Eastern Economic Journal, 2012, 38, (331–355) r 2012 EEA 0094-5056/12 www.palgrave-journals.com/eej/
Women, Men, and Job Satisfaction Cheryl J. Carleton and Suzanne Heller Clain Department of Economics, Villanova School of Business, Villanova University, 800 Lancaster Avenue, Villanova, PA 19085, USA.
We examine job satisfaction to determine whether gender differences found by previous researchers could be explained by constraints imposed by the specifications used. Applying those specifications to recent US data yields results similar to those previously found. However, clarification comes from applying specifications that allow for gender ^0 s) of personal/job differences in sample selectivity and in the relative weights (b characteristics in evaluating satisfaction. We find that gender differences in the job ^0 s. satisfaction of married workers can largely be attributed to gender differences in b However, more work is necessary to understand gender differences in job satisfaction among unmarried workers. Eastern Economic Journal (2012) 38, 331–355. doi:10.1057/eej.2011.17; published online 15 August 2011 Keywords: job satisfaction; gender differences JEL: J16; J28
INTRODUCTION Job satisfaction has been a subject of interest to labor economists for a good number of years. One strand of research has focused on the significant impact of job satisfaction on important aspects of worker performance, such as absenteeism [Vroom 1964; Mangione and Quinn 1975; Clegg 1983] and turnover [Flanagan et al. 1974; Freeman 1978; Akerlof et al. 1988; Clark 2001]. Another strand of research has in turn investigated the determinants of job satisfaction itself. Early studies among the latter [Borjas 1979; Bartel 1981; Hamermesh 2001] investigated the determinants of job satisfaction for men only. More recent studies have included women in the analysis [Clark 1997; Sloane and Williams 2000; SousaPoza and Sousa-Poza 2000; Donohue and Heywood 2004; Bender et al. 2005; Kristensen and Johansson 2008]. These have generally found indications in European Union, British and US data that job satisfaction is greater for women than for men. The finding has been seen as somewhat paradoxical, given the longstanding gender earnings gap that favors men. In their efforts to explain the source of the paradox, these researchers have considered many possibilities. A common approach to the issue has been to search for some previously overlooked factor that, once included in the analysis, causes the gender difference to disappear. Notable among the explanations is one by Clark [1997], suggesting that women’s higher job satisfaction is caused by women’s improved position in the labor force relative to their expectations. Clark came to this conclusion after first supposing the gender difference in expectations to be smaller for younger workers and for better-educated workers. In his empirical work, he found no gender differential in job satisfaction among the youngest workers (16–19) or the most educated workers in his sample, but differentials for other ages and levels of education. On the basis of this work, Clark was led to predict the gender difference in job satisfaction to be temporary, lasting only until expectations adjust so that there is no gender difference in rewards relative to expectations.
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
332
Donohue and Heywood [2004] similarly found that no gender gap in job satisfaction existed for a young (23–31) cohort of US workers, surveyed in 1988. Taking this to be evidence of a cohort effect, Donohue and Heywood hypothesized that the gender differences with respect to job satisfaction in the population as a whole might vanish as the young cohort ages. However, Donohue and Heywood noted that their findings were also consistent with a vintage effect, whereby the gender gap in job satisfaction remains a fact of life, regularly observable once workers age out of the young cohort. The present study re-examines the job satisfaction of men and women, using recent US data, with a primary focus on estimating job satisfaction separately by gender and marital status. Previous researchers tended to investigate gender differences in job satisfaction by pooling men and women together, regardless of marital status, and estimating a model with a dummy variable for gender. Such a specification could obscure the source of gender differences, if in fact tastes (i.e. slope coefficients) vary by gender. A secondary focus of the analysis is sample selectivity, an issue that Clark [1997] considered but discarded in his search for sources of gender differences in job satisfaction. Given research in the area of labor supply, it is reasonable to think that women (particularly married women) may have greater discretion than men (and probably unmarried women) in the decision to work for pay. Consequently, the decision of married women to work in the market may itself indicate a taste for market work and an expectation of a high level of job satisfaction. However, the decision of men to work in the market may reflect social conditioning and traditional gender roles more so than anything in particular about a taste for market work or an expectation of a high level of job satisfaction. It may be that men (and unmarried women) who are dissatisfied with work nevertheless tend to continue to work at their unsatisfying jobs, while married women who are unhappy with market work tend to quit unsatisfying jobs in order to pursue something more satisfying.1 If so, then these behaviors affect the underlying nature of the observed sample of job holders used to explore job satisfaction. As such, the differences in labor supply behaviors could be responsible for the appearance of greater job satisfaction among married working women, compared to working men (or unmarried working women). The observed samples of job holders with data reporting associated job satisfaction may be predestined to show gender differences in job satisfaction, based on the underlying gender differences in the nature of the decision to work for pay. In this paper, we first apply conventional specifications to all workers and then to subsamples separated by gender or marital status. Initial results duplicate the findings of previous researchers. However, results of estimation applied to various subsamples (males, females, marrieds, and unmarrieds) suggest a more complicated picture. The gender difference in job satisfaction is found to be less significant among unmarried workers than among married workers. Also, while job satisfaction of males is not significantly different by marital status, the job satisfaction of married females is significantly higher than the job satisfaction of unmarried females. In an effort to investigate the possible role of variation in taste, we apply the analysis to subsamples separated by gender and marital status (married men, married women, unmarried men, and unmarried women). Applying the analysis to such subsamples diverges from conventional specifications of job satisfaction, in that it allows for gender effects beyond intercept differences. We also expand the analysis to include techniques designed to correct for sample selection. In doing so, Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
333
we find that sample selectivity related to the decision to work appears to have a role in explaining the job satisfaction of married women but not to have a role in explaining the job satisfaction of married men, unmarried men, and unmarried women. Given that there are differences in sample selectivity, personal/job characteristics ^0 s or (X’s) and the impacts of personal/job characteristics on job satisfaction (b tastes) across subsamples, a relevant question concerns the extent to which each of these contributes to conventional findings of gender differences in job satisfaction. We address this question by re-estimating job satisfaction by marital status (marrieds, unmarrieds), this time using a full set of interaction terms of gender and X’s to capture gender differences. This analysis reaffirms that sample selectivity is a relevant factor for married but not unmarried workers, but suggests no significant difference in its role for married women and married men. It does pinpoint other ^0 s that differ significantly by gender. specific personal/job characteristics that have b In doing so, it also suggests that conventional findings can be fully attributed ^0 s of married workers; with these specifications there to gender differences in the b is no significant remaining gender difference in the intercept of the job satisfaction equation of married workers. However, a significant gender difference in the intercept of the job satisfaction equation of unmarried workers is found; it contributes an additional upward boost in the job satisfaction of male unmarried workers. This finding suggests that more work is needed to understand gender differences in the job satisfaction of unmarried workers. A brief discussion of the theoretical model and the methodology behind the analysis follows in the next section. The third and fourth sections describe the data and the analytical findings, respectively.
