Abstract
Given the beneficial effects of labor unions on “bread and butter issues,” union members should appraise their economic circumstances more favorably than nonunion members do. Yet, research on the anomaly of the dissatisfied union worker challenges this expectation. Using the General Social Survey’s Quality of Work Life module, this article examines whether union members appraise their economic circumstances less favorably than nonunion members. Results suggest that union membership is associated with more favorable appraisal of benefits but not wages. These findings may help to provide at least a partial explanation for the anomaly of the dissatisfied union worker. Because income constitutes a vital component of job satisfaction, this study contributes more broadly to our understanding of employee well-being. It also connects the study of occupational health to that on institutions, contributes to the growing literature on the political economy of health, and highlights latent effects of union security agreements.
Introduction
Job satisfaction is a major determinant of occupational health. A 2005 meta-analysis suggests that although the strength of associations vary, job satisfaction is linked with myriad dimensions of both physical and mental health. 1 A key element of job satisfaction is how employees evaluate their compensation. 2 Curiously, although there is a voluminous literature demonstrating that labor unions are associated with better pay and benefits,3–12 there is very little research examining how labor union membership relates to workers’ feelings about their pay and benefits. This gap in knowledge about the subjective dimensions of economic outcomes may arise from an assumption that subjective and objective economic outcomes will be the same. This assumption, however, is called into question by the widely observed finding that union members are less satisfied with their jobs than are nonunion workers.
One potential explanation for this discrepancy is how workers feel about their economic circumstances. This article examines whether union members, despite advantages in both areas,3,4,6–12 appraise their economic circumstances less favorably than nonunion members appraise theirs. Motivated by the need for explicit examination of how compensation appraisals relate to union membership, the Background section begins with an overview of the literature on labor unions and objective outcomes. The study then considers the potential causes of the gap in knowledge about labor unions and subjective outcomes. The Methods section details the research design and data source, the measures, and the statistical analyses. The Results section describes the sample characteristics, bivariate relationships, and the link between labor unions and compensation appraisals, with consideration to covariates. The Discussion section begins by summarizing the key findings and addressing limitations. The article concludes by situating the findings in relation to the connection between occupational health and institutions, the growing literature on the political economy of well-being, and labor market policy around collective bargaining.
Background
Although an immense amount is known about how union members fare on pay and benefits, very little is known about how union members feel about pay and benefits. Empirical inattention to union effects on subjective outcomes may arise in part because of the strength of the research on objective outcomes. Labor unions significantly increase the wages of their members across occupations.3,5–7,12 This is especially true in low-skilled occupations. 5 In 2011, the wage premium among private-sector union workers approached 30%. Rates were comparable at the local (22%) and state (36%) levels. 8 Labor unions are also associated with substantially better fringe benefits. 11 Especially in small firms, union members are more likely to have health insurance.4,9 Union members’ insurance is typically more affordable and more comprehensive.4,8 Furthermore, although the research is more mixed, labor unions also are associated with higher availability and generosity of retirement pensions.8,10
But the assumption that subjective outcomes align with objective outcomes is dubious. Knowing that union workers do well compared with nonunion workers does not answer the question of whether they feel that they do well. As job-satisfaction scholars have noted, “researchers working from an objective perspective . . . tend to be wary of measures of job satisfaction because they observe that workers often express satisfaction with jobs that are of objectively poor quality.” 13
This discrepancy is no more obvious than in research on the topic of labor unions and overall job satisfaction. In their seminal work on what unions do, Freeman and Medoff 14 observe that union workers expressed greater job dissatisfaction than nonunion workers. This dissatisfaction existed not only despite union workers’ having better employment conditions than nonunion workers ceteris paribus (with other conditions remaining the same) but also despite their expressing less willingness to change jobs. 15 Many studies have replicated this curious finding.16,17 The opposing expectations generated by labor economics research on union effects and union job satisfaction may be adjudicated in part by an explicit examination of union members’ appraisals of their pay and benefits. In other words, job satisfaction may suffer if union members do not regard their pay and benefits as good, regardless of whether their pay and benefits are better than their nonunionized counterparts. Arising from the literature around the dissatisfied union worker, we hypothesize that union members will be equally or less likely to appraise their economic situation favorably than nonunion members, all else equal.
