Job Satisfaction and Union Participation: The Role of Fit

Abstract

Job satisfaction’s role as an antecedent to union participation has often been proposed as a negative relationship, but empirical support is lacking. To clarify boundary conditions of this relationship, we turn to the exit-voice tradeoff and the attraction-selection-attrition framework. We suggest a negative job satisfaction–union participation relationship exists only among workers lacking fit with their colleagues (“person-workgroup fit”). We employed a distance-based measure of person-workgroup fit to analyze data from 777 workers across three unions (90 percent public sector) located in a large Midwestern city. Results indicate fit’s moderating role—relatively high fit workers participate in union activities irrespective of their job satisfaction, but workers with relatively low fit participate more when dissatisfied with their jobs. Our findings inform theory on antecedents of union participation and the strategic choices unions face in organizing and reinvigorating lay activism.

Keywords

job satisfaction union participation person-environment fit moderation RIASEC vocational interests exit-voice tradeoff attraction-selection-attrition

Introduction

It’s a long struggle to change an engrained practice where people don’t participate, and don’t expect to participate, and don’t expect their participation to make a difference. (Union officer quoted in Hickey 2005, 290)

Lay participation is essential to unions’ viability and growth, yet increasing participation is hard fought. Although job satisfaction’s role as an antecedent to union participation has been well studied, the most recent meta-analysis of the two variables revealed a nonstatistically significant relationship between them (Monnot, Wagner, and Beehr 2011). We seek an explanation for this surprising finding by examining an important missing variable—union members’ fit with their workgroup. We define a workgroup as individuals in similar occupations working within the same organization.

Correlates of union member participation have been the subject of numerous studies for decades. Participation has often been linked to job satisfaction measures such as the nature of the work the job occupant does (Cranny, Smith, and Stone 1992), job security (Artz and Ilker 2014), compensation (Kochan and Helfman 1981), working conditions (Reynolds and David 2012), employee autonomy (Kalleberg 1977), and union efficacy (Givan and Lena 2012). Union member activism has also been studied as a construct of trust among coworkers and leaders, and a predisposition toward participation (Hoell 2004). Scholars have also identified multiple dimensions for participation ranging from high-commitment activities like holding office to low-investment informal displays of support for the union’s agenda (Parks, Gallagher, and Fullagar 1995). Hickey, Kuruvilla, and Tashlin (2010) offered a heretical perspective by questioning whether rank-and-file activism was even necessary for union effectiveness. In all, single theory models have proven problematic, and observers have relied on multiple facets to explicate the relationship between union member participation and job satisfaction.

One strand of research, however, does offer particular insight into the role that person-environment (PE) fit contributes to job satisfaction and union member participation. Studies that have prioritized characteristics of the work environment that include the individual worker, his or her association with the work group, and the jobs performed offer a glimpse of how work fit can moderate union activism (Leicht 1989; Spinrad 1960). Where unions are able to build strong ties to the members, encourage individual workers to make personal contact with coworkers, shape characteristics of the work environment, and voice dissatisfaction with specific job characteristics, conditions for union participation are heightened.

Understanding all antecedents of union participation is, thus, important to both unions and labor scholars. Job satisfaction’s specific relationship to union participation has generally been thought to be negative, but this proposition lacks empirical support. Although Bamberger, Kluger, and Suchard (1999) established a statistically significant negative meta-analytic correlation of union participation and job satisfaction, correlations became more disparate as the literature expanded. Monnot, Wagner, and Beehr’s (2011) updated meta-analysis ultimately found a nonsignificant relationship between the two variables.

The null bivariate relationship reported in Monnot and colleagues’ meta-analysis suggests moving away from viewing job satisfaction as a union participation antecedent. They present path models developing a more holistic understanding of participation. We seek to understand if, when, and how a negative job satisfaction–union participation relationship exists to help inform union organizing efforts and models of participation. Indeed, rank-and-file participation and voluntary activism serve as the “very fabric of unions” (Gordon, Beauvais, and Ladd 1984, 480). Knowing workers most likely to require encouragement to participate in union activities helps focus union outreach, and theoretically derived moderators of the job satisfaction–union participation relationship are worth consideration. Naturally, if certain members participate more when they are less satisfied, unions should consider how to engage these workers and seek to understand why dissatisfied workers are generally their greatest participants. To this end, we believe accounting for “misfits,” those who do not share the interests and values of their workgroup, stands to determine whether a negative relationship exists between job satisfaction and union participation.

We suggest that workers most similar to their colleagues’ experience “natural” solidarity and stay involved in their union regardless of job satisfaction. Workers who identify with one another share a commitment to “mutual protection” and “shared norms.” At its core, solidarity is formed out of “shared experiences at work” and “the sense of involvement and attachment” that develops among workers (Goffee 1981). Critical to this formation is the fostering of “active work-group social relationships” and normative and cultural processes that generate individual participation and group solidarity (Fantasia 1988; Taylor and Whittier 1992). Although not always oppositional, this “groupness” allows for a collective identity that is separate from management (Hodson 1991).

Consider one union operating engineer who told a study author that working with “like-minded union brothers” provided a sense of belonging and pride motivating his union involvement (Marc Poulus, personal communication, September 10, 2016). Misfits who do not share the interests of their workgroup might still participate in their union but for additional reasons. Specifically, job satisfaction stands to meaningfully relate to union participation for these individuals. Consider a union teacher who had not immediately connected with her union brothers and sisters but was driven to participate to “protect [her] profession” from being treated as “cogs in a wheel” (personal communication from a member of a large teachers union who agreed to be interviewed by one of the authors, but wished to remain anonymous, April 4, 2013). Thus, although high-fit union members may participate regardless of their satisfaction, misfits are different. Those not satisfied at work are likely to either leave or participate in their union as an alternative to exit. In other words, dissatisfied misfits persisting with an employer are likely to participate in their union. Satisfied misfits, however, represent a unique subgroup, leading us to predict that they are less likely to participate in union activities as a group. In the remainder of this article, we develop these ideas in view of the exit-voice tradeoff and the attraction-selection-attrition framework.

We used primary data from members of three unions (90 percent public-sector) operating in a large, Midwestern city, to evaluate how members’ fit with their workgroup determines when a negative relationship exists between job satisfaction and union participation. This moderation analysis was performed using a distance-based measure of interest fit to compare both individual union members’ interests and the general interests characterizing their workgroups.