MODEL AND METHODOLOGY We begin by looking at frequency distributions for the reported job satisfaction of individuals working for pay, by gender and marital status. We perform tests for significant differences across these dimensions. It must be noted that many other factors that could influence job satisfaction are not being considered or held constant in any of these comparisons. The gender patterns observed when these other factors are not being held constant may be considerably different from those found when these other factors are held constant. The writings of previous researchers provide a basis for modeling the determinants of job satisfaction. Following the work of others before them, both Clark [1997] and Sloane and Williams [2000] relate job satisfaction to the familiar economic concept of utility.2 In so doing, they suggest job satisfaction (U) to be generally influenced by wage or income (Y), hours of work (H), personal characteristics (Xi), and job-specific characteristics (Xj). An extension of the basic model specification commonly has involved using both absolute and relative (or comparative) measures of wage or income (Y and Y*, respectively). That is, U ¼ uðXÞ ¼ uðY; Y ; H; Xi ; Xj Þ The basic set of personal characteristics (Xi) used by these authors includes age, gender, race, education, marital status, and home ownership. Variables reflecting health, job tenure, union membership, and work values are also used in some Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
334
specifications. The basic set of job-specific characteristics (Xj) used by these authors includes the opportunity for promotion, managerial or supervisory nature of job, permanency of job, and establishment size. Some specifications also used variables measuring the nature of employer-employee relations, gender mix at the worksite, employer expectation of work outside of normal hours, effect of co-workers on the nature of the job, and availability of alternative jobs, along with other controls for industry and/or occupation. The model employed here to explain variation in job satisfaction follows this general framework. The personal and job-specific characteristics included as determinants of job satisfaction largely reflect those used by other researchers. Specifically, age, gender, race, education, marital status, health, and job tenure are incorporated as key personal characteristics (Xi) influencing job satisfaction in this analysis. In addition to measures of hours of work and both absolute and comparative earnings, key job-specific characteristics (Xj) used in this analysis are general occupational category, general industry category, flexibility of work scheduling, perceptions of management and promotion practices, opportunities for input and development, and level of work-related stress. Because of the ordinal qualitative nature of the dependent variable in this analysis of job satisfaction, an ordered-probit technique is used to estimate the model, as is customary among such studies [Clark and Oswald 1996; Clark 1997; Sloane and Williams 2000; Sousa-Poza and Sousa-Poza 2000; Donohue and Heywood 2004; Bender et al. 2005]. That is, S ¼ bX þ e where S* is unobserved satisfaction and e is a stochastic disturbance term with a standard normal distribution. The reported level of job satisfaction (S) is related to unobserved satisfaction in the following manner: S ¼1 S ¼2
if S p0 if 0oS pm1
S ¼3
if m1 oS pm2
S ¼4
ifm2 oS
Estimates of the job satisfaction equation are generated for the full sample and for subsamples defined by gender or marital status (males, females, marrieds, and unmarrieds). Results are inspected for evidence of differences as captured by dummy variables for gender or marital status. The estimation is repeated for subsamples defined by gender and marital status (married males, married females, unmarried males, and unmarried females), to allow for effects beyond intercept differences. The ordered-probit technique is then modified to include a first-stage probit, in order to investigate possible sample selection bias associated with individuals’ decisions to work for pay and the restriction of the sample to job holders in the ordered-probit estimation.3 Individuals are job holders (Z ¼ 1) if the utility of working exceeds the utility of remaining unemployed or out of the labor force. The difference in the utilities (Z*) is unobservable, but assumed to be a function of personal characteristics (W). That is, Z ¼ aW þ u where W commonly includes age, gender, race, education, health, marital status, household composition, and unearned income4 and u is a stochastic error term with Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
335
a standard normal distribution. An individual is a job holder and included in the sample of jobholders if Z*>0.5 The error terms of the first and second stages of estimation (e and u) are assumed to be independent. The results of the first-stage probit provide the basis for imputing an additional independent variable (an inverse Mill’s ratio or lambda) for the second-stage ordered probit.6 Specifically, lambda (l) is calculated as jð^ aWÞ Fð^ aWÞ where j is the standard normal probability density function and F is the standard normal cumulative distribution function. If significant, this independent variable provides a necessary adjustment for the sample selectivity inherent in basing the second-stage estimation on a sample that inherently excludes non-workers, due to their lack of job and therefore unreported measure of job satisfaction. These efforts extend the work of Clark [1997], who considered the possibility of such sample selection bias, but found no significant evidence of it in his analysis. In his analysis, Clark applied a sample-selection correction in the context of a second-stage OLS-estimated job-satisfaction equation, but continued nevertheless to find a significant gender differential in job satisfaction. That is, despite the correction, he continued to find a significant coefficient on a gender dummy variable, in his estimate of an equation explaining job satisfaction for a sample of all workers. It must be noted, however, that Clark’s findings concerning sample selectivity are potentially flawed, inasmuch as his specification imposed a constraint reflecting an implicit assumption that sample selection, if it existed, would be operationally the same in its effects for all workers, male or female, married or not. That is, the functional form of the equation that Clark used in the second-stage of the estimation forced the estimated effect of the independent variable representing sample selectivity (the inverse Mill’s ratio or lambda) to be the same numeric value, regardless of gender or marital status, inasmuch as the sample consisted of all workers (male and female, married and unmarried). The assumption of a universal impact of sample selectivity runs counter to the conventional assumptions surrounding the effects of sample selection on estimation of wage equations, where findings of sample selection bias and the need to correct for such bias have been largely limited to samples of women (particularly married women). Here, we apply sample-selection correction in a manner that allows its effect to vary by gender and marital status. Specifically, we investigate the role of sample selection within four distinct samples of individuals classified by gender and marital status, to allow for differences in the effects of sample selection on the job satisfaction of different demographic groups. We then expand these estimation results (both with and without the sample ^0 s are the selection correction) to explore the extent to which gender differences in b source of previous findings of gender differences in the job satisfaction of workers. Specifically, we re-estimate job satisfaction by marital status (married, unmarried), this time using interaction terms of gender and X’s to capture gender differences. ^0 s are This analysis allows us to gauge the extent to which gender differences in b significant and the extent to which there is any remaining gender difference ^0 s or X’s. unassociated with b The data used in this investigation are described in the following section. Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
336
DATA This study utilizes data from the 2002 and 2006 versions of the General Social Survey (GSS) [Davis and Smith 2009]. These cross-sectional samples of data are much more recent than those used by other studies to examine gender differences in job satisfaction.7 The GSS is designed to collect data on demographic characteristics and attitudes of residents of the United States. It is claimed to be the most frequently analyzed source of information about American society, outside of the US Census.8 As the longest running project of the National Opinion Research Center, it was conducted almost annually from 1972 to 1994 (except in 1979, 1981, and 1992); since 1994 it has been conducted in even-numbered years. Historically, cross-sectional samples have been independently drawn each year.9 In 2008, however, the GSS began a transition toward a design that uses rotating panels. GSS sampling is designed to give each US household an equal probability of inclusion in the sample. Within a household, each eligible individual has an equal probability of being interviewed. However, one and only one eligible individual within a household is interviewed. The result is that persons living in large households are less likely to be interviewed. Prior to 2004, the GSS provides sampling weights that adjust for the selection of one adult per household. Beginning in 2004, the provided sampling weights adjust for both the selection of one adult per household and the sub-sampling of non-respondents. Survey response rates in 2002 and 2006 were 70.4 and 71.2 percent, respectively. A set of core questions are asked of GSS respondents every year, while topical modules of questions are included on a rotational basis.10 A topical module called “Quality of Working Life” was included in the 2002 and 2006 versions of the GSS; it asked several questions about job satisfaction and job aspects which might affect job satisfaction. The questions in this module were aimed at workers in general: those currently at work full-time or part-time, those with a job but not currently at work, those employed by others and those self-employed. In 2002, all survey questionnaires contained the Quality of Working Life module. In 2006, however, only a random subset of 75 percent of survey questionnaires contained the Quality of Working Life module. One particularly attractive feature of this module is that it asks respondents: “How fair is what you earn on your job in comparison to others doing the same type of work you do?” Given the findings of Sloane and Williams [2000], a measure of comparative earnings based on individual perception is expected to be more directly relevant to an individual’s job satisfaction than predicted earnings based on supplementary earnings equations, as formulated in Clark and Oswald [1996], Clark [1997], and Donohue and Heywood [2004].11 This GSS module also asks numerous questions about the flexibility of work scheduling. Bender et al. [2005] found job flexibility to be useful in explaining gender differences in job satisfaction that had previously been attributed to gender composition of the work place [Sloane and Williams 2000]. One of the weaknesses of the GSS in general is its failure to collect precise information about earnings. The information provided for income earned by the individual is strictly categorical and somewhat skewed in its design (i.e. choice of categories).12 That is, while it gives reasonably good detail about the lower end of the income distribution, detail about the mid- and upper portions of the income distribution are lacking. Given the incompleteness of the information and harsh criticisms of the practice of using midpoints of intervals as measurements of income Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
337 Table 1 Definitions of variables Job satisfaction Age Male White Married Health Education Tenure White collara Blue collara Manufacturingb Tradeb Serviceb Governmentb Full-time Ln hours Second job Low earnings Earningsofair Earnings>fair Fringes OK Fair promotions Caring supervisor Building abilities Having input Taking time off Not interfering Discrimination Mandatory OT High stress Changing schedule Lambda D06
=4 if very satisfied; =3 if somewhat satisfied; =2 if not too satisfied; =1 if not at all satisfied age in years =1 if male; =0 if female =1 if white; =0 if otherwise =1 if currently married; =0 if separated, widowed, divorced, or never married =1 if health excellent or good; =0 if health fair or poor =1 if has a BA or higher degree; =0 if otherwise years on the current job =1 if white collar occupation; =0 if otherwise =1 if blue collar occupation; =0 if otherwise =1 if manufacturing industry; =0 if otherwise =1 if wholesale or retail trade industry; =0 if otherwise =1 if service industry; =0 if otherwise =1 if public administration industry; =0 if otherwise =1 if full-time worker; =0 if part-time worker natural logarithm of hours worked at all jobs in week prior to survey =1 if has any other job besides main job; =0 if otherwise =1 if annual earnings from job are less than 15K; =0 if annual earnings from job are 15K or more =1 if individual believes earnings on job to be much less or somewhat less than deserves; =0 if otherwise =1 if individual believes earnings on job to be somewhat more or much more than deserves; =0 if otherwise =1 if believes “My fringe benefits are good” to be very true or somewhat true; =0 if otherwise =1 if believes “Promotions are handled fairly” to be very true or somewhat true; =0 if otherwise =1 if believes “My supervisor cares about me” to be very true; =0 if otherwise =1 if believes “I have an opportunity to develop my own special abilities” to be very true; =0 if otherwise =1 if agrees or strongly agrees that “I have a lot of say about what happens on my job”; =0 if otherwise =1 if taking time off from work for personal or family matters is not hard or not too hard; =0 if it’s somewhat or very hard =1 if demands of job rarely or never interfere with family life; =0 if they often or sometimes do =1 if feels discriminated against on the job because of gender, race, or age; =0 if otherwise =1 if working extra hours on the job is mandatory or required by employer; =0 if otherwise =1 if work is always or often stressful; =0 if otherwise =1 if often or sometimes allowed to change starting and quitting time on a daily basis; =0 if rarely or never allowed sample selectivity variable, estimated from probit equation results reported in Appendix Table A3 =1 if 2006; =0 if 2002
a
The omitted category for type of occupation is service. The omitted category for type of industry includes agriculture, forestry and fisheries; mining; construction; and transportation, communications and other public utilities. b
[Berg and Lien 2002], we simply use a dummy variable to indicate an income level in the lower tail of this distribution. The samples for the first-stage probit consist of individuals age 70 and younger who have provided the necessary information for variables used in a basic Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
338
employment equation. As previously mentioned, these variables include gender, age, race, marital status, education, health status, number of children in the household (by age group), work status, and education and work status of spouse, if married.13 The sub-samples used in the second-stage ordered probit are restricted to full-time or part-time employees who have supplied usable answers to the job-related questions used in the analysis. Owing to the rotation design used by the GSS in 2006, some workers that can be included in the first-stage probit analysis unfortunately do not have the working-life data required by the second-stage ordered probit analysis.14 As previously mentioned, the second-stage ordered-probit estimation attempts to control for many of the same variables used in other job-satisfaction studies, including age, gender, race, education, marital status, health, job tenure, type of occupation, type of industry, earnings, relative earnings, and various aspects of the job.15 Many of the questions, including how satisfied one is with one’s current job, have four or more categorical responses. Owing to small numbers in some cells and for ease of analysis, the responses were collapsed into two categories for a number of questions.16 Data for 2002 and 2006 are pooled in the analysis for similar reasons; to allow for distinct year effects, however, a dummy variable indicating year is included throughout the estimation. Table 1 provides precise definitions of the variables derived from these data and used in the ordered-probit analysis of job satisfaction. The omitted category for type of occupation is service; the omitted category for type of industry includes agriculture, forestry and fisheries; mining; construction; and transportation, communications and other public utilities.
