You are on Érudit's new platform. Enjoy! Switch to classic view

Articles

Dual Commitment to the Organization and the UnionA Multi-Dimensional ApproachLa double loyauté envers l’organisation et le syndicatUne approche multidimensionnelleDoble compromiso a la organización y al sindicatoUna perspectiva multidimensional

  • Aaron Cohen

…more information

  • Aaron Cohen
    School of Political Science,
    Department of Public Administration,
    University of Haifa,
    Haifa, Israel
    acohen@poli.haifa.ac.il

Article body

The concept of dual commitment to the union and the organization attracted the attention of scholars almost half a century ago, with the rise of unionism in the United States in the early 1950s. One of the concerns at that time was that unionization, and consequently commitment to the union, would result in diminished loyalty to the employer. This concern sparked a series of studies on the concept of “dual commitment,” namely commitment to both the organization and the union. Specifically, dual commitment refers to a worker’s positive or negative attachment to both a union and an employing organization. Unilateral commitment describes a positive attachment to a union or to an employer, but not to both of these (Stagner and Rosen, 1965). The studies on this concept concentrated on the relationship between commitment to the employer/organization and commitment to the union (Derber et al., 1953, 1954; Purcell, 1954; Stagner, 1954, 1961).

In subsequent years, the above works were criticized (Angle and Perry, 1986; Fucami and Larson, 1984; Gordon et al., 1980), mainly because of their methodological limitations. Criticism centred on the measurement problems of the two commitment scales, which utilized only one item. Therefore, more valid and reliable measurement tools were needed. Current research in the area of dual commitment improves the instruments used to measure organizational and union commitment. Yet, potential measurement problems arise from the use of one dimension of commitment, either organizational or union commitment, to examine dual commitment. The goal of this research is to apply a multi-dimensional approach to the examination of dual commitment. In line with the relevant literature on organizational and union commitments that apply multi-dimensional scales for examining each commitment, this study will use a multi-dimensional approach to examine dual commitment. By utilizing such an approach, it will examine whether the application of only one dimension to measure each commitment form results in the loss of relevant information about dual commitment.

Conceptual Frameworks for Dual Commitment to Union and Organization

Several conceptual frameworks exist for the concept of dual commitment; three of them delineated by Magenau, Martin and Peterson (1988). The first is based on the notion of conflict, which postulates difficulty in maintaining simultaneous commitments to organizations in conflict with one another. Dual commitment is possible when relations between union and management are positive, but the presence of strong conflict tends to push workers toward unilateral commitment to one side or the other. The literature (Strauss, 1977; Tannenbaum, 1965) suggests that union activists have a strong stake in their employment situation. Therefore, some forms of union participation may carry stronger potential for such conflict to occur than other forms. This conflict will create a situation where the union is opposed to the employer. Magenau, Martin and Peterson’s (1988) results supported the notion of conflict. They found that among union stewards, levels of unilateral commitment to the union (high levels of union commitment and low levels of organizational commitment) were much higher than among the rank-and-file. Walker and Lawler (1979) and Martin (1981) also found empirical support for this notion of conflict. Employees with high levels of organizational commitment eschewed joining militant unions, which might have put their personal welfare at risk.

The second explanation for dual commitment is based on Stagner’s (1954) early work, which described dual commitment as a phenomenon arising from people’s tendency to perceive their work situation as a unit rather than sharply differentiating the union role from the management role. According to Magenau, Martin and Peterson (1988), that explanation implies that conditions such as high job satisfaction and positive union-management relations should lead to dual commitment if they are attributable to the joint actions of the two parties. Accordingly, unilateral commitment appeared to be the rule among people more deeply involved with one side than with the other. A similar explanation of dual commitment is the frustration-aggression theory, mentioned by Klandermans (1986). This approach sees trade union participation as a reaction to frustration, dissatisfaction, or alienation in the work situation. Individuals and the organization are defined as systems striving for equilibrium. If the equilibrium is disturbed, they attempt to restore it. At the heart of this approach is the belief that unions are symptoms of employees’ incomplete integration into the company.