Theory and Hypothesis

Antecedents of union participation have been studied extensively by labor scholars culminating in two meta-analyses on the subject (Bamberger, Kluger, and Suchard 1999; Monnot, Wagner, and Beehr 2011) that found positive relationships with pro-union attitudes, union instrumentality perceptions, union commitment, and organizational commitment. Other antecedents have been proposed and substantiated thereafter (e.g., perceived behavioral control; Fiorito, Padavic, and Russell 2014). Job satisfaction’s role as an antecedent to union participation has often been proposed and evaluated as well. Yet, as reviewed previously, consistent empirical support is lacking.

Theoretical backing for a negative job satisfaction–union participation relationship follows from several theoretical developments including the exit-voice trade-off and the attraction-selection-attrition framework.

The exit-voice tradeoff suggests workers respond to discrepancies between desired and actual environmental circumstances by either exiting the undesirable situations or by expressing dissatisfaction through “voice.” This offered an explanation for lower mean job satisfaction among union workers compared with nonunion workers (Borjas 1979; Freeman 1980). In line with the present study, the exit-voice tradeoff has been extended to union-only samples by evaluating the relationship between individuals’ job satisfaction and the extent to which they participate in their union (e.g., Iverson and Currivan 2003).

Taken together, we suggest person-workgroup (PW) fit determines whether job satisfaction and union participation are negatively related. PW fit falls under the domain of PE fit, which describes the “congruence, match, or similarity between . . . [a] person and [their] environment” (Edwards 2008, 168). PW fit refers to the similarity of a worker’s interests to the interests generally held by his or her colleagues, specifically those working in similar occupations within a given organization.¹ This follows Holland’s (1966, 53) idea that environments reflect the people in them. In his words, “the dominant features of an environment are dependent upon the typical characteristics of its members” (emphasis in original). Holland (1997) eventually operationalized occupational environments as the distribution of personality types in a given occupation, an idea we extend to workgroups.

Theories of PE fit, including Holland’s conceptualization of interest fit, describe individuals as preferring and seeking out environments compatible with personal characteristics (Kristof-Brown, Zimmerman, and Johnson 2005). Individuals in compatible environments usually exhibit desired personal and organizational outcomes, including reduced stress, higher job satisfaction and performance, and lower turnover (Kristof-Brown, Zimmerman, and Johnson 2005; Nye et al. 2012).

PW fit provides a natural application for considering PE fit in unionized settings. A worker’s unionized colleagues provide an immediate, day-to-day connection to his or her union. Thus, when workers share the interests and values of their workgroup (i.e., experience high PW fit), they have reason to involve themselves in their union apart from their level of job satisfaction. Union participation provides connection and camaraderie; solidarity can form organically. However, as a whole, PW misfits do not enjoy the same connection to their colleagues as those with high fit. Camaraderie does not develop as easily, and solidarity does not follow so naturally. Thus, to the degree misfits find satisfaction in their work, they are less likely to participate in their union absent intentional union efforts to engage and include them.

To fully understand our propositions, consider the experiences of workers relatively high and low on both PW fit and job satisfaction. Individuals with high fit but low satisfaction are relatively likely to participate in their union. High fit suggests they have similar interests as their colleagues and are likely to find their work engaging and interesting, but low satisfaction indicates there may be circumstances and stressors related to their work for which they want to see change. Union participation provides a means to promote change to move their unsatisfying work toward a place of satisfaction. Participating in their union affords them voice to achieve this end while connecting with the colleagues they fit so well with.

Individuals with high fit and high job satisfaction are also relatively likely to participate in their union. As before, high fit again suggests they have similar interests to their colleagues and are likely to find their work engaging and interesting. Their high job satisfaction indicates agreeable work circumstances. Participation with their fellow union members follows high fit with their workgroup and enjoyment of the work itself. Those with similar interests as their workgroup most naturally select into and persist in the workgroup. Solidarity forms naturally and is stoked by union participation.

Taken together, union participation for high-fit individuals should remain relatively high regardless of union members’ job satisfaction. Similarity to their workgroup promotes solidarity and union involvement at all levels of job satisfaction. Individuals can participate in their union to promote change, or simply to connect with likeminded colleagues, a byproduct of sharing the dominant vocational interests of their workgroup.

Following the exit-voice tradeoff, relatively low-fit workers with low job satisfaction are likely to leave their employer unless they can change their work or working conditions. Participating in their union provides these workers the voice alternative to change the rather grim circumstance they find themselves in: poor fit and dissatisfying work. The reality of both poor fit and low satisfaction suggests these individuals who persist with their employer will be especially motivated to participate in their union to promote change. The exit-voice tradeoff describes low-fit, low-job satisfaction individuals well. Although neither fit nor satisfaction compels them to stay in their job, voice (i.e., union participation) can empower them to stay.

On the other hand, individuals with relatively low fit but high job satisfaction are less likely to participate in the union than their fellow union members. Although generally satisfied with their work, the employees’ low fit suggests a relative disconnect from other union members holding the dominant vocational interests of their workgroup. The reality that their job satisfaction remains high suggests these workers see little need for change at work. Because camaraderie and solidarity are also not expected to occur as naturally, union participation should be relatively low for satisfied misfits.

The exit-voice tradeoff is descriptive of relatively low-fit individuals who are not satisfied with their work. These workers are the most likely to either leave their firm or participate in their union. Thus, low-fit workers with low job satisfaction who remain with their employers should participate at higher-than-average levels with their union. On the other hand, union participation from low-fit workers with high job satisfaction should be low—solidarity comes less naturally for misfits, and they have less motivation to seek workplace change via union participation since they are already satisfied with their work. Thus,

Hypothesis 1: The relationship between job satisfaction and union participation is moderated by PW fit, such that for lower PW fit union members, the relationship between job satisfaction and union participation is more negative than for higher PW fit union members.

Figure 1 depicts the path analysis underlying our hypothesis. To summarize, our conceptualization of the job satisfaction–union participation relationship follows relevant developments on the antecedents of union participation and are grounded in the within-union application of the exit-voice tradeoff. We suggest that PW fit moderates the job satisfaction–union participation relationship. Those who fit with their workgroup have reason to participate in their union regardless of their level of job satisfaction, but relatively low-fit union members who are not satisfied are less likely to participate.

Figure 1.

Hypothesized path model distinguishing exit-voice tradeoff in union samples.