ANALYSIS Tables 2 and 3 show summary statistics for variables used in this study. In Table 2, the distributions of job satisfaction are presented by marital status and gender. Chisquare tests are performed to test if job-satisfaction distributions are independent of gender (for the full sample, for the married subsample, and for the unmarried subsample) or independent of marital status (for the male subsample and for the female subsample). The results of these tests indicate that there are significant differences in job satisfaction between men and women in the full sample (“All” in Table 2, Chi-square ¼ 6.888, Prob. ¼ 0.076) and between married women and unmarried women in the female subsample (Chi-square ¼ 18.516, Prob. ¼ 0.000). On the other hand, there is NO significant difference in job satisfaction between married men and married women in the married subsample (“Married” in Table 2, Chisquare ¼ 5.589, Prob. ¼ 0.133) and between unmarried men and unmarried women in the unmarried subsample (“Unmarried” in Table 2, Chi-square ¼ 3.982, Prob. ¼ 0.264). There is also no significant difference in job satisfaction between married men and unmarried men in the male subsample (Chi-square ¼ 3.483, Prob. ¼ 0.323). The finding of significant differences between men and women in the full sample is consistent with previous research; however, important clarification of differences in job satisfaction is noticeably gained when samples are stratified by marital status as well as gender. Table 3 provides summary statistics of variables that could influence job satisfaction. These statistics are shown separately by gender and marital status. An examination of these statistics shows many statistically significant differences by Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
339 Table 2 Job satisfaction by gender and marital status Total
Men
Women
Married Very Somewhat Not too Not at all Total
345 317 41 14
(48.1%) (44.2%) (5.7%) (2.0%) 717
163 178 23 7
(43.9%) (48.0%) (6.2%) (1.9%) 371
182 139 18 7
(52.6%) (40.2%) (5.2%) (2.0%) 346
Unmarried Very Somewhat Not too Not at all Total
299 357 78 26
(39.3%) (47.0%) (10.3%) (3.4%) 760
135 175 31 9
(38.6%) (50.0%) (8.9%) (2.6%) 350
164 182 47 17
(40.0%) (44.4%) (11.5%) (4.2%) 410
644 (43.6%) 674 (45.6%) 119 (8.1%) 40 (2.7%) 1,477
298 353 54 16
(41.3%) (49.0%) (7.5%) (2.2%) 721
346 321 65 24
(45.8%) (42.5%) (8.6%) (3.2%) 756
All Very Somewhat Not too Not at all Total
gender (at a 1, 5, or 10 percent level of significance). There are significant gender differences in the summary statistics of nine variables for both married and unmarried subsamples (percent white collar, percent blue collar, percent manufacturing industry, percent service industry, percent full-time, percent working a second job, percent low earners, percent where overtime is mandatory, and mean natural logarithm of hours worked). Married individuals show significant gender differences in two additional aspects (percent trade industry, percent reporting earnings less than fair17), while unmarried individuals show significant gender differences in six additional aspects (percent white, percent with a college degree, percent perceiving promotions as fair, percent saying job did not interfere with family, percent perceiving discrimination, and mean age). If one were to judge by the sheer number of significant gender differences in aspects that might be influential in affecting job satisfaction, one might guess that these patterns would have led to greater gender differences in job satisfaction among unmarried individuals than married individuals. Clearly, the varying magnitudes of both the gender differences in these aspects and their impacts on job satisfaction must be taken into consideration in explaining any pattern of gender differences in job satisfaction.
Regression results An analysis of job satisfaction holding these varying factors constant is required to answer questions about gender differentials in job satisfaction more definitively. For these purposes, ordered-probit estimation results for an initial model using dummy variables to capture gender and marital status (as is common) are presented in Table 4. An examination of the results for the sample of all workers in Column 1 (and the subsample of married workers in Column 2) suggests that job satisfaction Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
340 Table 3 Summary statistics of variables by marital status and gender Married
Percentages White Health Education White collar Blue collar Manufacturing Trade Services Government Full-time Second job Low earn Earnofair Earn>fair Fringe OK Fair prom Caring superv Build abilities Have input Take time off Not interfere Discrim Mand. OT High stress Change sched N Means (standard deviations) Age Tenure Ln hours N
Unmarried
Women
Men
p-value
Women
Men
p-value
82% 88 34 79 8 8 16 59 8 79 12 26 49 8 77 70 52 39 74 78 58 14 20 34 52 346
81% 88 32 49 42 23 12 29 11 97 18 5 39 8 75 74 52 39 73 77 53 10 31 31 55 371
0.82 1.00 0.57 0.00 0.00 0.00 0.09 0.00 0.13 0.00 0.03 0.00 0.01 0.89 0.48 0.18 1.00 0.99 0.84 0.70 0.23 0.13 0.00 0.38 0.42
67% 83 31 75 8 9 21 56 8 86 19 27 50 9 71 63 50 36 67 74 65 28 21 34 52 410
76% 84 23 44 42 20 19 32 8 90 25 19 46 8 73 69 46 37 69 72 59 17 33 29 47 350
0.01 0.70 0.02 0.00 0.00 0.00 0.37 0.00 0.96 0.08 0.04 0.01 0.37 0.90 0.61 0.08 0.25 0.89 0.61 0.61 0.06 0.00 0.00 0.16 0.14
42.3 (11.2) 7.46 (8.1) 3.60 (0.4) 346
42.6 (11.2) 8.17 (8.6) 3.83 (0.3) 371
0.70
40.1 (13.1) 6.20 (7.3) 3.66 (0.3) 410
36.2 (11.8) 5.83 (6.9) 3.74 (0.4) 350
0.00
0.25 0.00
0.47 0.01
for men (married men) is significantly lower than for women (married women), ceteris paribus, as reflected by the significant negative coefficient for the MALE dummy variable. By contrast, in the results for the subsample of unmarried workers in Column 3, there is no indication of a significant difference in job satisfaction between unmarried men and unmarried women, ceteris paribus. Interestingly, in the results for the sample of all workers in Column 1 (and the subsample of male workers in Column 4), there is no significant difference in job satisfaction between married (married male) workers and unmarried (unmarried male) workers, ceteris paribus, as reflected by the insignificant coefficient for the MARRIED dummy variable. However, an examination of the results for the subsample of female workers in Column 5 suggests that job satisfaction for married women is significantly higher than for unmarried women, ceteris paribus.18 Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
341 Table 4 Ordered probit results: Parameters of index function for job satisfaction by marital status and by gender
Constant Age Male White Married Health Education Tenure White collar Blue collar Manufacturing Trade Service Government Full-time Ln hours Second job Low earnings Earningsofair Earnings>fair Fringes OK Fair promotions Caring supervisor Building abilities Having input Taking time off Not interfering
All
Married
Not married