A third explanation is based on the exchange theory, which is the common approach to the understanding of the mechanism of multiple commitments in general, and of organizational commitment in particular (Cohen, 2003). According to this explanation, if an organization serves as a vehicle for the use of an individual’s abilities and satisfies his or her needs, the person reciprocates by commitment to the organization. If an organization fails to serve as such a vehicle, commitment to it is low. High dual commitment, then, should be related to a combination of variables reflecting a perception of a satisfying exchange relationship with both union and employer. On the other hand, unilateral union commitment should be related to a combination of the same variables, reflecting the perception of a satisfying relationship with the union and a less satisfying relationship with the employer. Unilateral employer commitment, the opposite of unilateral union commitment, should be related to a reverse set of conditions than those for unilateral union commitment.

Identification of Dual Commitment

Gordon and Ladd (1990) and Bemmels (1995) pointed out two approaches to identifying dual commitment: the taxonomic and the dimensional. The taxonomy approach classifies individual union members into taxons based on self-report about organizational commitment and union commitment. Subsequent steps in taxonomic research customarily involve the identification of other characteristics of the union members in a taxon, thereby providing a richer description of those individuals and more definite standards for classification. Accordingly, individuals are the units of analysis in the taxonomic approach (Gordon and Ladd, 1990). Works on the concept of dual commitment using the taxonomic approach classified individuals into four quadrants and constructed four-dimensional profiles of the concept. In most cases, they created high and low-commitment groups by cross-tabulating employer and union commitment measures, divided into values greater than, less-than, or equal to the scale midpoints (Bemmels, 1995). These cutoff scores were used to form a consistent definition of high and low commitment, which would be applied across all groups. The four-dimensional classification of employees consisted of those high in both commitments, those low in both commitments, and those low in one and high in the other. This line of research frequently combined the design and analysis of the profiles with discriminate analysis, which determined the variables that best differentiated the groups of profiles.

A summary of these works is presented by Cohen (2003). The findings indicate that concern for unilateral commitment to either union or company was apparently less of an issue than was expected in earlier literature, and the concept of dual commitment is not a zero-sum game (Thacker and Rosen, 1986). Yet, some exceptions to this general conclusion were also found (Beauvais, Scholl and Cooper, 1991; Sverke and Sjoberg, 1994).

The dimensional approach focuses on an organizational phenomenon rather than on individual differences. Instead of individuals being classified into categories determined by their relative union commitment and organizational commitment, dual commitment is evinced from the pattern of scores across individuals presumed to constitute a meaningful institutional entity. Specifically, the dimensional approach determines whether a linear relationship exists between union commitment and organizational commitment within a sample representing a particular organization. The existence of dual commitment is established through tests addressing the magnitude of the correlation between measures of union commitment and organizational commitment. A significant, positive correlation between the two commitments documents the existence of dual commitment; that is, in a given organizational unit, individuals’ convictions about their company tend to match their convictions about their union.

As mentioned above, this approach is based on examining the correlations between the two commitment forms. The accumulated data on this issue led to three meta-analyses on the relationship between the two variables. Reed, Young and McHugh (1994) found a mean corrected correlation of .42, which supported the notion of the existence of dual commitment. Bamberger, Kluger and Suchard (1999) found a corrected correlation of .36 in 41 samples involving a total of 17,935 respondents. Johnson, Johnson and Petterson’s (1999) meta-analysis was based on 31 samples with 84 independent correlations involving 22,012 unionized employees in seven countries. The corrected correlation between the two commitments was .37. All of the above findings based on the dimensional approach generally support the notion of the existence of dual commitment. The relatively high and positive corrected correlation between the two variables across the three meta-analyses strongly supports this conclusion.