Methods
Research Design and Data Source
This is an observational, cross-sectional individual-level study. It uses the National Opinion Research Center’s General Social Survey (GSS) for all analysis. Data about work environment and compensation appraisals are from the 2002, 2006, and 2010 Quality of Work Life modules. (Data are also available for 2014 but employ a new occupational coding schema that began after 2010.) This module was developed by the National Institute of Occupational Safety and Health. It aims to gather information on working conditions—especially on work organization variables, stressors, and a number of stress and health outcomes.
Measures
The GSS provides information on whether the respondent, the respondent’s spouse, or both respondent and spouse claim union membership. For the most straightforward analyses, the respondent’s union status has been chosen as the key independent variable (reference = nonunion member). To measure compensation appraisal, variables are used that tap both attitudes toward earnings and attitudes toward benefits. All measures are dichotomized and coded so that higher values represent more favorable endorsements. The earnings measure asks “How fair is what you earn on your job in comparison to others doing the same type of work you do?” Response options include much less than you deserve (coded “0”), somewhat less than you deserve (coded “0”), about as much as you deserve (coded “1”), somewhat more than you deserve (coded “1”), and much more than you deserve (coded “1”)? Fringe benefits are assessed using agreement with the statement “My fringe benefits are good.” Response options include not at all true (coded “0”), not too true (coded “0”), somewhat true (coded “1”), and very true (coded “1”).
A myriad of controls including and predicting objective economic outcomes have been included. Equivalized income adjusts for both inflation and family size. Years of education have been transformed into four levels: less than high school, high-school graduate, some college, and bachelor's degree or more (reference = high-school graduate). An ordinal measure of education has been chosen based on prior research suggesting that differences in income are largest at these points in the educational distribution. 18
A variety of potential confounders between union membership and aspects of job satisfaction have also been included. Past research has demonstrated an association between specific employment situations and job quality.19,20 Accordingly, work status consists of categories for full-time, part-time, and temporarily not working (reference = full-time). Additionally, self-employment status has been included under the premise that job satisfaction varies between those who work for others and those who work for themselves. 21 Many facets of job satisfaction vary by job characteristics22–24; thus, dummy variables have been constructed for both occupation and industry. Sensitivity analysis includes an indicator for survey year. Size and statistical significance of all coefficients remain the same with controls for survey year.
Finally, the present study includes a series of demographic covariates linked to economic circumstances. A dummy variable is included for female (reference = male) in all models, based on a preponderance of research on gender differences in job satisfaction. This research suggests that women have lower job quality but higher job satisfaction and that the predictors of job satisfaction differ between the sexes.25–27 Age as well as age-squared have also been included in all models, in recognition of the U-shaped pattern in job satisfaction over the life course. 22 Due to the differing attitudes toward pay across marital statuses, categories of married, divorced, widowed, separated, and never married (reference = married) have been included. 28 Since studies have shown widespread differences in employment satisfaction based on race, race categories of white, black, and other (reference = white) have been delineated. 29 Finally, geographic residence has been recoded from the nine U.S. census divisions into the four census regions of Northeast, Midwest, South, and West (reference = Midwest). This is because unionization rates, and potentially efficacy, vary across these areas. 30
Statistical Analyses
The analytic methods used in this analysis assess the contribution of union membership to compensation appraisals accounting for all potential sources of spuriousness and/or suppression. Table 1 provides means and standard deviations for all variables, excepting occupation and industry.
Sample Characteristics in General Social Survey, 2002, 2006, and 2010.
Note: N for all variables = 2534.
aEquivalized income is calculated as household income divided by the square root of the number of household members.
Table 2 shows the results of cross tabulations of union membership with each measure of compensation appraisal. This table assesses whether union membership is associated with the outcomes (compensation appraisal of earnings in Panel A and of benefits in Panel B). Columns 1 and 2 of Table 3 display the coefficients from a multivariate logistic regression estimating the association between compensation appraisal and union membership status, adjusting for relevant demographic and employment-related covariates. Log odds are exponentiated and presented as odds ratios to ease interpretation. These models test whether union membership is associated with compensation appraisals when accounting for all potential confounders.
Cross-Tabulation of Compensation Appraisals by Union Membership, General Social Survey 2002, 2006, and 2010.