Men
0.067 (0.88) 0.011*** (0.00) 0.138* (0.07) 0.163** (0.03) 0.069 (0.31) 0.340*** (0.00) 0.049 (0.55) 0.013** (0.01) 0.138 (0.18) 0.134 (0.28) 0.392*** (0.00) 0.227* (0.05) 0.064 (0.54) 0.044 (0.77) 0.013 (0.92) 0.216* (0.05) 0.058 (0.49) 0.181* (0.06) 0.423*** (0.00) 0.049 (0.69) 0.192** (0.01) 0.489*** (0.00) 0.375*** (0.00) 0.663*** (0.00) 0.225*** (0.00) 0.202** (0.01) 0.026 (0.72)
0.028 (0.97) 0.011** (0.02) 0.369*** (0.00) 0.044 (0.71)
0.038 (0.95) 0.015*** (0.00) 0.105 (0.31) 0.288*** (0.01)
0.404 (0.50) 0.008 (0.11)
0.392*** (0.01) 0.028 (0.81) 0.018*** (0.01) 0.229 (0.18) 0.270 (0.17) 0.305* (0.06) 0.131 (0.45) 0.013 (0.93) 0.215 (0.32) 0.077 (0.72) 0.249 (0.16) 0.125 (0.36) 0.172 (0.27) 0.467*** (0.00) 0.028 (0.88) 0.110 (0.35) 0.562*** (0.00) 0.288*** (0.00) 0.608*** (0.00) 0.218* (0.05) 0.131 (0.28) 0.035 (0.73)
0.295** (0.01) 0.044 (0.72) 0.002 (0.82) 0.064 (0.62) 0.011 (0.95) 0.518*** (0.00) 0.390** (0.02) 0.167 (0.28) 0.190 (0.38) 0.005 (0.97) 0.229 (0.12) 0.257** (0.02) 0.171 (0.17) 0.452*** (0.00) 0.089 (0.61) 0.249** (0.02) 0.347*** (0.00) 0.526*** (0.00) 0.736*** (0.00) 0.273*** (0.01) 0.264** (0.01) 0.031 (0.75)
0.211* (0.07) 0.051 (0.62) 0.264** (0.05) 0.009 (0.94) 0.008 (0.27) 0.305* (0.07) 0.223 (0.21) 0.341** (0.01) 0.104 (0.51) 0.135 (0.34) 0.185 (0.38) 0.910*** (0.00) 0.366** (0.02) 0.091 (0.42) 0.076 (0.63) 0.481*** (0.00) 0.089 (0.61) 0.229** (0.04) 0.556*** (0.00) 0.428*** (0.00) 0.627*** (0.00) 0.039 (0.72) 0.151 (0.19) 0.124 (0.22)
Women 0.407 (0.53) 0.013*** (0.00)
0.138 (0.21) 0.232** (0.02) 0.477*** (0.00) 0.128 (0.25) 0.021*** (0.00) 0.078 (0.57) 0.191 (0.41) 0.387 (0.10) 0.334* (0.09) 0.051 (0.78) 0.120 (0.61) 0.452*** (0.01) 0.208 (0.22) 0.075 (0.58) 0.347*** (0.01) 0.434*** (0.00) 0.278 (0.15) 0.104 (0.36) 0.474*** (0.00) 0.364*** (0.00) 0.697*** (0.00) 0.418*** (0.00) 0.346*** (0.00) 0.117 (0.26)
Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
342 Table 4 (Continued)
Discrimination Mandatory OT High stress Changing schedule D06 Mu (1) Mu (2) Chi-square Predicted correct N
All
Married
Not married
Men
Women
0.314*** (0.00) 0.055 (0.47) 0.438*** (0.00) 0.079 (0.25) 0.070 (0.28) 0.930*** (0.00) 2.817*** (0.00) 642.3*** (0.00) 63.3% 1,477
0.339** (0.02) 0.041 (0.72) 0.373*** (0.00) 0.077 (0.46) 0.101 (0.30) 0.836*** (0.00) 2.757*** (0.00) 264.3*** (0.00) 64.3% 717
0.307*** (0.00) 0.083 (0.43) 0.562*** (0.00) 0.068 (0.48) 0.035 (0.70) 1.042*** (0.00) 2.965*** (0.00) 380.1*** (0.00) 62.2% 760
0.179 (0.20) 0.008 (0.94) 0.262** (0.02) 0.033 (0.75) 0.066 (0.48) 0.960*** (0.00) 2.978*** (0.00) 288.5*** (0.00) 63.0% 721
0.414*** (0.00) 0.086 (0.47) 0.638*** (0.00) 0.169* (0.09) 0.212** (0.02) 0.979*** (0.00) 2.854*** (0.00) 423.2*** (0.00) 66.3% 756
P-values for two-tailed tests are in parentheses. *indicates significance at the 10 percent level, **indicates significance at the 5 percent level, and ***indicates significance at the 1 percent level. Sampling weights provided by the GSS are used in this estimation process.
Controlling for personal characteristics and job aspects in this manner yields results that are only partially consistent with the basic findings of the chi-square tests of independence previously applied to the distributions of job satisfaction by gender and marital status (Table 2). However, these results could themselves be misleading, because (i) the effects of gender (marital status) are limited to intercept differences and (ii) the model does not allow for possible sample selectivity associated with gender differences in decisions to work. ^0 s associated with X’s) vary by To accommodate the possibility that slopes (i.e. b gender and marital status, the ordered-probit analysis is first performed separately for married men, married women, unmarried men, and unmarried women. To accommodate an investigation of sample selectivity associated with gender differences in the decisions to work, the ordered-probit analysis is then preceded by a first-stage probit analysis of the decision to work, applied to samples consisting of both workers and non-workers. The first-stage estimation results for this probit analysis of work status are reported in Appendix Table A3.19 The second-stage ordered-probit estimation results both with and without corrections for sample selectivity are presented in Table 5, where it can be seen that applying sample-selection correction within job-satisfaction equations estimated by gender and marital status has results that vary across demographic groups. In subsamples of married men, unmarried men, and unmarried women, inverse Mill’s ratios (represented by lambda) have no significant effect on job satisfaction. In the subsample of married women, however, lambda has a significant negative effect on job satisfaction. Given the formula used for lambda, married women with a greater probability of being at work have a lower value of lambda.20 The negative coefficient on lambda in the job-satisfaction equation thus implies that women with a greater probability of being at work have greater levels of job satisfaction than do married women who Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
343
have a lower probability of being at work.21 No such pattern is implied by the equations estimated for any of the other three samples (married men, unmarried men, and unmarried women).22,23 One might ask to what extent the previous finding of gender differences in the job satisfaction of married workers is in fact driven by gender differences in the slope coefficients of the job-satisfaction equation. In order to investigate the matter, we reestimate job satisfaction by marital status (married, unmarried), this time using a full set of interaction terms of gender and X’s to capture gender differences. This analysis allows us to gauge the extent to which gender differences in slope coefficients are significant and the extent to which there is any remaining gender difference unassociated with these slope coefficients or their X’s. The results of this analysis are reported in Table 6. These results reaffirm that sample selectivity (lambda) is a significant factor for married but not unmarried workers, but surprisingly the insignificance of the interaction of gender and lambda suggest no significant gender difference in the role of sample selectivity for married workers. ^0 s that These results highlight other specific personal/job characteristics that have b differ significantly by gender. Among married workers, these include tenure, fulltime status, hours of work, low income and having input about one’s job. The coefficients of these interaction terms imply that married men on average value having additional years of tenure, having full-time status, having low income and having input about one’s job less than married women, ceteris paribus. The coefficient on the interaction term for hours of work implies that married men on average value having additional hours of work more than married women, ceteris paribus. Simultaneity bias may