A third approach, not mentioned above, is the formulation of a specific scale for dual commitment. Angle and Perry (1986), who developed a five-item scale for dual commitment, applied such an approach. Their findings in a sample of rank-and-file employees of municipal bus companies showed that their dual commitment scale was positively and significantly related to both union commitment (r = .40) and organizational commitment (r = .48). However, this approach does not accord with the notion of commitment as presented in earlier studies. Moreover, very few studies have applied it to dual commitment. The dimensional approach will be applied in this study as one way of examining dual commitment.

Correlates of Dual Commitment

Naturally, an important aspect of the concept of dual versus unilateral commitment is its relationship with determinants and outcomes. Knowledge of the relationship with determinants is important for understanding the causes underlying unilateral or dual commitment. Technically, many researchers have used what is termed the parallel models approach (Fukami and Larson, 1984), where organization and union commitment are each regressed on a set of independent variables to identify common predictors of both union and organizational commitment. These predictors, in turn, are then considered predictors of dual commitment. Dissimilarity in the determinants and/or outcomes might indicate that the two commitments are related to different processes, providing stronger support for the existence of unilateral commitment. A summary of the findings of such research based on one dimension of both union and organizational commitment is presented by Cohen (2003). Fukami and Larson’s (1984) findings that variables related to organizational commitment were not related to union commitment support the notion that the two commitments represent divergent models. Barling, Wade and Fullagar (1990), who replicated Fukami and Larson’s (1984) study, obtained similar findings, as did Magenau, Martin and Peterson (1988), who, in four samples of rank-and-file union members and stewards, found few common predictors of commitment to the union and the organization.

The findings of Sherer and Morishima (1989) and Bemmels (1995) showed more variety in the pattern of the relationship between the determinants and each of the two commitments. Sherer and Morishima found that while work-related variables are usually related to organizational commitment and not to union commitment, variables that represent other factors such as external market and industrial relations may be related to both commitments. Bemmels’ (1995) findings that two styles of leadership behaviours, consideration and initiative structure, were related to both commitments, but in opposite directions, supported this conclusion. This approach will be applied in this study, but with one major difference as compared to previous studies. Instead of using only one dimension for measuring each commitment form, this study will apply a multi-dimensional measurement for union as well as for organizational commitment.

Goals of this Study

The main problem with dual commitment research is that it generally assumes that union commitment and organizational commitment can be measured by using one scale for each. Current research neglects the fact that the standard and the common scales for measuring union commitment and organizational commitment are multi-dimensional. The approach and measurement instrument used by Gordon and his colleagues (1980) are considered the leading tools in the study of union commitment, and most studies on this subject rely to some extent on them (Cohen, 2003). A factor analysis of the scale revealed basically four factors, representing four dimensions of union commitment. The first and pre-eminent dimension was union loyalty, in which two features were reflected. The first was a sense of pride in association with and membership in the union. The second was a clear awareness of the benefits accruing to the individual from union membership. The second dimension was responsibility to the union. It measured the degree of willingness to fulfil the day-to-day obligations to the union and the duty of a member to protect the union’s interests. The third dimension was willingness to work for the union. The fourth dimension was a belief in the concept of unionism.

Using the argument that organizational commitment can be better understood as a multi-dimensional concept, Meyer and Allen (1984) proposed a two-dimensional method to measure it. Conceptually their distinction between the two dimensions paralleled that of the side bet calculative approach of Becker (1960) and the attitudinal approach of Porter and his colleagues (1974). The first dimension was termed affective commitment, and was defined as “positive feelings of identification with, attachment to, and involvement in, the work organization” (Meyer and Allen, 1984: 375). The second was termed continuance commitment, and was defined as the “extent to which employees feel committed to their organizations by virtue of the costs that they feel are associated with leaving (e.g., investments or lack of attractive alternatives)” (Meyer and Allen, 1984: 375). In a subsequent paper, Allen and Meyer (1990) added a third dimension they called normative commitment, defined as the employees’ feelings of obligation to remain with the organization. The factor analysis of Allen and Meyer (1990) supported the proposed three-dimensional scales. In their assessment of the scales, Hackett, Bycio and Hausdorf (1992) generally supported the existence of three dimensions. In subsequent years the three-dimensional approach replaced the OCQ and is currently considered the dominant one. The vast amount of research based on these measures was demonstrated in the meta-analysis summarizing the research that has used them (Meyer et al., 2002).