As an example of fit’s moderating role of the job satisfaction–union participation relationship, consider teachers (the largest workgroup in our sample). Misfit K-12 teachers may include teachers hired through the Teach for America (TFA) program. TFA teachers are drawn from noneducational academic backgrounds and have not undergone teaching preparatory programs (“Is TFA for You?” 2016). Most of the TFA recruits have noneducational career interests and are pursuing degrees in applied fields. Unlike teachers from undergraduate educational programs, TFA “corps” members enter short two-year teaching commitments and most then move on to their preferred vocational path. Leadership is a strongly sought after characteristic of TFA recruits and most descriptive of individuals high on Holland’s (1997) “enterprising interest,” an interest not typically highest for teachers. If such misfits are ultimately satisfied with their work as teachers, their vocational disconnect from other union members might preclude them from participating fully in their union.

Data and Analysis

In 2015, we invited leaders of seven unions located in a large Midwestern city to ask members of their respective unions to participate in our study. Leaders from five unions representing several industries, and both the private and public sectors, agreed to participate. These unions’ members were emailed directly by their respective leadership with an invitation to complete our study’s linked online survey. Two of the five unions only yielded a single response and were excluded from the sample. The three unions making up our sample include one public- and two private-sector unions providing 1,232 respondents.

All respondents were located in a Midwestern, free collective bargaining (i.e., non-right-to-work) state. The clear majority of individuals in the final sample (90%) were members of the public-sector union. This union represents employees in a large, metropolitan school district. Noted differences exist between private and public unions, including the inability of many public-sector unions to strike. In the present case, however, the public-sector union was in a strike-permissive state. Indeed, the studied union struck in a previous bargaining period and passed a strike-vote in the most recent bargaining period, narrowly avoiding a strike occurrence. All considered, our results hold relevance for both public-sector unions in strike-permissive states and private-sector union members. Furthermore, our results held after controlling for sector.

Measures

Vocational interests

Holland’s (1959, 1997) model of interests provides an operational basis for the person and environment components of PW fit. Holland described the following six interests: realistic, investigative, artistic, social, enterprising, and conventional, which are often referred to by their acronym, RIASEC. Table 1 describes work preferences and occupations characteristic of each interest. The interrelatedness of RIASEC interests is often depicted using a hexagon (see Figure 2; for example, Holland 1997; Tracey and Rounds 1993): adjacent interests (e.g., realistic and conventional) are more related than alternate interests (e.g., realistic and enterprising), which are more related than opposite interests (e.g., realistic and social).

Table 1.

Holland’s (1997) RIASEC Interests and Example Occupations.

Interest	Individuals with this interest seek and prefer . . .	Occupations
Realistic	activities involving the “manipulation of objects, tools, machines, and animals” (p. 43).	Bricklayer, electrician, plumber
Investigative	the “creative investigation of physical, biological, or cultural phenomena” (p. 44).	Chemist, psychologist, biologist
Artistic	“ambiguous, free, unsystematized activities and competencies to create art forms or products” (p. 45).	Graphic designer, actor, craft artist
Social	“inform[ing], train[ing], develop[ing], cur[ing], or enlighten[ing]” others (p. 46).	Nurse, teacher, clergy
Enterprising	“attain[ing] organizational or self-interest goals” (p. 46).	Retail salesperson, Lawyer, CEO
Conventional	the “systematic manipulation of data such as keeping records, filing [and reproducing] materials, organizing written and numerical data according to a prescribed plan, [and] operating business and data processing equipment” (p. 47).	Accountant, librarian, editor

RIASEC = realistic, investigative, artistic, social, enterprising, and conventional.

Figure 2.

Holland’s RIASEC model of interests with Prediger’s people/things and data/ideas dimensions.

Person interests were measured using the Department of Labor’s public-domain Occupational Information Network (O*NET) Interest Profiler Short Form (Rounds et al. 2010). The form is a sixty-item measure with ten items corresponding to each RIASEC interest. Respondents were asked to decide the degree to which they would like or dislike doing a type of work regardless of whether they had education or training to do the work, how it was related to their current job, or how much money they would make doing the work. Responses ranged from −3 (dislike very much) to 3 (like very much) with a neutral midpoint (0 = neither like nor dislike). “Lay brick or tile” represents a realistic item and “perform rehabilitation therapy” a social item. Scale scores were formed by taking the mean of all scale items. Reliabilities (Cronbach’s α) ranged from .82 (social) to .92 (investigative) for RIASEC scale scores.

We evaluated the environmental interests characteristic of the following occupation groups: teachers, clinicians, and paraprofessionals within a metropolitan teachers union, operators from a local operating engineers union, and wiremen from an electrical workers local. The teachers’ group included kindergarten to twelfth-grade teachers, including special education faculty. Clinicians included counselors, social workers, physical and occupational therapists, and speech pathologists. Technicians, assistants, clerks, and secretarial staff made up the paraprofessionals group. Heavy equipment operators and mechanics made up the group of operators, and apprentice and journey wiremen and electricians comprise the group of wiremen.

Occupational interest profiles were formed from the mean RIASEC scores of all respondents within a given occupational group. As in previous studies of interest fit (e.g., Su 2012), occupational interest profiles for each individual were formed from the mean of all RIASEC scores except the individual’s own so as not to inflate the subsequent evaluation of fit. This leave-one-out aggregation technique provides slightly different environmental interest profiles for each individual within a given occupational group. By evaluating person and environment separately, and with an aggregated environment, we are able to calculate an indirect, objective measure of fit (Kristof-Brown and Guay 2011), reducing common method bias in our results (Podsakoff et al. 2003).

PW fit

We evaluated the difference between an individual’s RIASEC interest scores and the average RIASEC interest scores of others working in similar occupations within their organization. Since Holland (1963) first recommended comparing first letters of person and environment RIASEC profiles for agreement, numerous indices have been introduced to measure RIASEC fit. Camp and Chartrand (1992) and Brown and Gore (1994) offer reviews of many RIASEC-specific fit indices. Unfortunately, most of the RIASEC fit indices developed fail to consider entire RIASEC profiles. Zener and Schnuelle (1976) eventually expanded fit indices to consider the first three letters of person and environment RIASEC interest scores, but subsequent indices generally failed to move past this three-letter comparison threshold. Profile correlations allow for a comparison of all six scores but fail to directly account for scale-level differences between person and environment. In other words, so long as person and environment scores correlate with each other, fit scores are high, regardless of the similarity or difference of relative strengths of each component measure. Calculating the Euclidean distance between person and environment scores based on all RIASEC interests accounts for scale-level differences by comparing the distance between all person and environment scores. We operationalized fit as the Euclidean distance between person and environment scores due to this superior property and the parsimony of a single fit metric.

