With women entering law and other professions in numbers rivaling and exceeding those of men, it makes sense to ask how they are faring and how they are feeling about their experiences. Reference Menkel-MeadowMenkel-Meadow (1989) posed the issue that would dominate several following decades of legal practice for women, observing that “whether a feminist critique of the legal profession will emerge and transform the profession or structural obstacles will silence feminist concerns or force assimilation is a question that must be answered empirically and theoretically” (1989:289; see also Reference BarnettBarnett 1990; Reference GlendonGlendon 1994; Reference RhodeRhode 2000; Reference SchiltzSchiltz 1999). There is growing evidence that while the legal profession has structurally assimilated women, it has neglected many of their broader concerns, leaving conventional legal practice untransformed in many important respects (Reference Schultz and ShawSchultz & Shaw 2003). Yet the findings from studies of job satisfaction among lawyers continue to confuse this picture.
On the one hand, research indicates that while women have entered law in large numbers that have helped greatly expand the profession's client base of individuals and especially corporations, women still lag behind men lawyers in rates of partnership, remuneration, and retention in firm settings (Reference Hagan and KayHagan & Kay 1995; Reference Kay and HaganKay & Hagan 1998). For example, Reference EpsteinEpstein et alia (1995) conclude from their mid-1990s research in New York City that at the highest levels, elite law firms remain stubbornly resistant to the advancement of women. They observe that “women … seem to be leaving large law firms disproportionately more than men, meaning that the profile of the ‘Wall Street lawyer’ at the very top partnership levels looks nothing like the distribution of lawyers at the bottom of the pyramid” (Reference EpsteinEpstein et al. 1995:439). This continues to be true beyond New York: although the composition of the North American legal profession is now nearly one-third female, only 17 percent of law firm partners are women (see http://www.nalp.org/content), and about 5 percent of managing partners are women (Reference RhodeRhode 2001:14; see also Reference Hagan, Kay and WiltonHagan & Kay 1996; Reference SpanglerSpangler 1986).
Put simply: women enter firms at rates comparable to those for men, but they are more likely to leave (Reference BrockmanBrockman 1994; Reference Reichman and SterlingReichman & Sterling 2002), and sooner (Reference KayKay 1997), while they earn less (Reference HaganHagan 1990; Reference RhodeRhode 2001; Reference Stager and FootStager & Foot 1988) and are less likely to become partners (Reference DonnellDonnell et al. 1998; Reference EpsteinEpstein et al. 1995; Reference Hagan and KayHagan & Kay 1995; Reference RadfordRadford 1990). Reference RhodeRhode (2001) concludes a recent American Bar Association (ABA) survey by remarking that “in accounting for these persistent and pervasive disparities, a wide array of research reveals common patterns … : Women's opportunities are limited by traditional gender stereotypes, by inadequate access to mentors and informal networks of support, by inflexible workplace structures, and by other forms of gender bias in the justice system” (2001:14; see also Reference Schultz and ShawSchultz & Shaw 2003). Rhode further notes that about one-quarter of female lawyers and 3 percent of male lawyers think the prospects for advancement in the profession are greater for men than for women (Reference SambornSamborn 2000). Depending on your point of view, this may be a professional picture that is one-quarter empty or three-quarters full.
A parallel body of research paradoxically shows that women as well as men lawyers respond to surveys and interviews by reporting that they are highly satisfied with the substance of their work (Reference ChambersChambers 1989; Reference DinovitzerDinovitzer et al. 2004; Reference KayKay et al. 2004b; Reference Mattessich and HeilmanMattessich & Heilman 1990). There is evidence of specific issues such as job setting and advancement where women lawyers indicate somewhat less satisfaction than men lawyers (e.g., American Bar Association 2006; Reference DinovitzerDinovitzer et al. 2004; Reference KayKay et al. 2004a, Reference Kay, Sheehy and McIntyre2006; Reference WallaceWallace 2004), but the findings are nonetheless paradoxical in the overall expressions of satisfaction among both men and women lawyers. Thus in their recent study of Chicago lawyers (see also Reference Heinz and LaumannHeinz & Laumann 1982; Reference HeinzHeinz et al. 2005), Reference HeinzHeinz et al. (1999) observe that
One of the persistent puzzles is why the overall job satisfaction of female lawyers, in spite of their circumstances (work setting, income, work/family conflicts), is equivalent to that of males. When we examined their evaluations of more specific aspects of their employment, such as their chances for advancement, salary, and level of responsibility, we found that women were significantly (though not hugely) more likely than men to say that they would not again choose to become a lawyer. So we should be careful that we do not overstate the equivalence of men's and women's job satisfaction. Here again, however, the findings regarding lawyers mirror more general findings. Women's job satisfaction, regardless of occupational setting, has usually been found to be equal to—sometimes, greater than—that of men (Reference HeinzHeinz et al. 1999:755).