However, most research on dual commitment has used only one scale for each commitment. Such an approach might limit our understanding of the concept of dual commitment. If both commitments are multi-dimensional, using only one dimension to examine dual commitment provides only a partial understanding of the phenomenon. Questions can be raised as to why one dimension was selected as opposed to another. One can also question whether the findings based on only one dimension provide an accurate understanding of dual commitment. What is the meaning of a correlation between one dimension of organizational commitment and another dimension of union commitment? Would the correlation have been different if a different dimension had been applied to a given commitment? The above questions remain unanswered because few studies have examined dual commitment using multiple dimension scales for each of the commitments.

The goal of this research is to examine dual commitment using multi-dimensional scales. First, using one of the commonly accepted approaches to test for dual commitment, I will examine the correlations among the various dimensions of the scales. Second, I will examine the relationship between all of the dimensions of both union and organizational commitment and a set of organizational and union correlates including turnover intentions regarding the organization, the job and the occupation, job satisfaction, life satisfaction, participation in union activities, union militancy, and propensity to strike. By utilizing this approach, I hope to accomplish two goals. First, I will test the utility of the concept of dual commitment by looking at its relationships with variables that represent union and organizational correlates. Second, if dimensions from both union and organizational commitment are related to the outcomes, this finding will strengthen the utility of the multi-dimensional approach to dual commitment over the use of one dimension for each commitment.

A different but related question that will be examined here is whether a multi-dimensional approach to dual commitment is superior to a uni-dimensional one. Cohen (2003) argued that a multiple commitment approach should apply one dimension from each commitment focus. This approach precludes the need for overly long scales for multiple commitments that increase the probability of common method errors. More specifically, Cohen (2003) recommended that the affective commitment scale be the only scale for organizational commitment and that the union loyalty factor represent union commitment. One way to empirically examine this suggestion is to compare the relationship of the two-dimensional model and the correlates with the relationship of the other dimensions to the correlates. If the other dimensions, namely all of them, excluding affective organizational commitment and union loyalty, explain the variance of the correlates above and beyond the variance explained by affective commitment and union loyalty, then one can conclude that the omission of the other dimensions limits our understanding of dual commitment. It should be noted that no conceptual hypotheses will be advanced regarding the relationship between commitment dimensions and the outcome variables. The main goal of this research is a methodological one whose purpose is to clarify the concept of dual commitment. As such, conceptual hypotheses are not the main focus of this study.

Research Design

Sample

This study focuses on members of the Union of Nurses in Israel and examines employees’ dual commitments toward the occupational union which represents them and toward the organization which employs them. Of the 489 nurses surveyed, the majority were divided almost equally between two hospitals in the northern part of Israel, with 204 working in one hospital and 207 in another. The remaining 78 worked in various clinics also located in the northern part of Israel. Questionnaires were distributed to all of the nurses in the two hospitals and the clinics. To increase respondent participation (Silverblatt and Amann, 1991; Sudman, 1976), the data was collected by research assistants who approached co-workers. Questionnaires were distributed only to those who agreed to participate. More than 65% of those who received the questionnaire returned it. Demographically, 91% of the respondents were females, and 80% of the respondents were married. The mean age for the total sample was 39.2; the mean tenure in the organization was 12.5 years, and the mean number of years in the occupation was 16.3. The average educational level was “some college.”