Euclidean distance calculations of person and workgroup RIASEC scores followed Tracey, Allen, and Robbins’ (2012) calculations. First, complete person and environment RIASEC profiles were transformed to points in Prediger’s (1982) two-dimensional People-Things (PT) and Data-Ideas (DI) space. This space represents a parsimonious conception of Holland’s hexagon by transforming the six RIASEC interests into a single point on these two dimensions. The two dimensions can be directly overlaid on the RIASEC hexagon (note their axes in Figure 2).

The PT dimension was calculated as 2*R + I − A − 2*S − E + C such that positive values represent Things interest and negative values represent People interest (letters in the equation represent the first letter of each interest). The DI dimension was calculated as 1.73*E + 1.73*C − 1.73*I − 1.73*A such that positive values represent Data interest and negative values represent Ideas interest. Euclidean distance was then calculated as follows:

d i s t a n c e = \sqrt{{(P T_{p e r s o n} - P T_{e n v i r o n m e n t})}^{2} + {(D I_{p e r s o n} - D I_{e n v i r o n m e n t})}^{2}} .

Very high raw distance values represent low fit (i.e., the person and workgroup scores are far apart), and very low raw values (i.e., approaching zero) represent high fit (i.e., the person and workgroup scores are close together). To aid in interpretability, we multiplied the raw Euclidean distance calculations by −1 and mean centered the results. Thus, negative values represent below average fit in the study sample, and positive values represent above average fit.

Job satisfaction

Studies evaluating overall job satisfaction generally use either Brayfield and Rothe’s (1951) five-item scale or the Job in General Scale (Ironson et al. 1989) and its abridged version (Russell et al. 2004).² In all cases, these scales were not created with union environments expressly in view. In the present study, we used a measure that would account for overall job satisfaction including the unique concerns of unionized workers.

We developed job satisfaction items that assess the concerns of union workers in their vernacular. These items were reviewed and refined with feedback from union members and labor studies faculty before being administered. Respondents were asked to evaluate their agreement with items (sample item: “Since starting my job, I have less time to do the job I was hired to do”) using a seven-point scale (−3 = strongly disagree, 3 = strongly agree) with a neutral midpoint (0 = neither agree nor disagree). Scale items are provided in the appendix.

To evaluate convergent and discriminant validity, Brayfield and Rothe’s (1951) scale was also administered in the present study. Our new scale correlated .60 with Brayfield and Rothe’s measure (using the same seven-point scaling), suggesting we are evaluating a somewhat similar but ultimately unique job satisfaction construct. Our union-specific job-satisfaction scale exhibited strong reliability (Cronbach’s α = .88).

Union participation

Scholars differ on the dimensionality of union participation. Some use a single factor (e.g., Anderson 1979), and others suggest multidimensional models (e.g., McShane 1986; Monnot, Wagner, and Beehr 2011; Parks, Gallagher, and Fullagar 1995). A distinction is most often made between nonmilitant participation, which involves activities that do not require direct interference with work (e.g., helping with union organizing activities), and militant participation, which involves high-intensity activities interfering with one’s work (e.g., striking). As both types of participation are relevant to the present study, we sought a parsimonious measure with items addressing both militant and nonmilitant participation.

Existing scales of both militant (e.g., Martin 1986) and nonmilitant (e.g., McShane 1986) union participation use a mix of Likert, dichotomous, and open-ended numeric items (e.g., “number of union meetings attended”). The mix of scaling within and between instruments makes scale-level interpretation difficult. Instead of combining such items, we began with the existing scales and again drew on the experience of labor studies faculty and union members to create a Likert scale assessing overall union participation. Nine items comprised our scale of overall union participation; they are provided in the appendix. Two items (marked with asterisks) were adapted directly from Martin’s (1986) Militancy scale.

Our scale covered general participation in union activities and service along with militant actions, both legal and those involving acts of civil disobedience or illegal work stoppages. Respondents were asked to evaluate their agreement with four items (e.g., “I would never engage in violence during a strike”; reverse-coded) using a seven-point scale (−3 = strongly disagree, 3 = strongly agree) with a neutral midpoint (0 = neither agree nor disagree) and their participation relative to other employees in their union with five items (e.g., “I help with union organizing efforts”), also using a seven-point scale (−3 = extremely below average, 3 = extremely above average) with a neutral midpoint (0 = an average amount). Scale reliability (Cronbach’s α) was .82.

Controls

We controlled for differences in individuals’ employment sector, status as a racial minority, and level of education. The public-private sector control provides guidance on whether our findings generalize beyond the public-sector union comprising much of our sample. The racial majority control (1 = racial majority) informs whether fit and misfit with the racial in-group should be considered in our fit conceptualization. Finally, the ordinal accounting of education level (1 = some high school to 7 = doctorate) seeks to account for education and education-related status differences in union participation.

Empirical Findings

We estimated the effect of job satisfaction on union participation conditional on fit using ordinary least squares in SAS 9.4 (SAS 2012). The full equation can be written as follows:

\begin{array}{l} P a r t i c i p a t i o n = b_{0} + b_{1} (J S) + b_{2} (F i t) + b_{3} (J S \times F i t) \\ + b_{4} (R M) + b_{5} (E d u) + b_{6} (S e c) + e, \end{array}

where “JS” represents job satisfaction, “RM” is a racial majority group dummy, “Edu” represents education level, “Sec” is a public-sector employment dummy, and “e” represents the error term. Following Aiken and West (1991), simple slopes of union participation at plus and minus one standard deviation of fit and job satisfaction were calculated and plotted along with their 95 percent confidence intervals. Following their guidance for interpreting interactions between continuous variables, a significant interaction between job satisfaction and fit (b₃) and a more negative job satisfaction and union participation simple slope estimate when fit is relatively low (−1 SD) than high (+1 SD) supports our hypothesis.

Table 2 provides descriptive statistics and correlations of our study variables. Fit is mean centered, and job satisfaction and union participation are both centered on their neutral scale midpoint to aid in interpretability. Respondents who correctly answered at least four of five randomly placed quality control items in our survey comprised our final sample of 777 union members (70% female and 30% racial minorities).³ Ninety percent held a bachelor’s degree or higher, and 71 percent held a master’s degree or higher. Respondents’ mean age and organizational tenure was 43.46 (SD = 11.11) and 11.93 (SD = 8.70), respectively. Mean job tenure and current union tenure was 9.63 (SD = 8.60) and 13.22 (SD = 9.35), respectively. Consistent with the job-satisfaction-fit literature, individuals’ job satisfaction positively related with their PW fit (r = .09, p < .05).