Note. All analyses involve two-tailed tests of significance.
aPearson chi-square = 0.004. p = .947.
bPearson chi-square = 36.311. p = .000.
Odds Ratios From Logistic Regression of Compensation Appraisals on Union Membership and Relevant Controls, General Social Survey 2002, 2006, and 2010.
Notes: Robust z-statistics in parentheses. References = Married, White, Midwest, High School, and Full-time Employment. All analyses involve two-tailed tests of significance.
***p < .01, **p < .05, *p < .1.
Both outcome variables contained about 50% missing values. This owed primarily to the fact that these questions were asked only of working respondents, but also to changes in the GSS sampling frame in 2004, which affected the sampling universe for the 2006 National Institute of Occupational Safety and Health survey items. Just under 30% of respondents did not contribute data on union membership status due to the split-ballot design implemented by the GSS. Missingness was 5.54 and 6.36, respectively, for occupation and industry and below 1% for all remaining variables. We employed listwise deletion to address missingness. All analyses were conducted in Stata 14.
Results
Sample Characteristics
Well over half of the sample (57%) believe that their earnings are fair. Almost three-quarters (71%) believe that their fringe benefits are good. About 12% of the sample are members of a labor union. This is in line with estimates of union density during the range of survey years. Descriptive statistics generally mirror those of the U.S. population at large. The sample is 51% female, 49% married, 78% white, and 12% self-employed. A majority of respondents (65%) have begun or have completed postsecondary schooling. The average age is forty-two. Average income, equivalized for family size, is roughly US$26,500 ($26,521).
Bivariate Associations
First, the study examines the distribution of favorable compensation appraisals across union and nonunion members (Table 2). Union members are significantly more likely to feel that their benefits are good (Panel B: 86% vs. 69%; p =.000), but they are no more likely to feel that their earnings are fair. In fact, the proportion of union and nonunion members who regard their earnings as fair is almost identical (Panel A: 56.8% vs. 57%; p = .947).
Multivariate Regression
Table 3 displays a multivariate logistic regression estimating odds ratios for each indicator of compensation appraisal. All estimates adjust for relevant demographic and employment-related covariates. These models test whether union membership is associated with favorable compensation appraisals once accounting for potential confounders. Coefficients above zero suggest a higher likelihood of positive appraisal relative to the reference group (nonunion members); coefficients below zero suggest a lower likelihood of positive appraisal. The p values associated with the chi-square test statistics from likelihood ratio tests indicate that both models (earnings and benefits) are significant overall and that, compared with unconditional models, the fully adjusted models fit significantly better.
Column 1 of Table 3 evaluates whether and how union membership is significantly associated with the appraisal of fair earnings. Union members are no more likely than nonunion members to appraise their earnings as fair. This echoes the findings at the bivariate level in Table 2. Column 2 examines the extent and direction of an association between union membership and the appraisal that fringe benefits are good. Unlike in the case of earnings, union membership increases the odds that an individual worker will perceive benefits as good. The odds ratio is, in fact, quite large (2.66; p < .01).
Discussion
At the bivariate level, union members are more likely to have favorable appraisals of benefits but not earnings. These associations persist in logistic regression models accounting for the full set of potential labor market and demographic confounders. These results lend partial support for the study's hypothesis. This hypothesis, which proposes that union members will be less likely to appraise their economic situation favorably than nonunion members despite increases to income and benefits, is supported by the earnings findings, where there is no apparent advantage for union members. The hypothesis gathers less support from the benefits finding, where union members are more likely than nonunion members to have favorable appraisal of benefits.
The findings differ somewhat from the little work that examines attitudes toward pay. White and Bryson 31 find “a substantial degree of correspondence between job satisfaction attitudes and real economic outcomes with respect to pay, hours and security.” The difference in operationalization may be important here. White and Bryson focus on a traditional measure of satisfaction with earnings, whereas the current study uses an item tapping the respondent’s appraisal that his or her earnings are fair. It is possible, of course, to be satisfied with earnings that one does not judge fair. More likely, however, is the possibility that the authors focus on union coverage rather than membership, as does the present study. It is the case that individuals covered by, but not a member of, a labor union tend to have greater job satisfaction. 32 This can be understood in the context of the free-rider problem, where covered members enjoy most of the benefits of union membership but few of the costs.