Definition, Measurement and Analysis Procedures

Commitment forms

Union commitment was measured by the full multi-dimensional scale developed by Gordon et al. (1980). Sixteen items were used to measure union loyalty, seven items to measure responsibility to the union, four items to measure willingness to work for the union, and three items to measure belief in unionism. All items were coded so that higher scores reflected greater commitment. Organizational commitment was measured by the three-dimensional scale created by Meyer and Allen (1991). Eight items were used to measure affective commitment, eight items to measure calculative commitment, and eight to measure normative commitment.

Dependent variables

Turnover intention was measured by three items based on Mobley et al.’s (1979) definition and following similar measures applied in the literature (Michaels and Spector, 1982; Miller, Katerberg, and Hulin, 1979). Accordingly, respondents were asked to indicate their agreement with the following three items on a five-point scale: (1) “I think a lot about leaving the organization,” (2) “I am actively searching for an alternative to the organization,” (3) “As soon as it is possible, I will leave the organization.” There has been a tendency in recent literature to view turnover as a multi-dimensional construct (Cohen, 1993). Therefore, data on three dimensions of turnover intentions were collected in this study—from the organization, the job, and the occupation. This data collection was accomplished by using the same items, but replacing the term “organization” with “job” or “occupation.” The scale ranged from 1 (strongly disagree) to 5 (strongly agree), so a higher score meant weaker turnover intentions. Life satisfaction was measured by five items based on Vrendenburgh and Sheridan (1979). The respondents were asked to indicate on a scale from 1 (very dissatisfied) to 5 (very satisfied) how satisfied they were with the following five aspects of their life: residence, non-work-activities, family life, friendship, and health. As for job satisfaction, respondents were asked to indicate how satisfied they were with six aspects of their job: current job; co-workers; supervisors; current salary; opportunities for promotion; work in general. The scale ranged from 1 (very dissatisfied) to 5 (very satisfied). This measure was developed by Schriesheim and Tsui (1980). Cronbach’s alpha in this sample (.71) was quite similar to the one reported by Tsui, Egan and O’Reilly (1992) in an American sample (.73).

Union activity was measured using ten items partially based on the Gordon et al. (1980) measure. Respondents were asked how active they were in ten activities related to their union: voting, attendance at general membership meetings, knowledge of the union contract, applying to the union for assistance, reading the union’s newsletter, membership in the union local, membership in any other institution of the union, participation in any cultural training or educational activity initiated by the union in the last year, applying for any of the union offices in the past year, support of union activities. In accordance with previous scales of union participation (Gordon et al., 1980), the scale for each item was dichotomous with 0 = not active and 1 = active. Therefore, the final scores for both formal and informal types of activities ranged from 0 to 10.

The measure of propensity to strike was similar to that used by Martin (1986). Respondents were asked how many days they would participate in: (1) any strike called by their union, regardless of the issue (or issues); (2) a strike for a small (10%) wage increase; and (3) a strike for a large (25%) wage increase. The response scale for each question was: 1 = zero days; 2 = for one week; 3 = for two weeks; 4 = for one month; 5 = for two months; 6 = for more than two months. Attitudinal militancy was defined and measured by four items. Respondents were asked to predict their behaviour should their union strike: (1) “I would never be a ‘strikebreaker;’” (2) “I would participate in a strike even if I knew that it was an illegal strike;” (3) “I would not hesitate to picket for the union;” (4) “I would be ready to warn strikebreakers.” The respondents were asked to indicate their degree of agreement with each question on a scale ranging from 1 = strongly disagree to 5 = strongly agree.

Finally, because the data were collected from three organizations, I controlled for the possibility that the values for both the independent and dependent variables varied systematically. This was done by creating dummy variables that represented the organizations. The first one was the clinics, the second one was hospital 1, and the reference group was hospital 2. The dummy variables were entered in the first step of each of the regression equations.