Table 2.

Descriptive Statistics and Correlation Matrix of Study Variables.

Variable	M	SD	Range	1	2	3	4	5	6	7	8	9	10	11
1. Job Satisfaction	−0.80	1.45	−3 to 3	.88
2. Realistic	−0.61	1.37	−3 to 3	.07*	.90
3. Investigative	0.32	1.48	−3 to 3	−.02	.57***	.92
4. Artistic	0.84	1.29	−3 to 3	−.05	.21***	.35***	.88
5. Social	1.36	0.99	−3 to 3	−.05	.08**	.25***	.38***	.82
6. Enterprising	−0.14	1.22	−3 to 3	.00	.28***	.30***	.29***	.41***	.86
7. Conventional	−0.44	1.34	−3 to 3	.02	.41***	.30***	.10***	.22***	.55***	.90
8. PW Fit	−0.01	2.82	−11.16 to 5.23	.09**	.04	.07*	.04	.06*	.12***	.07*
9. Union Participation	−0.27	1.06	−3 to 3	−.09***	.22***	.16***	.08**	.03	.01	−.05	.06*	.82
10. Racial majority^a	0.70	0.46	0 to 1	−.03	.07*	−.01	−.08**	−.08**	−.13***	−.21***	−.07*	.04
11. Level of education	5.59	0.84	1 to 7	−.30***	−.24***	−.03	.12***	.26***	.02	.01	−.09**	−.11***	−.09**
12. Public sector^a	0.90	0.31	0 to 1	−.23***	−.31***	−.07*	.15***	.30***	.00	.01	−.13***	−.20***	−.14***	.75***

N = 759-777. Reliabilities (Cronbach’s α) are provided on the diagonal when applicable. PW = Person-Workgroup.

1 = yes.

p < .10. **p < .05. ***p < .01.

Multidimensional scaling (Borg and Groenen 2005) performed in SAS 9.4 (SAS 2012) verified the structure of RIASEC interests for individuals. As shown in Figure 3, the ordering of the RIASEC scales followed the circular ordering of Holland’s hexagon. Confirmatory factor analysis (CFA) was conducted to examine the construct validity of vocational interests, job satisfaction, and union participation. After loading all items onto their corresponding latent constructs, fit indices demonstrated that this model adequately fit the data (root mean square error of approximation [RMSEA] = .07, standardized root mean square residual [SRMR] = .08, confirmatory fit index [CFI] = .73).

Figure 3.

Two-dimensional scaling of person RIASEC interests.

Overall, the characteristic interests of workgroups followed expectations. RIASEC means and PT/DI scores for each of the five workgroups are shown in Table 3. Following O*NET interest profiles of occupations underlying our workgroups (Peterson et al. 1999), we expected teachers, clinicians, and paraprofessionals to all be high on social interest. Indeed, the mean score for social interest was highest for these workgroups and exhibited the least variability across individuals in those workgroups.

Table 3.

Workgroup Interest Profiles.

Variable	Teachers	Clinicians	Paraprofessionals	Operators	Wiremen
Realistic	−0.73	−1.27	−0.44	0.31	0.81
Investigative	0.29	0.07	0.60	0.17	0.85
Artistic	0.90	0.69	1.51	0.23	0.27
Social	1.43	1.76	1.64	0.47	0.50
Enterprising	−0.18	−0.06	0.73	−0.29	−0.08
Conventional	−0.47	−0.63	0.89	−0.89	−0.28
People-Things^a	−5.22	−7.25	−4.91	−0.98	1.00
Data-Ideas^b	−3.18	−2.51	−0.85	−2.73	−2.56
N	621	51	24	27	54

People-Things and Data-Ideas calculations are provided in the text.

Positive values represent Things.

Positive values represent Data.

Wiremen and operators were both expected to be high on realistic interest and were. Compared with the teacher’s union groups, whose average realistic scores represented a “slight dislike,” wiremen and operators both exhibited positive mean realistic scores representing a degree of liking. The mean score for investigative interest was slightly higher than realistic interest for wiremen, but the practical significance of this difference was trivial (.04 scale points). For operators, the mean score for social interest was slightly higher than realistic, but again, the practical significance of this difference was very small (.16 scale points). Figure 4 plots each of the workgroups in Prediger’s (1982) two-dimensional PT/DI space. Work groups’ PT/DI scores followed expectation as teachers, clinicians, and paraprofessionals all exhibited stronger People scores than operators and wiremen.

Figure 4.

Workgroup interest profiles on People-Things, Data-Ideas dimensions.

Moderation results supported our hypothesis. Table 4 provides results of the regression analysis with and without controls. In both cases, the job satisfaction–PW fit interaction was significant as well as the negative simple slope of job satisfaction and union participation when fit was relatively low (–1 SD). The relationship between job satisfaction and union participation when fit was relatively high (+1 SD) was not significant. Table 5 provides the effects of these simple slope results, and Figure 5 displays a plot of the slopes and 95 percent confidence intervals for job satisfaction and fit without controls. A significant interaction term indicates that slopes at ± 1 SD of the moderator (i.e., “simple slopes”) are significantly different from each other (Aiken and West 1991). Simple slope estimates in Table 5 determine whether a slope is significantly different than 0. The relatively low and high fit regions are significantly different for all positive job satisfaction scores.

Table 4.

Determinants of Union Participation.

Variable	Without controls		With controls
Variable	Model 1	Model 2	Model 3	Model 4	Model 5
Intercept	−0.33*** (.044)	−0.34*** (.044)	−0.14 (.291)	0.07 (.295)	0.02 (.295)
Job Satisfaction	−0.07*** (.027)	−0.07*** (.026)		−0.11*** (.027)	−0.11*** (.027)
PW Fit	0.02* (.014)	0.05** (.016)		0.02 (.014)	0.03** (.016)
JS × PW Fit		0.02** (.010)			0.02** (.010)
Racial majority (1 = yes)			0.03 (.084)	0.02 (.083)	0.02 (.083)
Level of education			0.12* (.069)	0.06 (.069)	0.06 (.069)
Public sector (1 = yes)			−0.89*** (.188)	−0.88*** (.188)	−0.88*** (.188)
F	5.36***	5.68***	10.56***	9.74***	8.87***
R ²	.01***	.02***	.04***	.06***	.07***
Δ R²		.01**		.02***	.01**

N = 763 without controls; 743 with controls. Unstandardized coefficients are reported. Standard errors are in parentheses. PW = Person-Workgroup; JS = Job Satisfaction.

p < .10. **p < .05. ***p < .01.