Heinz et al. give this finding further perspective by observing that “it is not immediately apparent why one should expect lawyers to be more disgruntled than people who sell cars or fix teeth” (1999:755).
A recent national study (Reference DinovitzerDinovitzer et al. 2004) of more than 5,000 U.S. lawyers in their first two to three years of practice adds to the puzzle about the satisfaction of women compared to that of men lawyers. The findings from this research again indicate that women and men are both generally satisfied with their decision to become a lawyer, but also that there are further differences in their comparative sources of satisfaction.Footnote 1 Dinovitzer et al. report that “women are significantly more satisfied than men with the substance of their work” (2004:58), but that this source of satisfaction must be weighed against “other dimensions of satisfaction where women's ratings are significantly lower than those of men: namely job setting, social index of work, and the power track” (2004:58). They further report (see also Reference Hagan and KayHagan & Kay 1995) that women are more likely than men to intend changing jobs in the next two years. Yet both men and women in this national study report generally high levels of work satisfaction, “with no evidence … of any pervasive unhappiness” (Reference DinovitzerDinovitzer et al. 2004:47).
The finding that women across a range of jobs express relatively high overall levels of satisfaction with their work experience has been recognized in the fields of social psychology and work and occupations for several decades (Reference Firebaugh and HarleyFirebaugh & Harley 1995), and it is dubbed by Reference CrosbyCrosby (1982; see also Reference HodsonHodson 1989) “the paradox of the contented female worker.” Still, as Reference Mueller and WallaceMueller and Wallace (1996) and Reference HullHull (1999) recently have noted, gender equality in contentment among lawyers—an occupational group whose members are trained regardless of gender to be contentious—is an especially notable paradox (see also Reference ChuiChui 1998; Reference MacCorquodale and JensenMacCorquodale & Jensen 1993).
We suggest in this article that this paradox of similarity in satisfaction obscures a more notable difference in depressed feelings between men and women lawyers, and that these feelings are as much work- as gender-related. We argue that analysis of this difference can be a source of insight into the complex orientations of women lawyers to their profession. Our point is not that all or even most women lawyers are seriously depressed or distressed about their work, but that their more frequent indications of depressed or despondent feelings nonetheless deserve attention.
We further suggest that it is especially important to pursue the link between gender and work-related feelings in the legal profession. Reference MillsMills (1956), ironically if not prophetically, called the profession's firms of the 1950s “law factories” and argued that its employees and their offices mimicked in an exaggerated way the practices of the modern corporation. The timing of partnership decisions in law firms comes at a crucial time in women's decisionmaking about childbearing and today creates a special conjuncture between corporatism and professionalism that seems especially likely to influence women's prospective and retrospective feelings about their work (see, for example, Reference O'BrienO'Brien 2006 and Reference Reichman and SterlingReichman & Sterling 2002).Footnote 2 A woman lawyer in the interview phase of this research summarized her confrontation with this conjuncture in this way:
The profession continues to discriminate on the basis of gender: it is more difficult for females to obtain partnerships, and often the only way a female lawyer who is a parent can maintain her status in a firm is by denying the importance of her parenting responsibilities.
A central premise of this article is that the feelings of satisfaction women as well men often take in modern lawyering can obscure feelings of despondency and depression that also accompany legal careers, especially for women confronted with conflicting priorities of partnership and parenting. The fact that feelings of depression are so often channeled inward, while feelings of dissatisfaction are so often projected outward, may help explain why the feelings of women lawyers about their work are often misconstrued.