Findings

Table 1 presents the intercorrelations among research variables together with their reliabilities, means, and standard deviations. It should be noted that the fourth scale of union commitment, namely belief in unionism, revealed very low reliability: .46. Gordon et al. (1980), who developed the union commitment scales, suggested that this factor might be considered purely an artifact of negatively worded questions. In this study, we decided to omit this scale because of its low reliability. As can be seen from Table 1, reliabilities of all other scales are acceptable and are all above .65.

Table 1

Basic Statistics and Correlation Matrix (reliabilities in parentheses)

Basic Statistics and Correlation Matrix (reliabilities in parentheses)

N = 413–489

*p ≤ .05; ** p ≤ .01; *** p ≤ .001

-> See the list of tables

The intercorrelations among the scales show a relatively high correlation between normative organizational commitment and continuance commitment (r = .63), as well as with affective commitment (r = .74). These correlations are quite typical of the relationships among these variables (Meyer et al., 2002), although the one between normative and continuance commitments is higher than that found in most research. Perhaps the relationship among the three dimensions is higher in countries such as Israel than in the North American countries studied by Meyer et al. (2002). As for the union commitment dimensions, the only strong correlation is the one between willingness to work for the union and responsibility for the union (r = .56). It should be noted that Blalock (1972) argued that whenever there are high correlations between independent variables, it is necessary to have both large samples and accurate measurement. Both conditions were met in this research. The sample was well above 400, and both scales are considered established ones with good reliabilities. All other correlations among the independent variables were acceptable.

To establish further the discriminant validity of the scales applied here, I performed a confirmatory factor analysis using the AMOS structural equation modeling program. In this analysis, I compared the fit of a three-factor model for organizational commitment and union commitment to the alternative fit of three two-factor models and a single, one-factor model. For organizational commitment, the results for the three-factor model (the model incorporating the three independent subscales) revealed the following fit indices: X2 = 1055.3, df = 249; X2/df = 4.24; GFI = .82; CFI = 0.75; TLI = .72. In the second model that was tested, all 24 items were loaded onto a single factor. The findings revealed a X2 = 1308.33, df = 252; X2/df = 5.19; GFI = .78; CFI = .67, NFI = .62, and TLI = .63. In the third model, three two-factor models were tested. All showed a poorer fit than did the three-factor model. The findings thus support the superiority of the three-factor model over the one or the two-factor models.

The same pattern of findings was evident for union commitment: X2 = 1630.55, df = 321; X2/df = 5.08; GFI = .77; CFI = .64; TLI = .60. In the second model that was tested, all 27 items were loaded onto a single factor. The findings revealed a X2 = 1907.21 df = 323; X2/df = 5.90; GFI = .72; CFI = .56, NFI = .52, and TLI = .52. In the third model, three two-factor models were tested. All showed a poorer fit than the three-factor model. It should be noted that the fit indices are not high. Yet, those for the three-factor models are superior to any alternative model for both commitments and thus support the use of the three dimensional scales.

In addition to validating the methodology, the correlation analysis also provides some preliminary information relevant to the goals of this study. The main finding of the correlation analysis is the positive and significant correlation among all except one of the six dimensions of union and organizational commitment. (The one exception is the correlation between responsibility to the union and affective commitment.) This finding supports the existence of a dual commitment based on the multi-dimensional approach. Naturally, a solid conclusion cannot be drawn based on a bivariate analysis. Therefore, several regression analyses were performed to obtain multivariate data that will provide the basis for stronger conclusions.

Tables 2 and 3 present the results of the hierarchical regressions. In step 1, the dummy variables representing the organizations examined here were entered. In step 2, two commitment dimensions were entered into the equation—affective organizational commitment and union loyalty. In step 3, the other four dimensions of both organizational and union commitments were entered into the equation. The rationale for these hierarchical regressions was to test whether and how the other dimensions of commitment were related to the outcomes and contributed to the explained variance above and beyond the variance that was explained by affective commitment and union loyalty. The difference between the tables is that Table 2 presents the relationship between commitment dimensions and work and non-work correlates, while Table 3 presents the relationship between dimensions of commitment and union correlates.