Table 5.

Slope Estimates of Job Satisfaction on Union Participation at Low and High Levels of PW Fit.

Simple slope	Without controls	With controls
Low PW Fit (−1 SD)	−0.14*** (.038)	−0.16*** (.038)
High PW Fit (+1 SD)	−0.01 (.038)	−0.05 (.039)

See notes for Table 4. A significant interaction term (provided in Table 4) indicates that slopes at ± 1 SD of the moderator (i.e., “simple slopes”) are significantly different from each other. Simple slope estimates determine whether a slope is significantly different than 0. PW = Person-Workgroup.

Figure 5.

Job satisfaction by person-workgroup fit on union participation.

To verify that our results hold across nonmilitant and militant forms of union participation, we conducted analyses separately for items 1 to 5 and 6 to 9. In both analyses, simple slope findings did not differ from that of the combined scale (i.e., the slopes were significant and negative for relatively low fit [–1 SD] and were not significant for relatively high fit [+1 SD]). We also compared our results with results using the Brayfield and Rothe measure to evaluate job satisfaction and found that our simple slope findings did not differ. Notably, our job satisfaction scale explained 71 percent more variation of the data in our hypothesis tests (r²) than the Brayfield and Rothe measure.

Discussion and Conclusion

Across five different workgroups in three labor unions covering both public and private sectors, results support fit’s moderating role of job satisfaction and union participation. Specifically, a negative job satisfaction–union participation relationship exists only among relatively low-fit workers. Our hypothesis followed consideration of the exit-voice tradeoff’s application to union-only samples and the role of PE fit’s relationship with participation. Following the attraction-selection-attrition framework, we proposed that high-fit individuals participate in unions regardless of job satisfaction because camaraderie and solidarity form more naturally than among poor fit workers. On the other hand, low-fit workers that do not experience satisfaction are relatively likely to either leave their employer or exercise voice by participating in their union to promote change. Low-fit individuals who are satisfied, but have no natural link to their coworkers, stand out as a group less likely to participate—they have neither dissatisfaction nor workgroup fit to drive their participation.

Our results inform model development of union participation antecedents. Failing to account for moderators of the job satisfaction–union participation relationship can lead to erroneous conclusions. For example, if we had only evaluated job satisfaction and union participation broadly, we would have observed a degree of support for the exit-voice tradeoff. Yet conditioning this relationship on fit rendered the relationship insignificant when fit was high and stronger when fit was relatively low. Thus, as evidenced by past meta-analyses, a negative job satisfaction–union participation relationship does not appear descriptive of all unionized workers. The bounds of job satisfaction’s role as an antecedent to participation should be reconsidered conditioning on workgroup fit.

Unions may benefit from focusing organizing efforts on reaching workers with relatively low fit and job satisfaction. Our finding supports the idea that unions offer a direct “voice” alternative to leaving an employer on account of low job satisfaction and poor work fit. This corresponds with and extends Borjas’s (1979) original union application of the exit-voice tradeoff. He suggested discrepancies between desired and actual work circumstances effectively activate the exit-voice tradeoff. High-fit individuals lack these PE discrepancies, but low-fit individuals do not.

The lack of a negative job satisfaction–union participation relationship for high-fit individuals sheds a positive light on union participation. If a negative job satisfaction–union participation relationship universally existed for workers, as suggested by the exit-voice tradeoff, the role of unions appears rather bleak: individuals who would otherwise leave their employer due to dissatisfaction can instead stay due to the voice their participation affords them, but their satisfaction remains relatively low. Our findings, however, suggest such universal application of the exit-voice tradeoff may be overly broad. Indeed, high-fit individuals participate with their union at relatively high levels without feeling dissatisfaction. Indeed, identifiable conditions exist where members are both satisfied with their work and engaged with their union. Unions and labor scholars should account for such boundary conditions.

Our study is not without limitations. Our sample is largely made up of public-sector union members, which may limit the generalizability of results. Nevertheless, since the public-sector union members were in a strike-permissive state, they arguably share means of participation more common to private-sector unions than public-sector unions in states not permitting strikes. We also controlled for the public- and private-sector distinction, and our findings are consistent with and without these controls. All considered, we suggest our results hold relevance for both public-sector unions in strike-permissive states and private-sector union members.

We are also limited in our ability to make causal claims regarding the effect of job satisfaction or fit on union participation due to the cross-sectional nature of our sample. Future studies might examine our hypothesis using longitudinal data. A more systematic examination of fit’s moderation of the exit-voice tradeoff would include attrition data as well. An event history analysis similar to that of Iverson and Currivan (2003) that accounts for fit would thereby better serve to further clarify fit’s role in the exit-voice tradeoff. Finally, for job satisfaction and union participation, we note that our measures were survey self-reports and, as such, were individuals’ perceptions and not objective or absolute levels of the given constructs.

Future studies would also do well to examine the job satisfaction–fit interaction alongside other antecedents of union participation. Past model development benefited from the strong statistical power characteristic of meta-analyses (e.g., Bamberger, Kluger, and Suchard 1999; Monnot, Wagner, and Beehr 2011), but examinations of union participation and fit are rare and not likely to benefit from meta-analysis until more studies occur. Finally, future studies should consider whether fit distinguishes between job satisfaction differences in union and nonunion samples. It may be that relatively low-fit workers enabled by unions to stay on the job fully account for mean differences in job satisfaction between union and nonunion groups. If this were the case, job satisfaction for high-fit workers would be the same for union and nonunion workers—again providing a more hopeful take on the role unions play in the lives of their members.

In conclusion, this article expanded and linked two literatures not often studied together—union participation and PE fit. Considering both led to proposing fit’s moderating role of job satisfaction and union participation, which we examined using a primary sample of three diverse metropolitan unions. Our findings established boundary conditions to a relationship generally assumed negative. Our study helps guide the development of union participation models in future studies and informs the strategic choices unions face in organizing and reinvigorating lay activism. Unions stand to increase participation by engaging lower-fit, satisfied workers. Unions should also take heart in the distinction high fit affords. Participation in the union can occur alongside members’ satisfaction just as easily as their dissatisfaction.