explain why tenure appears on this list. If job satisfaction is a larger factor in the decision to remain at a job for married women than for married men, then the estimated effect of this variable on job satisfaction may be biased significantly upward for married women in comparison to married men.24 Given persisting differences in gender roles within families, it seems reasonable that married men and married women place different psychic values on level of pay. The significant gender difference in the effect of hours of work might be similarly explained. For unmarried workers, there are significant gender differences in the effects of numerous factors (age, health status, working in a service industry, full-time status, feeling earnings more than fair, feeling fringe benefits are good, finding work stressful, and year of survey) on job satisfaction. For example, ceteris paribus, unmarried men value good fringe benefits more than unmarried women do, while unmarried women value having earnings that are more than they deserve more than unmarried men do, on average.25,26 Finally, the results also suggest that conventional findings for married workers ^0 s of married workers; with these can be fully attributed to gender differences in the b specifications there is no significant remaining gender difference in intercept, as reflected by the coefficient of MALE. However, a significant gender difference in the intercept of the job satisfaction equation of unmarried workers is found; it implies an additional upward boost in the job satisfaction of male unmarried workers. This latter finding could reflect an underlying truth. However, it would also be consistent with any specification error that has a greater positive (or smaller negative) constant impact on the job satisfaction of unmarried men than on that of unmarried women. For example, it might be related to greater heterogeneity in the Eastern Economic Journal 2012 38
344
Ordered probit results: Parameters of index function for job satisfaction with and without sample selection correction Married men
Constant Age White Health Education Tenure White collar Blue collar Manufacturing Trade Service Government
Married women
Unmarried men
Unmarried women
Without
With
Without
With
Without
With
Without
With
0.726 (0.47) 0.015** (0.02) 0.085 (0.61) 0.418** (0.04) 0.110 (0.55) 0.008 (0.34) 0.306 (0.29) 0.264 (0.36) 0.178 (0.34) 0.030 (0.90) 0.035 (0.86) 0.254 (0.39)
0.645 (0.52) 0.021** (0.02) 0.071 (0.67) 0.284 (0.26) 0.126 (0.50) 0.009 (0.32) 0.314 (0.27) 0.255 (0.38) 0.168 (0.37) 0.032 (0.89) 0.040 (0.85) 0.248 (0.41)
0.625 (0.52) 0.004 (0.55) 0.048 (0.80) 0.327 (0.12) 0.047 (0.77) 0.038*** (0.00) 0.189 (0.42) 0.480 (0.18) 0.720** (0.04) 0.536* (0.08) 0.283 (0.29) 0.074 (0.84)
1.118 (0.27) 0.007 (0.34) 0.015 (0.94) 0.198 (0.38) 0.125 (0.47) 0.038*** (0.00) 0.237 (0.31) 0.591 (0.11) 0.704* (0.05) 0.505* (0.10) 0.237 (0.38) 0.002 (1.00)
0.933 (0.26) 0.002 (0.82) 0.373** (0.03) 0.070 (0.71) 0.146 (0.46) 0.005 (0.69) 0.393* (0.08) 0.203 (0.40) 0.680*** (0.00) 0.314 (0.17) 0.424* (0.05) 0.033 (0.92)
0.961 (0.27) 0.002 (0.82) 0.373** (0.03) 0.055 (0.82) 0.141 (0.49) 0.005 (0.69) 0.393* (0.08) 0.204 (0.40) 0.681*** (0.00) 0.313 (0.17) 0.423* (0.05) 0.034 (0.92)
1.523 (0.12) 0.021*** (0.00) 0.317** (0.02) 0.579*** (0.00) 0.165 (0.31) 0.003 (0.77) 0.036 (0.84) 0.300 (0.35) 0.016 (0.96) 0.108 (0.70) 0.228 (0.40) 0.078 (0.81)
1.989* (0.06) 0.020*** (0.00) 0.328** (0.02) 0.747*** (0.00) 0.041 (0.83) 0.003 (0.75) 0.013 (0.94) 0.275 (0.39) 0.007 (0.98) 0.134 (0.63) 0.207 (0.44) 0.088 (0.79)
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
Eastern Economic Journal 2012 38
Table 5
Full-time Ln hours Second job Low earnings Earningsofair Earnings>fair Fringes OK Fair promotions Caring supervisor Building abilities
Taking time off Not interfering
Mandatory OT High stress
0.451* (0.09) 0.071 (0.78) 0.253 (0.27) 0.524** (0.01) 0.566*** (0.00) 0.068 (0.82) 0.226 (0.23) 0.556*** (0.00) 0.249 (0.11) 0.711*** (0.00) 0.475*** (0.00) 0.378** (0.05) 0.165 (0.31) 0.453** (0.03) 0.108 (0.56) 0.535*** (0.00)
0.427 (0.11) 0.033 (0.89) 0.212 (0.36) 0.573*** (0.01) 0.572*** (0.00) 0.127 (0.67) 0.272 (0.15) 0.562*** (0.00) 0.237 (0.13) 0.729*** (0.00) 0.499*** (0.00) 0.359* (0.06) 0.171 (0.29) 0.448** (0.03) 0.092 (0.62) 0.567*** (0.00)
0.803*** (0.01) 0.314 (0.13) 0.332** (0.04) 0.181 (0.37) 0.557*** (0.00) 0.254 (0.33) 0.539*** (0.00) 0.324* (0.06) 0.541*** (0.00) 0.758*** (0.00) 0.097 (0.55) 0.323* (0.05) 0.046 (0.76) 0.223 (0.24) 0.059 (0.70) 0.388** (0.02)
0.807*** (0.01) 0.316 (0.13) 0.332** (0.04) 0.185 (0.36) 0.557*** (0.00) 0.254 (0.33) 0.540*** (0.00) 0.324* (0.06) 0.543*** (0.00) 0.760*** (0.00) 0.094 (0.57) 0.323* (0.05) 0.047 (0.76) 0.221 (0.25) 0.058 (0.71) 0.389** (0.02)
0.375 (0.12) 0.474* (0.07) 0.316* (0.07) 0.163 (0.35) 0.408*** (0.00) 0.562** (0.04) 0.043 (0.78) 0.346** (0.02) 0.517*** (0.00) 0.726*** (0.00) 0.414*** (0.00) 0.390** (0.01) 0.011 (0.94) 0.437*** (0.00) 0.170 (0.28) 0.761*** (0.00)
0.405* (0.10) 0.486* (0.07) 0.322* (0.06) 0.150 (0.39) 0.392*** (0.00) 0.561** (0.04) 0.023 (0.88) 0.335** (0.02) 0.509*** (0.00) 0.711*** (0.00) 0.415*** (0.00) 0.381** (0.02) 0.012 (0.93) 0.455*** (0.00) 0.145 (0.37) 0.771*** (0.00)
345
Eastern Economic Journal 2012 38
Discrimination
1.240** (0.02) 0.659** (0.02) 0.045 (0.80) 0.539* (0.06) 0.423*** (0.00) 0.091 (0.72) 0.011 (0.95) 0.694*** (0.00) 0.371*** (0.01) 0.538*** (0.00) 0.011 (0.95) 0.026 (0.88) 0.169 (0.23) 0.115 (0.63) 0.099 (0.51) 0.252 (0.10)
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
Having input
1.242** (0.02) 0.677** (0.01) 0.050 (0.78) 0.569** (0.05) 0.427*** (0.00) 0.105 (0.68) 0.008 (0.96) 0.686*** (0.00) 0.372*** (0.01) 0.533*** (0.00) 0.022 (0.89) 0.027 (0.87) 0.157 (0.27) 0.149 (0.52) 0.114 (0.44) 0.248 (0.11)
346
Married men
Changing schedule D06 Lambda Mu (1) Mu (2) Chi-square Predicted correct N
Married women
Without
With
Without
0.046 (0.76) 0.029 (0.83) — — 0.922*** (0.00) 2.944*** (0.00) 130.6*** (0.00) 62.5% 371
0.058 (0.69) 0.087 (0.55) 0.511 (0.35) 0.922*** (0.00) 2.948*** (0.00) 131.5*** (0.00) 62.5% 371
0.208 (0.20) 0.172 (0.25) — — 0.842*** (0.00) 2.828*** (0.00) 167.6*** (0.00) 69.1% 346
With 0.192 (0.24) 0.127 (0.41) 0.749* (0.10) 0.848*** (0.00) 2.842*** (0.00) 170.3*** (0.00) 68.5% 346
Unmarried men Without 0.033 (0.82) 0.200 (0.16) — — 1.098*** (0.00) 3.307*** (0.00) 190.5*** (0.00) 63.7% 350
With 0.034 (0.82) 0.202 (0.16) 0.049 (0.91) 1.097*** (0.00) 3.308*** (0.00) 190.6*** (0.00) 64.0% 350
Unmarried women Without 0.214 (0.11) 0.288** (0.02) — — 1.122*** (0.00) 2.993*** (0.00) 248.3*** (0.00) 63.4% 410
With 0.197 (0.14) 0.295** (0.02) 0.520 (0.23) 1.125*** (0.00) 2.999*** (0.00) 249.7*** (0.00) 62.7% 410
P-values for two-tailed tests are in parentheses. *indicates significance at the 10 percent level, **indicates significance at the 5 percent level, and ***indicates significance at the 1 percent level. Sampling weights provided by the GSS are used in this estimation process.