The findings showed that in general, many of the dimensions of the two commitment forms were related to the outcomes. This relationship was true for six out of the eight outcomes. The exception was attitudinal militancy and propensity to strike. Only dimensions of union commitment were related to these two outcomes. All of the other outcomes were related to dimensions from both commitments (see Tables 2 and 3, step # 3). Affective organizational commitment seems to be the dominant factor in its relationship to outcomes, as can be seen from its relationship with the six outcomes. Normative commitment is also related to four outcomes and does have an impact on these variables, although less forcefully than affective commitment. It is significant to note that the continuance commitment was not related to any of the correlates.

As for the union commitment dimensions, both union loyalty and willingness to work for the union are related to the correlates examined here. Union loyalty seems to have a stronger relationship with the variables relevant to the organization and the work setting such as job satisfaction and withdrawal intentions (see Table 2, step # 3). Willingness to work for the union is tied more closely to variables related to the union setting, such as participation in union activities, attitudinal militancy and propensity to strike (see Table 3, step # 3). Responsibility to the union is related to only two variables of the eight examined here.

Table 2

Results of Three Step Regression Analyses (standardized beta) of Dimensions of Organizational and Union Commitment on Work and Non-Work Correlates

Results of Three Step Regression Analyses (standardized beta) of Dimensions of Organizational and Union Commitment on Work and Non-Work Correlates

N = 413–489

*p ≤ .05; ** p ≤ .01; *** p ≤ .001

1. Reference group = Hospital 2

-> See the list of tables

Table 3

Results of Three Step Regression Analyses (standardized beta) of Dimensions of Organizational and Union Commitment on Union Outcomes

Results of Three Step Regression Analyses (standardized beta) of Dimensions of Organizational and Union Commitment on Union Outcomes

N = 413–489

*p ≤ .05; ** p ≤ .01; *** p ≤ .001

1. Reference group = Hospital 2

-> See the list of tables

A comparison of the relationship between the two dominant commitment dimensions to the correlates (see Table 2 and 3, step # 2) and the other commitment dimensions to the correlates (see Tables 2 and 3, step # 3) revealed some interesting patterns. The findings showed that while affective organizational commitment and union loyalty were related to most of the correlates examined, the additional dimensions also had a strong relationship to these correlates. With regard to many correlates, the additional dimensions explained variance much better than did the two dimensions alone.

For example, in the case of union participation, the variance explained by the four dimensions above and beyond the 2% explained by affective commitment and union loyalty was 22%. In the case of attitudinal militancy and propensity to strike, an additional 5% and 4% respectively were explained over and above the 1%–2% variance explained by affective commitment and union loyalty. The four dimensions of commitments added 2% to 8% to the variance already explained by affective commitment and union loyalty for the three correlates of turnover intentions. Only in the cases of job satisfaction and life satisfaction were the contributions of the four commitment dimensions insignificant, with the only relevant variables in those cases being affective commitment and union loyalty.

Finally, because the relatively high correlation among some of the independent variables raises the possibility of multicollinearity, I examined the regressions using the Variance Inflation Factor (VIF). This statistic provides one of the more common measurements for multicollinearity. The rule of thumb for this measure is that a VIF greater than 10 indicates multicollinearity. Some researchers argue for a more rigid criterion, setting a value as low as 5 as an indication of multicollinearity (Haan, 2002). The findings in this study showed that in no case was the VIF higher than 3.5. In the case of normative commitment, the value was 3.33; in the case of continuance commitment, the VIF was 2.36. For all the other independent variables, the VIF was less than 2. Therefore, we may conclude that there is very little chance that this study’s findings have been tainted by multicollinearity.