The absence of literature on employee appraisals of benefits does not permit useful comparison to the current study. This is unfortunate, as it could shed some light on the different findings between pay and benefits. Whereas union members are not more likely to appraise their earnings as fair in unadjusted regressions, they are more likely to appraise their benefits as good. What might account for these divergent findings? It could be that the union benefit premium is generally larger than the union wage premium.
It is also possible that under current labor market conditions, workers view good benefits as a scarcer resource than fair pay. Employer-provided health-insurance rates have decreased during the period of observation. Between 1999 and 2014, offer rates decreased five percentage points, from 71% to 66%. 33 When available, it is more expensive. The average employee contribution to health insurance increased from $318 in 1999 to $1081 in 2014. 33 Beginning in 2008, out-of-pocket costs, such as deductibles and co-pays, began to shift as well. 33 Employer-sponsored pension plans have followed a similar trend. 34 The defined-benefit plans that predominated up through the 1980s have given way to defined-contribution plans, shifting responsibility to individual employees and introducing much greater risk. 35 With benefits of any sort—especially those of high quality—having become increasingly decoupled from employment (even full-time employment), workers may appraise any solid benefit package as good.
Although this study has many strengths (nationally representative data of the United States, survey data spanning over a decade, and survey designed to understand the distribution and correlates of work-related variables), it is not without limitations. The most pressing of these constraints relates to the fact that we cannot know from these data how union members feel relative to nonunion members in the context of union support. Union members may not believe their wages are fair, yet still regard them as better (or less unfair) than they might otherwise be. Future research could use more fine-grained measures of compensation appraisals not only to replicate these findings but also to more fully explore how labor unions relate to different subjective economic outcomes.
This study makes three primary empirical and theoretical contributions. First, it may help to explain the anomaly of dissatisfied union workers. This has historically been attributed to the characteristics of the workers and the workplaces that tend to unionize.15,22 Others have described it as a result of organized labor's ability to provide a voice for union members to express complaints (Dowding et al., 2000; Freeman, 1980; Hirschman, 1970).36–38 This study cannot rule out these possibilities. But it does provide some clues that evaluations about pay, whatever their association with worker/workplace characteristics and collective voice, are likely implicated in this curious anomaly. Moreover, because income constitutes a vital component of job satisfaction among unionized and nonunionized workers alike, this study contributes more broadly to our understanding of employee well-being in underscoring the independence of subjective and objective evaluations of pay.
Second, it incorporates research on institutions into broader conversations about individuals’ subjective labor market experiences. Many studies of labor unions have been relegated to the fields of labor/labor movements, where studies have tended to center on a specific workplace, occupation, or industry. 39 These inquiries present unions as an organizational feature rather than as a sociological phenomenon, both in its origins and its consequences. In examining union membership among a nationally representative sample spread across many levels of the labor market, the analysis here treats labor unions as an institution and asks how it relates to vital components of people’s lived experiences. This responds to the criticism lodged by Meardi et al. 40 that as social sciences expanded during the 1970s and 1980s, research on the macro-level correlates of work had diverged significantly, and regrettably, from that on the organization of work.
Second, this investigation adds to the growing literature on the political economy of well-being. Scholars of health have long emphasized the importance of stratification for health and health inequalities.41–43 Scholars of politics have offered rich insights into how macro-level political factors affect stratification processes around income inequality and poverty.44–49 A recent proliferation of research has integrated these two approaches. This area of research highlights the way in which power relations and institutions relate to health and health inequalities,50–59 life satisfaction, 60 and happiness 61 among workers and the population at large. This work underscores the importance of macro-level factors for well-being.
In terms of policy contributions, these findings suggest that labor policies that dampen union activity will jeopardize a major source of support for workers. This is especially meaningful in the United States context, where healthcare is both not universally subsidized by the government and considerably more expensive than elsewhere. 62 Moreover, given the links between compensation, job satisfaction, and health, 1 further labor union contraction could compromise the physical and, especially, mental well-being of the U.S. workforce. Thus, the implementation of state right-to-work laws, which prohibit labor unions from securing union dues from all employees in a unionized workplace regardless of membership, may generate unanticipated negative externalities for the labor market.
Footnotes
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.