Footnotes

Appendix

Acknowledgements

The authors gratefully acknowledge Fritz Drasgow and Ryan Lamare for their helpful reviews of earlier drafts of this article.

Authors’ Note

An earlier version of this paper was presented at the Labor and Employment Relations Association 69th Annual Meeting in Anaheim, California (June 2017) and published in the Meeting’s Proceedings.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

ORCID iD

Justin P. Wiegand

Author Biographies

Justin Wiegand leads Sift Human Resources LLC. He received his PhD in Human Resources and Industrial Relations from the School of Labor and Employment Relations (LER) at the University of Illinois, Urbana-Champaign. He conducts research on person-environment fit, vocational interests, union participation, measurement, and narcissism.

Robert Bruno is director of the Labor Studies Program and a Professor of Labor and Employment, as well as the director of the Project for Middle Class Renewal in the School of Labor and Employment Relations (LER) at the University of Illinois, Urban-Champaign.

References

Aiken

Leona S.

West

Stephen G.

1991. Multiple Regression: Testing and Interpreting Interactions. Thousand Oaks: SAGE.

Anderson

John C.

1979. “Local Union Participation: A Re-examination.” Industrial Relations 18 (1): 18-31.

Artz

Benjamin

Ilker

Kaya

. 2014. “The impact of job security on job satisfaction in economic contractions versus expansions.” Applied Economics. Vol. 46, Issue 24, 2873-2890.

Bamberger

Peter A.

Kluger

Avraham N.

Suchard

Ronena

. 1999. “The Antecedents and Consequences of Union Commitment: A Meta-analysis.” The Academy of Management Journal 42 (3): 304-18. http://www.jstor.org/stable/256921.

Borg

Ingwer

Groenen

Patrick J. F.

2005. Modern Multidimensional Scaling: Theory and Applications (2nd ed.). New York: Springer.

Borjas

George J.

1979. “Job Satisfaction, Wages, and Unions.” The Journal of Human Resources 14 (1): 21-40. doi:10.2307/145536.

Brayfield

Arthur H.

Rothe

Harold F.

1951. “An Index of Job Satisfaction.” Journal of Applied Psychology 35 (5): 307-11. doi:10.1037/h0055617.

Brown

Steven D.

Gore

Paul A.

Jr.

1994. “An Evaluation of Interest Congruence Indices: Distribution Characteristics and Measurement Properties.” Journal of Vocational Behavior 45 (3): 310-27. doi:10.1006/jvbe.1994.1038.

Camp

Charissa C.

Chartrand

Judy M.

1992. “A Comparison and Evaluation of Interest Congruence Indices.” Journal of Vocational Behavior 41 (2): 162-82. doi:10.1016/0001-8791(92)90018-U.

10.

Cranny

C.J.

Smith

Patricia Cain

Stone Eugene

1992. Job Satisfaction: How People Feel about Their Jobs and How It Affects Their Performance. New York: Lexington Books.

11.

Edwards

Jeffrey R.

2008. “Person-Environment Fit in Organizations: An Assessment of Theoretical Progress.” Academy of Management Annals 2 (1): 167-230. doi:10.1080/19416520802211503.

12.

Fantasia

Rick

. 1988. Cultures of Solidarity: Consciousness, Action, and Contemporary American Workers. Berkeley: University of California Press.

13.

Fiorito

Jack

Padavic

Irene

Russell

Zachary A.

2014. “Union Beliefs and Activism: A Research Note.” Journal of Labor Research 35 (4): 346-57. doi:10.1007/s12122-014-9186-4.

14.

Freeman

Richard B.

1980. “The Exit-Voice Tradeoff in the Labor Market: Unionism, Job Tenure, Quits, and Separations.” Quarterly Journal of Economics 94 (4): 643-73.

15.

Givan

Rebecca Kolins

Hipp

Lena

. 2012. “Public Perceptions of Union Efficacy A Twenty-Four Country Study.” Labor Studies Journal, 37, 1: 7-32.

16.

Goffee

Robert

. 1981. “Incorporation and Conflict: A Case Study of Subcontracting in the Coal Industry.” Sociological Review 29:475-97.

17.

Gordon

Michael E.

Beauvais

Laura L.

Ladd

Robert T.

1984. “The Job Satisfaction and Union Commitment of Unionized Engineers.” Industrial & Labor Relations Review 37 (3): 359-70. doi:10.2307/2523731.

18.

Hickey

Robert

. 2005. “Running Your Local.” In A Troublemaker’s Handbook 2, edited by Slaughter

Jane

, 290–305. Detroit: Labor Notes.

19.

Hickey

Robert

Kuruvilla

Sarosh

Lakhani

Tashlin

. 2010. “No panacea for success: Member activism, organizing, and union renewal.” British Journal of Industrial Relations. 48:1, pp. 53-83.

20.

Hodson

Randy

. 1991. “The Active Worker: Autonomy and Compliance at the Workplace.” Journal of Contemporary Ethnography 20 (1): 47-78.

21.

Hoell

Robert

. 2004. “How Employee Involvement Affects Union Commitment.” Journal of Labor Research. Spring 25, 2: 267-77.

22.

Holland

John L.

1959. “A Theory of Vocational Choice.” Journal of Counseling Psychology 6 (1): 35-45. doi:10.1037/h0040767.

23.

Holland

John L.

1963. “Explorations of a Theory of Vocational Choice and Achievement: A Four-Year Prediction Study.” Psychological Reports 12 (2): 547-94. doi:10.2466/pr0.1963.12.2.547.

24.

Holland

John L.

1966. The Psychology of Vocational Choice. Waltham: Blaisdell Publishing.

25.

Holland

John L.

1997. Making Vocational Choices: A Theory of Vocational Personalities and Work Environments (3rd ed.). Odessa: Psychological Assessment Resources.

26.

Ironson

G. H.

Smith

P. C.

Brannick

M. T.

Gibson

W. M.

Paul

K. B.

1989. “Construction of a Job in General Scale: A Comparison of Global, Composite, and Specific Measures.” Journal of Applied Psychology 74 (2): 193-200.

27.

Judge

Timonthy A.

Hulin

Charles L.

Dalal

Reeshad S.

2012. Job Satisfaction and Job Affect. In Kozlowski

Steve W. J.

(Ed.), The Oxford Handbook of Organizational Psychology (1st ed., Vol. 1). New York, NY: Oxford University Press.

28.

Kalleberg

Arne L

. Feb. 1977. “Work Values and Job Rewards: A Theory of Job Satisfaction. American Sociological Review.” 42 (1): 124-143.