Discussion

The goal of this study was to examine the multi-dimensionality of the dual commitment to the organization and the union. Most studies that have examined this concept have used only one dimension for each commitment. Organizational commitment has generally been studied using either the nine-item short version of the OCQ (Porter et al., 1974) or the affective commitment scale of Meyer and Allen (1984). Union commitment has been measured using either a self-developed scale or the loyalty dimension of Gordon et al.’s scale.

The findings of this study strongly suggest that examining dual commitment by applying only one dimension for each commitment might result in the loss of valuable information on dual commitment. While affective commitment and union loyalty are related to the correlates examined here, the contribution of the other dimensions is meaningful and cannot be ignored. This conclusion is demonstrated by the regression analyses (Tables 2 and 3) showing that dimensions other than affective commitment and union loyalty are also related to the correlates. The relationship of these dimensions to the correlates seems to be strong, as demonstrated by their standardized betas. Of the organizational commitment dimensions, normative commitment is related to four correlates, and of the union commitment dimensions, it is willingness to work for the union that is strongly related to the correlates examined here.

The above conclusion was strengthened by the analysis performed in Tables 2 and 3, steps # 3. This analysis examined several recommendations that argued for simplifying the concept of multiple commitments in general, and dual commitment in particular (Cohen, 2003). The simplification meant using only one dimension for each commitment—the affective one for organizational commitment and union loyalty for union commitment. The findings here showed that using only two dimensions to study dual commitment might result in only a partial understanding of this phenomenon. In many correlates, the additional dimensions added much significant variance to those already explained by affective commitment and union loyalty.

Two commitment forms are worth elaborating on in this discussion. The first one is willingness to work for the union. This dimension had a very strong relationship to the correlates. In one case, it was related to the correlates in a direction opposite to that of union loyalty (intention to leave the job). In other cases, willingness to work for the union was related to outcomes that union loyalty was not (participation in union activities, attitudinal militancy, and propensity to strike). The importance of this dimension to a thorough understanding of dual commitment was strongly demonstrated by the data. However, a question that might or should be asked regarding this dimension is whether the fact that it is a behavioural tendency rather than an attitude should prevent its future use. Commitment foci are generally attitudes that reflect a particular psychological contract. Porter et al.’s (1974) OCQ was abandoned because it included items that had to do with behavioural tendencies. Future conceptualizations of dual commitment should deal with this dilemma of whether to include behavioural tendency factors as part of the dual commitment construct.

Another commitment dimension that deserves attention here is the continuance commitment. As demonstrated by the regression analyses (see Tables 2 and 3), this dimension was not related to any of the eight outcomes. Many questions have been raised as to whether this dimension has enough construct validity as well as predictive validity to be included as one of the multiple organizational commitments constructs (Cohen, 2003). This study provided additional findings suggesting that the answer might be no. While one may argue that this dimension can be better understood as multi-dimensional in itself (Meyer et al., 2002), there are more and more findings demonstrating that this dimension adds too little to our understanding of commitment. This study is no exception in that regard.

In sum, the findings of this study exemplify the complexity of the construct of dual commitment to the organization and the union. The development of multi-dimensional definitions of and scales with which to measure organizational and union commitment affect this construct. Indeed, it would have been more surprising had that not been the case. Future researchers studying dual commitment need to justify more completely the dimension they select to represent, either organizational or union commitment. As always, the dilemma is to choose between the more simplified approach of one dimension for each commitment and the more thorough approach that better represents the notion of dual commitment. The findings here showed that whatever approach is adopted in future research, it must be justified in terms of its limitation compared with the approach that was not selected.

The study has some limitations. First, it is cross-sectional and as such might suffer from common method error. Also, due to its design, the study cannot draw causal conclusions. Second, the sample is comprised of nurses in Israel, and we may not be able to generalize the findings to a different population. Despite its limitations, the study makes important contributions to the continuation of research on dual commitment to the union and to the organization. It demonstrates the multi-dimensionality of this construct that has to be dealt with in future research on this important and interesting concept.

Appendices