of Labour Research, 6, 364—373- Cook,J. O. Hepworth, S. J., Wall, T. D. & Warr, P. B. (1981). The Experience of Work: A Compendium and Review of 249 Measures and their Use. London: Academic Press. Dean, L. R. (1954). Union activity and dual loyalty. Industrial and Labor Relations Review, 1, 526-536. DeCotiis, L. A. & LeLouarn, J. (1981). A predictive study of voting behavior in a representation election using union instrumentality and work perceptions. Organizational Behavior and Human Performance. 27, 103-110. DeCotiis, T. A. & Summers, T. P. (1987). A path analysis of a model of the antecedents and consequences of organizational commitment. Human Relations, 40, 445-470. Dunlop, J. T. (1987). Industrial relations — Old and new. The First Annual Don Wood Visiting Lectureship in Industrial Relations, Queen's University, Kingston, 19 November. England, G. W. (I960). Dual allegiance to company and union. Personnel Administration, 23, 20—25. Ephlin, D. F. (1988). Revolution by evolution: The changing relationship between GM and the UAW. Academy of Management Executive, 2, 63-66. Fryxeil, G. E. & Gordon, M. E. (1989). Workplace justice and job satisfaction as predictors of satisfaction with union and management. Academy of Management Journal, 32, 851-866. Fukami, C. V. & Larson, E. W. (1984). Commitment to company and union: Parallel models. Journal of Applied Psychology, 69, 367-371. Fullagar, C. (1986). A factor analytic study of the validity of a union commitment scale. Journal of Applied Psychology, 71, 129-136. Fullagar, C. & Barling, J. (1987). Towards a model of union commitment. In D. Lewin, D. Lipsky & D. Sockell (Eds), Advances in Industrial and Labor Relations, vol. 4, pp. 43-78. Greenwich, CT: JAI Press. Fullagar, C. & Barling, J. (1989). A longitudinal test of a model of the antecedents and consequences of union ityjAty. Journal of Applied Psychology, 14, 213-227. Gordon, M. E., Beauvais, L. L. & Ladd, R. T. (1984). The job satisfaction and union commitment of unionized engineers. Industrial and Labor Relations Review, 43, 359—370. Gordon, M. E., Philpot, J. W., Burt, R. E., Thompson, C. A. & Spiller, W. E. (1980). Commitment to the union: Development of a measure and an examination of its conelates. Journal of Applied Psychology, 65, 479^99. Hamner, W. C. & Smith, F. J. (1978). Work attitudes as predictors of unionization SiCt'mty. Journal of Applied Psychology, 63, 4 1 5 ^ 2 1 . Jans, N. A. (1982). T\3.em.t.\xte3.a.dm.easviKm.eatoi-wor\iio.voh7emeat. Journal of OccupationalPsychology. 55, 57-67. Kanungo, R. (1982). Work Alienation: An Integrative Approach, New York: Praeger. Klandermans, B. (1989). Union commitment: Replications and tests in the Dutch context. Journal of Applied Psychology, 74, 869-875. Kochan, T. A., Katz, H. C. & McKersie, R. B. (1986). The Transformation of American Industrial Relations, New York: Basic Books. Dual loyalty to organization and union 143 Ladd, R. T., Gordon, M. E., Beauvais, L. L. & Morgan, R. L. (1982). Union commitment: Replication and extension. Journal of Applied Psychology, 67, 640—644. Lodahl, T. & Kejner, M. (1965). The definition and measurement of job myobitraent. Journal of Applied Psychology. 49. 24-33. Magenau, J. M. & Martin, J. E. (1985). Patterns of commitment among rank-and-file union members: A canonical analysis. Proceedings of the 37th Annual Meeting of the Industrial Relations Research Association, December 1984, Dallas, pp. 455-464. Magenau, J. M., Martin, J. E. & Peterson, M. M. (1988). Duai and unilateral commitment among rank-and-file union members. Academy of fAanagement Journal. 3 1 , 359—376. Martin, J. E., Magenau, J. M. & Peterson, M. F. (1982). Variables differentiating patterns of commitment among union stewards. Proceedings ofthe Academy of Management, 42, 302-306. Martin, J. B., Magenau, J. M. & Peterson, M. F. (1986). Variables related to patterns of union stewards' commitmeat. Journal of Lahor Research. 7, 323-336. Moos, R. H. (1981). Work Environment Scale: Manual. Palo Alto, CA: Consulting Psychologists Press. Mowday, R. T., Porter, L. W . & Steers, R. M. (1982). Employee-Organization Linkages: The Psychology of Commitment, Absenteeism and Turnover. New York: Academic Press. Porter, L. W. c& Smith, F. J. (1970). The etiology of organizational commitment. Unpublished manuscript. University of California, Irvine. Premack, S. L. & Hunter, J. E. (1988). Individual unionization decisions. Psychological Bulletin, 103, 223-234. Purcell, T. V. (1954). Dual allegiance to company and union - packinghouse workers. A Swift-UPWA study in a crisis situation, 1949-1952. Personnel Psychology. 7, 4 8 - 5 8 . Purcell, T. V. (I960). Blue-collar Man: Patterns of Dual Allegiance in Industry. Cambridge, MA: Harvard University Press. Rogan,J. C. &Kesselman, H . J . (1977). Is the ANOVA F-test robust to variance heterogeneity when sample sizes are unequal.': An investigation via a coefficient of variation. American Educational ResearchJournal. 14, 493-498. Rizzo, J., House, R. J. & Littzman, S. I. (1970). Role conflict and ambiguity in complex organizations. Administrative Science Quarterly, 15, 150-163. Schriesheim, C. A. (1978). Job satisfaction, attitudes towards unions, and voting in a union representation Aectioa. Journal of Applied Psychology, 73, 548-552. Stagner, R. (1954). Dual allegiance as a problem in modern society. Personnel Psychology, 7, Al-Al. Stagner, R. & Rosen, H. (1965). Psychology of Union—Management Relations. Belmont, CA: Wadsworth. Warr, P., Cook, J. & Wall, T. (1979). Scales for the measurement of some work attitudes and aspects of psychological well-being. Journal of Occupational Psychology, 52, 129-148. Wilson, K. V. (1956). A distribution-free test of analysis of variance hypotheses. Psychological Bulletin, 53, 96-101. Zaiesny, M. D. (1985). Comparison of economic and non-economic factors in predicting faculty vote preferences in a union representation election. Journal of Applied Psychology, 70, 243—256. Received 19 July 1989; revised version received 12 November 1990