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
Eastern Economic Journal 2012 38
Table 5 (continued)
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
347 Table 6 Ordered probit results: Parameters of index function for job satisfaction with and without sample selection correction Married
Constant Age White Health Education Tenure White collar Blue collar Manufacturing Trade Service Government Fulltime Ln hours Second job Low earnings Earningsofair Earnings>fair Fringes OK Fair promotion Caring supervisor Building abilities Having input Taking time off Not interfering Discrimination
Unmarried
Without
With
Without
With
0.662 (0.49) 0.004 (0.54) 0.050 (0.79) 0.332 (0.11) 0.048 (0.77) 0.038*** (0.00) 0.191 (0.41) 0.487 (0.18) 0.724** (0.04) 0.542* (0.07) 0.285 (0.28) 0.074 (0.84) 0.455* (0.09) 0.069 (0.78) 0.254 (0.27) 0.528*** (0.01) 0.570*** (0.00) 0.070 (0.81) 0.227 (0.22) 0.562*** (0.00) 0.249 (0.10) 0.716*** (0.00) 0.479*** (0.00) 0.380** (0.04) 0.167 (0.30) 0.457** (0.03)
1.154 (0.25) 0.007 (0.33) 0.014 (0.94) 0.202 (0.37) 0.126 (0.46) 0.039*** (0.00) 0.239 (0.31) 0.599 (0.10) 0.707** (0.05) 0.509* (0.09) 0.238 (0.37) 0.001 (1.00) 0.430 (0.11) 0.032 (0.90) 0.213 (0.36) 0.577*** (0.00) 0.576*** (0.00) 0.128 (0.67) 0.274 (0.15) 0.568*** (0.00) 0.237 (0.12) 0.734*** (0.00) 0.503*** (0.00) 0.360* (0.06) 0.173 (0.28) 0.451** (0.03)
1.579 (0.12) 0.021*** (0.00) 0.333** (0.02) 0.609*** (0.00) 0.170 (0.32) 0.004 (0.74) 0.037 (0.84) 0.303 (0.37) 0.017 (0.96) 0.108 (0.72) 0.242 (0.39) 0.078 (0.82) 0.399 (0.12) 0.482* (0.08) 0.332* (0.07) 0.168 (0.36) 0.423*** (0.00) 0.581** (0.04) 0.047 (0.77) 0.366** (0.02) 0.545*** (0.00) 0.752*** (0.00) 0.431*** (0.00) 0.411** (0.01) 0.003 (0.99) 0.455*** (0.00)
2.056* (0.06) 0.020*** (0.00) 0.343** (0.02) 0.781*** (0.00) 0.043 (0.83) 0.004 (0.73) 0.015 (0.94) 0.277 (0.41) 0.008 (0.98) 0.136 (0.65) 0.220 (0.44) 0.088 (0.80) 0.430* (0.09) 0.494* (0.08) 0.338* (0.06) 0.155 (0.40) 0.407*** (0.00) 0.581** (0.04) 0.027 (0.87) 0.355** (0.02) 0.536*** (0.00) 0.737*** (0.00) 0.431*** (0.00) 0.401** (0.01) 0.004 (0.98) 0.473*** (0.00)
Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
348 Table 6 (Continued) Married
Mandatory OT High stress Changing schedule D06 Lambda Male M*Age M*White M*Health M*Education M*Tenure M*White collar M*Blue collar M*Manufacturing M*Trade M*Service M*Government M*Full-Time M*Ln hours M*Second Job M*Low earnings M*Earningsofair M*Earnings>fair M*Fringes OK M*Fair promotions M*Caring supervisor M*Building abilities
Eastern Economic Journal 2012 38
Unmarried
Without
With
Without
With
0.109 (0.55) 0.541*** (0.00) 0.210 (0.19) 0.172 (0.25) — — 1.41 (0.31) 0.011 (0.27) 0.036 (0.89) 0.082 (0.78) 0.061 (0.81) 0.030** (0.04) 0.114 (0.76) 0.224 (0.63) 0.549 (0.17) 0.573 (0.13) 0.321 (0.34) 0.326 (0.49) 1.684*** (0.00) 0.603 (0.11) 0.205 (0.48) 1.088*** (0.00) 0.146 (0.49) 0.176 (0.65) 0.220 (0.37) 0.117 (0.61) 0.121 (0.56) 0.186
0.094 (0.61) 0.573*** (0.00) 0.194 (0.23) 0.126 (0.41) 0.753* (0.10) 1.823 (0.20) 0.014 (0.23) 0.086 (0.74) 0.079 (0.81) 0.001 (1.00) 0.030** (0.04) 0.073 (0.84) 0.345 (0.46) 0.541 (0.18) 0.543 (0.15) 0.278 (0.41) 0.248 (0.60) 1.659*** (0.01) 0.623* (0.10) 0.168 (0.56) 1.108*** (0.00) 0.155 (0.47) 0.221 (0.57) 0.264 (0.29) 0.120 (0.60) 0.132 (0.53) 0.198
0.171 (0.31) 0.798*** (0.00) 0.222 (0.12) 0.297** (0.03) — — 2.517** (0.05) 0.020** (0.04) 0.016 (0.94) 0.549** (0.03) 0.313 (0.22) 0.001 (0.95) 0.416 (0.14) 0.507 (0.21) 0.664 (0.11) 0.184 (0.61) 0.639* (0.07) 0.121 (0.80) 1.173*** (0.00) 0.185 (0.59) 0.018 (0.94) 0.000 (1.00) 0.105 (0.59) 0.813** (0.03) 0.559** (0.01) 0.067 (0.76) 0.029 (0.89) 0.034
0.145 (0.39) 0.808*** (0.00) 0.205 (0.15) 0.305** (0.02) 0.535 (0.24) 3.015** (0.03) 0.019** (0.05) 0.007 (0.98) 0.732** (0.02) 0.182 (0.51) 0.001 (0.95) 0.394 (0.16) 0.483 (0.24) 0.655 (0.11) 0.156 (0.67) 0.616* (0.08) 0.131 (0.78) 1.208*** (0.00) 0.195 (0.57) 0.024 (0.92) 0.016 (0.95) 0.121 (0.54) 0.812** (0.03) 0.539** (0.01) 0.056 (0.80) 0.018 (0.93) 0.017
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
349 Table 6 (Continued) Married
M*Having input M*Taking time off M*Not interfering M*Discrimination M*Mandatory OT M*High stress M*Changing schedule M*D06 M*Lambda Mu(1) Mu(2) Chi-square Predicted correct N
Unmarried
Without
With
Without
With
(0.40) 0.457* (0.05) 0.353 (0.16) 0.323 (0.13) 0.310 (0.32) 0.221 (0.35) 0.294 (0.19) 0.164 (0.45) 0.142 (0.48) — — 0.883*** (0.00) 2.889*** (0.00) 304.5*** (0.00) 65.4% 717
(0.37) 0.493** (0.04) 0.335 (0.18) 0.341 (0.11) 0.338 (0.28) 0.190 (0.42) 0.323 (0.15) 0.136 (0.53) 0.039 (0.85) 0.245 (0.73) 0.886*** (0.00) 2.898*** (0.00) 308.2*** (0.00) 65.3% 717
(0.87) 0.341 (0.10) 0.104 (0.64) 0.043 (0.83) 0.243 (0.30) 0.117 (0.60) 0.430* (0.05) 0.186 (0.35) 0.488** (0.01) — — 1.110*** (0.00) 3.140*** (0.00) 443.5*** (0.00) 63.2% 760
(0.94) 0.342 (0.10) 0.094 (0.68) 0.042 (0.84) 0.262 (0.27) 0.091 (0.68) 0.438** (0.05) 0.168 (0.40) 0.497*** (0.01) 0.568 (0.36) 1.112*** (0.00) 3.143*** (0.00) 444.8*** (0.00) 62.9% 760
P-values for two-tailed tests are in parentheses. *indicates significance at the 10 percent level, **indicates significance at the 5 percent level, and ***indicates significance at the 1 percent level. Sampling weights provided by the GSS are used in this estimation process.
unmarried group than in the married group and omitted variables related to such heterogeneity.27 Alternatively, it might be associated with the lack of precise income data in this analysis. In any case, it suggests that more work may be needed to understand gender differences in the job satisfaction of unmarried workers.
CONCLUSION In this paper, gender differences in the job satisfaction of workers are investigated by marital status using recent US data. Applying the specifications of previous researchers to these data yields results somewhat similar to those found by previous researchers. The gender difference that had been found by previous researchers using these specifications with older US data [Bender et al. 2005] — that is, greater job satisfaction among women — appears, but is limited to married workers. In an effort to investigate the possible role of variation in taste, we apply the analysis to subsamples separated by gender and marital status (married men, married women, unmarried men, and unmarried women). Applying the analysis to such subsamples diverges from conventional specifications of job satisfaction, in that it allows for gender effects beyond intercept differences. We also expand the Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
350
analysis to include techniques designed to correct for sample selection. In doing so, we initially find that sample selectivity related to the decision to work appears to have a role in explaining the job satisfaction of married women but not to have a role in explaining the job satisfaction of married men, unmarried men, and unmarried women. This pattern is strikingly similar to that commonly found by researchers studying the role of sample selectivity in the earnings equations of workers separated by gender and marital status. ^0 s by pooling men and women by marital We test for significant differences in b status and estimating the job satisfaction equation with a full set of interaction terms of gender and X’s. The findings suggest that sample selectivity has similar effects for ^0 s of other X factors married men and married women; they point to differences in b as largely responsible for the conventional finding that married women report greater job satisfaction than married men. While significant gender differences in ^0 s of X factors exist for unmarried workers also, a gender intercept difference the b remains significant; it provides an additional boost to the job satisfaction of unmarried men. The finding suggests that further work needs to be done to understand the nature of gender differences in the job satisfaction of unmarried workers. The findings concerning differences in the relevance of sample selection for married and unmarried workers seem logical. There can understandably be greater discretion for married individuals in decisions about working in the market, given opportunities for specialization and exchange within the households of married couples. Moreover, changing social attitudes about gender roles may be making sample selectivity issues more relevant for analysis of the behavior of married men. ^0 s in job satisfaction equations estimated for However, gender differences in b married workers seem consistent with the existence of some remaining gender differences in the household roles of married individuals. The relationship between ^0 s and reflecting the job characteristics and job satisfaction (as captured by the b extent to which the individual feels that the job helps the individual meet one’s objectives in life) may well differ by gender, if priorities and objectives regarding work and earnings still differ by gender, as they would if working is still more typically a primary role for married men than for married women. ^0 s appear more widespread in job satisfaction Significant gender differences in b equations estimated for unmarried workers. The finding may be due to greater heterogeneity in the subsample. Applying this analysis to panel data could shed additional light on the subject. Panel data would make it possible to investigate the potential endogeneity of job tenure by considering the effects of job satisfaction on decisions to continue working at a particular job. Panel data would also make it possible to explore the effects of changes in marital status on job satisfaction and allow some control for individual heterogeneity. Additional research along these lines could advance the understanding of gender differences in job satisfaction, beyond what can be done here with cross-sectional data.
Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
351
Appendix Table A1 Definitions of additional variables Number of kids w/kidso6 yrs Spouse’s educ Spouse work ft
=number of household members under age 18 =1 if any household members are under age 6; =0 if otherwise =1 if spouse has a BA or higher degree; =0 if otherwise =1 if spouse is a full-time worker; =0 if otherwise
Table A2 Summary statistics of variables for first-stage probit, by marital status and gender Married
Percentages White Health Education w/kidso6 yrs Spouse’s educ Spouse work FT N Means (standard deviations) Age #of kids N
Unmarried
Women
Men
p-value
Women
Men
p-value
82% 81 27 21 29 75 886
79% 80 30 20 31 47 685
0.11 0.63 0.18 0.90 0.48 0.00 —
65% 71 23 14 — — 1,327
72% 77 22 5 — — 1,114
0.00 0.00 0.70 0.00 — — —
44.2 (12.6) 0.86 (1.2) 886
45.9 (12.5) 0.81 (1.2) 685
0.01
42.0 (14.6) 0.62 (1.1) 1,327
39.5 (14.3) 0.24 (0.6) 1,114
0.00
0.38 — —
0.00 — —
Table A3 Probit results: Parameters of index function for being at work
Constant Age Age squared White Health Education Number of kids w/kidso6 yrs Spouse’s educ Spouse work ft
Married men
Married women
Unmarried men
Unmarried women
0.271 (0.79) 0.058 (0.17) 0.001** (0.01) 0.182 (0.25) 0.812*** (0.00) 0.079 (0.64) 0.298*** (0.00) 0.681*** (0.01) 0.291* (0.10) 0.028 (0.83)
2.005*** (0.00) 0.111*** (0.00) 0.001*** (0.00) 0.133 (0.25) 0.341*** (0.00) 0.241** (0.04) 0.178*** (0.00) 0.202 (0.17) 0.206* (0.07) 0.365*** (0.00)
2.667*** (0.00) 0.173*** (0.00) 0.002*** (0.00) 0.014 (0.88) 0.565*** (0.00) 0.125 (0.27) 0.181*** (0.01) 0.121 (0.54) — — — —
1.972*** (0.00) 0.118*** (0.00) 0.002*** (0.00) 0.078 (0.33) 0.553*** (0.00) 0.401*** (0.00) 0.040 (0.33) 0.416*** (0.00) — — — —
Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
352 Table A3 (Continued) Married men D06 Chi-square Predict correct N
0.407*** (0.00) 174.3*** (0.00) 85.0% 685
Married women 0.045 (0.61) 108.0*** (0.00) 69.1% 886
Unmarried men 0.074 (0.40) 149.7*** (0.00) 75.1% 1,114
Unmarried women 0.043 (0.58) 204.4*** (0.00) 69.4% 1,327
This equation is the first stage of a two-stage Heckman-like method of sample selection correction, for the ordered-probit results reported in Tables 5 and 6. P-values for two-tailed tests are in parentheses. *indicates significance at the 10 percent level, **indicates significance at the 5 percent level, and ***indicates significance at the 1 percent level. Sampling weights provided by the GSS are used in this ^0 s of the first-stage probit equations for married estimation process. Significant gender differences in the b individuals are found for the estimated coefficients of health status (5 percent level of significance), number of children (1 percent level of significance), presence of children under the age of six years (10 percent level of significance), work status of spouse (5 percent level of significance), education of spouse (5 percent level of significance) and year (1 percent level of significance). Significant gender ^0 s of the first-stage probit equations for unmarried individuals are found for the differences in the b estimated coefficients of education (10 percent level of significance), age and age squared (5 percent level of significance), number of children (10 percent level of significance) and presence of children under the age of six years (5 percent level of significance). It is not surprising that there is a significant gender difference (1 percent level of significance) in the average values of lambda for both married and unmarried workers.
Notes 1. As Budig and England [2001] have found, there is a wage disadvantage for married mothers, which lowers their opportunity cost of non-market work, so they may choose not to work. 2. The traditional thinking among economists is that measurements of utility are subjective and interpersonal comparisons of such measurements are not useful. However, these researchers point to the numerous findings of strong relationships between job satisfaction and various aspects of worker behavior, as evidence that the study of interpersonal comparisons of self-reported measures of job satisfaction is indeed meaningful. For a survey of research on happiness in general, see Frey and Stutzer [2002]. 3. For seminal work on sample selectivity, see Heckman [1976]. 4. The specification for this portion of the analysis is fairly standard. For example, in estimating participation equations by gender, Christofides et al. [2003] use independent variables that reflect age, education, presence of children, whether disabled, whether married, whether a family head and whether an immigrant, along with controls for region of residence. 5. While there is some overlap in X and W, there are certain variables included in X that are not included in W and vice versa. X contains characteristics of one’s job, while W does not. Household composition and unearned income (reflected here by number of children, presence of children under the age of 6, spouse’s education level, and spouse’s work status) are typically components of W but not of X. These patterns are true for the participation equations and the job satisfaction equations appearing in Clark [1997] and Sloane and Williams [2000], for example. One notable exception can be found in Donohue and Heywood [2004], where the presence of young children was an independent variable in one specification of a job satisfaction equation. The authors explained its inclusion by saying that they anticipated that the variable could serve as a control for sample selection issues associated with women; in fact, they found it to be related to higher job satisfaction for both white-collar female and blue-collar male workers. Here, we choose to control for sample selection issues by addressing job satisfaction and participation in separate equations. 6. Details about this technique can be found in Heckman [1976]. 7. In studies of US workers, Donohue and Heywood [2004] used the 1988 National Longitudinal Survey of Youth, while Bender et al. [2005] used the 1997 National Study of the Changing Workforce. In studies of British workers, Clark [1997] used the 1991 British Household Panel Survey, while Sloane Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
353
8. 9.
10.
11.
12.
13.
14.
15.
16.
17. 18.
19.
and Williams [2000] analyzed the 1986 Social and Economic Life Initiative Survey. In an international comparison of workers, Sousa-Poza and Sousa-Poza [2000] used data from the 1997 International Social Survey Program (ISSP); the US data of the ISSP have been collected as a module of the GSS since 1994. http: //www.norc.org/GSS+Website/About+GSS/ Historically, survey results were based on independently drawn samples of English-speaking persons 18 years of age or over, living in non-institutional arrangements within the United States. Starting in 2006 Spanish-speaking individuals were added to the target population. Since the beginning, the GSS employed a rotation design; permanent items appeared on every survey, but rotating items appeared on two out of every three surveys. The importance of the rotating scheme has increased over the years; more items have been shifted from permanent to rotating status, to make room for topical modules of special interest [Davis and Smith 2009, GSS codebook, Appendix Q]. Sloane and Williams [2000] found that neither absolute pay nor objective (predicted) comparison pay were significant when entered jointly into a job-satisfaction equation. However, they did find that both absolute and subjective comparison pay were highly significant when entered jointly. For an income variable available in both the 2002 and 2006 versions of the GSS, the categories for annual earnings are: less than $1,000; between 1,000 and 3,000; between 3,000 and 4,000; between 4,000 and 5,000; between 5,000 and 6,000; between 6,000 and 7,000; between 7,000 and 8,000; between 8,000 and 10,000; between 10,000 and 15,000; between 15,000 and 20,000; between 20,000 and 25,000; and 25,000 and above. The combined total number of respondents to the 2002 and 2006 GSS was 7,275. We excluded 845 who were over the age of 70. Both health and education of spouse were “rotating items” in the GSS in these years. As a result, a total of 1,900 respondents were never asked about health and 760 married individuals were never asked about spouse’s education. The losses of observations largely explain the reduction to the 4,012 observations used in the first stage of estimation. Only 2,732 of the 4,012 observations used in the first stage of estimation were part- or full-time workers. Only 1,477 of these are used in the second stage of estimation. The reduction in numbers is largely due to the fact that about 1,100 workers were never asked about quality of working life in 2006 and in both years combined over 500 respondents with income did not provide information about that income. Additionally we excluded 163 part- or full-time workers with data because they were self-employed. While it would be interesting to explore the job satisfaction of self-employed workers separately, the numbers of selfemployed workers were too small to support such an investigation here. In previous studies of job satisfaction, researchers have used either health status as represented by reported limitation on activity [Donohue and Heywood 2004; Bender et al. 2005] or self-reported evaluations of health [Clark 1997]. In the 2002 and 2006 GSS data, only the latter is available for use here. When asked about the fairness of earnings in comparison to others doing the same type of work, respondents were asked to indicate if earnings were much less than deserved, somewhat less than deserved, about what was deserved, somewhat more than deserved or much more than deserved. For four variables, a statement about working environment was made (e.g. “My fringe benefits are good.”) and respondents were asked whether the statement was “very true,” “somewhat true,” “not too true,” or “not at all true.” In regards to amount of employee input on the job, respondents were asked whether they “strongly agree,” “agree,” “disagree,” or “disagree strongly” with the statement “I have a lot of say about what happens on my job.” In terms of ability to take time off, respondents were asked to indicate whether taking time off from work for personal or family matters is “not hard,” “not too hard,” “somewhat hard,” or “very hard.” As for the frequency with which the job interferes with family life and the worker can make changes in starting or quitting times on a daily basis, respondents were asked to indicate “often,” “sometimes,” “rarely,” or “never.” In answering a question about whether the respondent finds work stressful, respondents were asked to indicate “always, ” “often, ” “sometimes, ” “rarely,” or “never. ” Table 1 indicates how responses were collapsed into fewer categories for 11 related variables. This finding could very well reflect the larger wage penalty associated with married mothers, found by Budig and England [2001]. These findings with respect to MALE and MARRIED are robust across different specifications (with and without sample weights, with and without HEALTH, with and without AGE squared, with and without D06) and across different samples (70 years old and younger, 60 years old and younger). More specifically, Appendix Table A3 shows the results of first-stage probit estimation for the likelihood of being at work full-time or part-time (as opposed to being with a job but not at work, unemployed, laid off, in search of a job, retired, or in school, keeping house or doing something else), by gender and marital status. Appendix Table A1 defines additional independent variables used in the Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
354
20. 21. 22.