29.

Kochan

Thomas A.

Helfman

David E.

1981. “The Effects of Collective Bargaining on Economic and Behavioural Job Outcomes.” In Research in Labor Economics, Vol. 4 edited by Ehrenberg

Ronald

, 321-65. Greenwich, Conn.: JAI Press.

30.

Is TFA for You? 2016. July 18. https://www.teachforamerica.org/join-tfa/is-tfa-for-you.

31.

Iverson

Roderick D.

Currivan

Douglas B.

2003. “Union Participation, Job Satisfaction, and Employee Turnover: An Event-History Analysis of the Exit-Voice Hypothesis.” Industrial Relations: A Journal of Economy and Society 42 (1): 101-105. doi:10.1111/1468-232X.00279.

32.

Kristof-Brown

Amy L.

Guay

Russell P.

2011. “Person–Environment Fit.” In APA Handbook of Industrial and Organizational Psychology, edited by Zedeck

Shelden

, 3–50. Washington, DC: American Psychological Association.

33.

Kristof-Brown

Amy L.

Zimmerman

Ryan D.

Johnson

Erin C.

2005. “Consequences of Individuals’ Fit at Work: A Meta-analysis of Person–Job, Person–Organization, Person–Group, and Person–Supervisor Fit.” Personnel Psychology 58 (2): 281-342. doi:10.1111/j.1744-6570.2005.00672.x.

34.

Leicht

Kevin

. 1989. “Unions, Plants, Jobs, and Workers: An Analysis of Union Satisfaction and Participation.” The Sociological Quarterly, 30, 2: 331-362.

35.

Locke

Edwin A

. 1969. “What Is Job Satisfaction?” Organizational Behavior and Human Performance 4 (4): 309–36. https://doi.org/10.1016/0030-5073(69)90013-0.

36.

Martin

James E.

1986. “Predictors of Individual Propensity to Strike.” Industrial & Labor Relations Review 39 (2): 214-27. doi:10.2307/2523460.

37.

McShane

Steven L.

1986. “The Multidimensionality of Union Participation.” Journal of Occupational Psychology 59 (3): 177-87. doi:10.1111/j.2044-8325.1986.tb00223.x.

38.

Monnot

Matthew J.

Wagner

Stephen

Beehr

Terry A.

2011. “A Contingency Model of Union Commitment and Participation: Meta-analysis of the Antecedents of Militant and Nonmilitant Activities.” Journal of Organizational Behavior 32 (8): 1127-46. doi:10.1002/job.736.

39.

Nye

Christopher D.

Rong

Rounds

James

Drasgow

Fritz

. 2012. “Vocational Interests and Performance: A Quantitative Summary of Over 60 Years of Research.” Perspectives on Psychological Science 7 (4): 384-403. doi:10.1177/1745691612449021.

40.

Parks

Judi Mclean

Gallagher

Daniel G.

Fullagar

Clive J. A.

1995. “Operationalizing the Outcomes of Union Commitment: The Dimensionality of Participation.” Journal of Organizational Behavior 16 (7): 533-55.

41.

Peterson

Norman G.

Mumford

Michael D.

Borman

Walter C.

Jeanneret

P. Richard

Fleishman

Edwin A.

, eds. 1999. An Occupational Information System for the 21st Century: The Development of O*NET. Washington, DC: American Psychological Association.

42.

Podsakoff

Philip M.

MacKenzie

Scott B.

Lee

Jeong-Yeon

Podsakoff

Nathan P.

2003. “Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies.” Journal of Applied Psychology 88 (5): 879-903. doi:10.1037/0021-9010.88.5.879.

43.

Prediger

Dale J.

1982. “Dimensions Underlying Holland’s Hexagon: Missing Link between Interests and Occupations?” Journal of Vocational Behavior 21 (3): 259-87. doi:10.1016/0001-8791(82)90036-7.

44.

Reynolds

Megan M.

David

Brady

. March 2012. “Bringing You More than the Weekend: Union Membership and Self-rated Health in the United States.” Social Forces. Volume 90, Issue 3, 1. 1023-1049.

45.

Rounds

James

Rong

Lewis

Phil

Rivkin

David

. 2010. O*NET Interest Profiler Short Form Psychometric Characteristics: Summary. Washington, DC: The National Center for O*NET Development.

46.

Russell

Steven S.

Spitzmüller

Christiane

Lin

Lilly F.

Stanton

Jeffrey M

Smith

Patricia C.

Ironson

Gail H.

2004. “Shorter Can Also Be Better: The Abridged Job in General Scale.” Educational and Psychological Measurement 64 (5): 878-93. doi:10.1177/0013164404264841.

47.

SAS. 2012. (Version 9.4) [Windows]. Cary, NC: SAS Institute Inc.

48.

Spector

Paul E.

1997. Job Satisfaction: Application, Assessment, Causes, and Consequences. Thousand Oaks, CA: SAGE Publications.

49.

Spinrad

William

. April 1960. “Correlates of Trade Union Participation: A Summary of the Literature.” American Sociological Review. Vol. 25: 237-244.

50.

Rong

. 2012. “The Power of Vocational Interests and Interest Congruence in Predicting Career Success.” PhD diss., University of Illinois Urbana–Champaign. http://hdl.handle.net/2142/34329.

51.

Taylor

Verta

Whittier

Nancy

. 1992. “Collective Identity in Social Movement Communities: Lesbian Feminist Mobilization.” In Frontiers in Social Movement Theory, edited by Morris

Aldon

McClurg

Carol

, 104-29. New Haven: Yale University Press.

52.

Tracey

Terence J. G.

Allen

Jeff

Robbins

Steven B.

2012. “Moderation of the Relation between Person–Environment Congruence and Academic Success: Environmental Constraint, Personal Flexibility and Method.” Journal of Vocational Behavior 80 (1): 38-49. doi:10.1016/j.jvb.2011.03.005.

53.

Tracey

Terence J. G.

Rounds

James B.

1993. “Evaluating Holland’s and Gati’s Vocational-Interest Models: A Structural Meta-analysis.” Psychological Bulletin 113 (2): 229-46. doi:10.1037/0033-2909.113.2.229.

54.

Zener

Thelma B.

Schnuelle

Leslie

. 1976. “Effects of the Self-directed Search on High School Students.” Journal of Counseling Psychology 23 (4): 353-59. doi:10.1037/0022-0167.23.4.353.