23.
24.
25.
26.
27.
analysis; Appendix Table A2 provides associated descriptive statistics for the sample used in this firststage of estimation. See Heckman [1976, p. 479]. Interestingly, a significant negative coefficient for lambda is also frequently found in earnings equations for women [ Dolton and Makepeace 1987; Wright and Ermisch 1991; Sloane and Williams 2000]. These results are based on first-stage samples that include workers who were not asked questions about job satisfaction and who were subsequently excluded in the samples of workers used in the second-stage estimation, because of missing data. The reduction in the number of workers from the first to the second stage of estimation is unfortunate. To avoid the loss of valuable information about the decision to work, however, we retained the full set of workers for the first-stage estimation, rather than omitting them throughout the analysis. The estimation is repeated with different specifications (without HEALTH; with AGE squared; with TENURE squared) and a different sample (those under the age of 66). When the sample is limited to those under the age of 66, the results are similar except that the P-value for the lambda of married women slips to 0.12. When HEALTH is omitted from the ordered-probit stage, the lambdas of both married men and married women are significant at conventional levels of significance. When TENURE squared is added, it is associated with a negative but insignificant effect; with its addition, the P-value of the lambda for married women slips to 0.13 while the lambdas for all other groups remain insignificant. When AGE squared is added, a significant U-shaped relationship between age and job satisfaction is found for married men only and only when the model includes a correction for sample selection; as a result of its inclusion, the lambda for married women becomes rather insignificant (P-value ¼ 0.22) while the lambda for married men becomes highly significant (P-value ¼ 0.03). Given the oddness of the latter results and the nature of the variable itself, we feel it more relevant for the first-stage probit and less relevant than TENURE squared for the second-stage ordered probit. A number of studies have found that dissatisfied workers are more likely to quit their jobs. (See Bo¨ckerman and Ilmakunnas [2009] as a recent example that provides a listing of others.) However, it is not clear that any of these have investigated whether the responsiveness of quit rates to job dissatisfaction is any greater among married women than married men. These findings about significant gender differences in coefficients could be misleading, if there is multicollinearity among the X variables. That is, significant gender differences in other coefficients could be masked by multicollinearity among X variables. Calculations of variance inflationary factors (VIFs) for these variables do not seem to indicate severe multicollinearity, however. For married women, the highest VIFs for the X variables in the job satisfaction equation are under 3.5 (trade 2.5, service 3.4, ln hours 2.5 and full-time status 2.4). For married men, the highest VIFs are a bit higher (white collar 5.5, blue collar 5.6, lambda 3.5 and age 2.8, with the latter two occurring in the model with sample selection correction only). For unmarried workers, primarily the VIFs for types of occupations or industries exceed 2.5 (men white collar 3.2, men blue collar 3.6, women trade 4.0, women service 5.1). Additionally, for unmarried women in the model with sample selection correction, the VIF for lambda is 2.9. These findings could also be misleading, if there is misspecification in the model. Misspecification of the model would be problematic in this regard, in that the estimated coefficients of identified variables (e.g. FT) could be biased because of some omitted relevant variable(s) correlated with them. If any such omitted variables differ in importance to the job satisfaction of men and women, then the extent of the bias would vary by gender and the differing effects of these omitted variables, rather than the identified variables, could in fact be the source of gender differences in job satisfaction. This comment applies to the discussion that follows as well. Notably, within the unmarried group there are significant gender differences in household composition: number of children (averaging 0.63 for women but only 0.24 for men) and presence of children under the age of six years (for 13 percent of women but only for 6 percent of men). The direction of these differences is exactly the opposite in the case of married workers; in comparison to married men who work, married women who work have fewer children on average (0.75 as opposed to 0.9) and are less likely to have children under the age of six years (16 percent as opposed to 25 percent). As a result of different life circumstances, unmarried women who work may feel less satisfaction from their jobs than unmarried men who work, in a manner unrelated to gender ^0 s associated with those X’s. The finding of a differences in the X’s being used here or in the b significant positive coefficient for MALE is not altered by simply including measurements of household composition (number of children and presence of children under the age of six) among the X’s in the job satisfaction equation for unmarried workers, however. The results of such an effort are not shown in Table 6, but are available upon request.
Eastern Economic Journal 2012 38
Cheryl J. Carleton and Suzanne Heller Clain Women, Men, and Job Satisfaction
355
References Akerlof, George A., Andrew K. Rose, and Janet L. Yellen. 1988. Job Switching and Job Satisfaction in the US Labor Market. Brookings Papers on Economic Activity, 2(2): 495–582. Bartel, Ann P. 1981. Race Differences in Job Satisfaction: A Reappraisal. Journal of Human Resources, 16(2): 294–303. Bender, Keith A., Susan M. Donohue, and John S. Heywood. 2005. Job Satisfaction and Gender Segregation. Oxford Economic Papers, 57(3): 479–496. Berg, Nathan, and Donald Lien. 2002. Measuring the Effect of Sexual Orientation on Income: Evidence of Discrimination? Contemporary Economic Policy, 20(4): 394–414. Bo¨ckerman, Petri, and Pekka Ilmakunnas. 2009. Job Disamenities, Job Satisfaction, Quit Intentions, and Actual Separations: Putting the Pieces Together. Industrial Relations, 48(1): 73–96. Borjas, George J. 1979. Job Satisfaction, Wages and Unions. Journal of Human Resources, 14(1): 21–40. Budig, Michelle J., and Paula England. 2001. Wage Penalty for Motherhood. American Sociological Review, 66(2): 204–225. Christofides, Louis, Qi Li, Zhenjuan Liu, and Insik Min. 2003. Recent Two-Stage Sample Selection Procedures with an Application to the Gender Wage Gap. Journal of Business & Economic Statistics, 21(3): 396–405. Clark, Andrew E. 1997. Job Satisfaction and Gender: Why Are Women So Happy at Work? Labour Economics, 4(4): 341–372. _______ 2001. What Really Matters in a Job? Hedonic Measurement Using Quit Data. Labour Economics, 8(2): 223–242. Clark, Andrew E., and Andrew J. Oswald. 1996. Satisfaction and Comparison Income. Journal of Public Economics, 61(3): 359–381. Clegg, Chris W. 1983. Psychology of Employee Lateness, Absence, and Turnover: A Methodological Critique and an Empirical Study. Journal of Applied Psychology, 68(1): 88–101. Davis, James A., and Tom W. Smith. 2009. General Social Surveys, 1972–2008, [machine-readable data file] /Principal Investigator, James A. Davis; Director and Co-Principal Investigator, Tom W. Smith; Co-Principal Investigator, Peter V. Marsden; Sponsored by National Science Foundation. — NORC ed. — Chicago: National Opinion Research Center [producer]; Storrs, CT: The Roper Center for Public Opinion Research, University of Connecticut [distributor], 1 data file (53,043 logical records)+1 codebook (2,656 pp.). — (National Data Program for the Social Sciences, no. 18). Dolton, Peter J., and Gerald H. Makepeace. 1987. Marital Status, Child Rearing and Earnings Differentials in the Graduate Labor Market. Economic Journal, 97(388): 897–922. Donohue, Susan M., and John S. Heywood. 2004. Job Satisfaction and Gender: An Expanded Specification from the NLSY. International Journal of Manpower, 25(2): 211–234. Flanagan, Robert J., George Strauss, and Lloyd Ulman. 1974. Worker Discontent and Work Place Behavior. Industrial Relations, 13(2): 101–123. Freeman, Richard B. 1978. Job Satisfaction as an Economic Variable. American Economic Review, 68(2): 135–141. Frey, Bruno S., and Alois Stutzer. 2002. What Can Economists Learn from Happiness Research? Journal of Economic Literature, 40(2): 402–435. Hamermesh, Daniel S. 2001. The Changing Distribution of Job Satisfaction. Journal of Human Resources, 36(1): 1–30. Heckman, James 1976. The Common Structure of Statistical Models of Truncation, Sample Selection and Limited Dependent Variables and a Simple Estimator for Such Models. Annals of Economic and Social Measurement, 5(4): 475–492. Kristensen, Nicolai, and Edvard Johansson. 2008. New Evidence on Cross-Country Differences in Job Satisfaction Using Anchoring Vignettes. Labour Economics, 15(1): 96–117. Mangione, Thomas W., and Robert P. Quinn. 1975. Job Satisfaction, Counterproductive Behavior, and Drug Use at Work. Journal of Applied Psychology, 60(1): 114–116. Sloane, Peter J., and Hector Williams. 2000. Job Satisfaction, Comparison Earnings, and Gender. Labour, 14(3): 473–502. Sousa-Poza, Alfonso, and Andre´s A. Sousa-Poza. 2000. Taking Another Look at the Gender/ Job-Satisfaction Paradox. KYKLOS, 53(2): 135–152. Vroom, Victor H. 1964. Work and Motivation. New York: Wiley. Wright, Robert, and John F. Ermisch. 1991. Gender Discrimination in the British Labor Market: A Reassessment. Economic Journal, 101(406): 508–522.
Eastern Economic Journal 2012 38