The Gendered and the Generic
That women lawyers contemplate leaving their jobs from their first years of practice and then actually do leave firm practices at rates higher than men could certainly be a hint that all is not well with regard to gender issues in the legal profession; but at the same time it remains puzzling that women do not report being dissatisfied when asked about their overall satisfaction with their work. Reference Mueller and WallaceMueller and Wallace (1996) report an important analysis of this paradox among practicing lawyers in a Canadian city. They find that on average, women lawyers make nearly $28,000 less than male lawyers and correspondingly perceive fewer opportunities and resources for advancement. Yet the women in this sample of lawyers are not significantly different in their reports of job satisfaction from the men. The paradox here as elsewhere (e.g., Reference VarcaVarca 1983) is that women lawyers who should be less satisfied than men lawyers, given their work conditions, tend to be more generic than gendered in their reported feelings of overall satisfaction.
Mueller and Wallace's study is useful and important because they go on to try to account for the paradoxical “contentment” of women lawyers in terms of a variety of relevant factors. Their first notable finding in doing so involves estimating an overall job satisfaction equation that includes measures of work motivation and “negative affectivity,” or as it is more commonly understood, depression. Work motivation and depression are strongly correlated with job satisfaction, but in opposite ways, with motivation increasing and depression reducing satisfaction. Statistically taking these variables into account in Mueller and Wallace's analysis of work satisfaction renders the effect of gender unexpectedly significant and indicates that women are more satisfied than men. Descriptive data presented earlier in the Mueller and Wallace article confirm that women lawyers in the sample have more “negative affect”—indicating that they are more depressed or despondent—while they are similarly motivated as men. These findings therefore imply that when the effect of women being more depressed than men is statistically removed, the remaining effect is of women lawyers being more satisfied with their jobs than the men lawyers.
Mueller and Wallace then introduce a number of other variables into their work satisfaction equations. These variables include workplace rewards and perceptions of job conditions. As noted above, the men in this sample receive greater rewards and perceive their job conditions more favorably than women. When these variables are introduced into the overall job satisfaction analysis, Mueller and Wallace find that the effect of gender on job satisfaction is reduced back to about its original nonsignificant level (see also Reference PhelanPhelan 1994). When attention is shifted to pay satisfaction as a more specific outcome measure of work satisfaction, a similar pattern of results emerges, with the positive effect of gender being reduced, although not to statistical nonsignificance (see also Reference Witt and NyeWitt & Nye 1992). In the end it is probably most important to keep in mind that no matter what specific variables are included in Mueller and Wallace's analyses, and regardless of the marginal movements up and down of the resulting gender coefficients, women never appear less satisfied than men. The paradox persists: women appear as much or more satisfied as men with their jobs.
An underemphasized contribution of the Mueller and Wallace study is the inverse association found between negative affect or depression and work satisfaction. A part of the paradox of similar levels of job satisfaction among women and men lawyers involves this link between depression and job satisfaction. One key to understanding this paradox may be that the depression of women lawyers in conventional practice settings reflects an internalization of negative feelings instead of an externalization of these feelings through protests about work dissatisfaction. If this is so, it may be important that we learn more about the relationship between gender, depression, and reported job satisfaction among women and men lawyers. At a minimum, the literature on lawyers must acknowledge and incorporate measures of depression into its study of work satisfaction. Given the strong relationship found between depressed affect and work satisfaction in the research of Mueller and Wallace, failure to measure and include this variable in the literature on lawyers methodologically constitutes model misspecification and substantively risks masking important feelings that are associated with the practice of law, especially among women. This is why although the latest and otherwise most systematic multivariate research on gender and legal practice today reports that gender does not have a significant effect on lawyers' work satisfaction (Reference HeinzHeinz et al. 2005:271), further analysis may be required.
Reference HullHull's (1999) analysis of Chicago lawyer data on “the paradox of the contented female lawyer” does not point specifically to this potential role of depression we have highlighted, but it does suggest the need to consider treating job satisfaction as an independent rather than dependent variable. As well, Hull suggests that qualitative research based on interviews and observations may be necessary to gain fuller insights into the reactions of women lawyers to their work (see also Reference Dinovitzer and GarthDinovitzer & Garth 2007). Perhaps the most important feature of Hull's analysis is the attention it directs to the context and content of legal work, including the characteristics of employing organizations. Although among lawyers working in Chicago Hull finds no significant gender differences in satisfaction with work content, she does find that women lawyers are less satisfied than men lawyers with the features of the contexts in which they work. This attention to the context of the work experience may be quite important, and Hull may also be correct in suggesting that issues of causal order should be considered as well. Despite the specific finding with regard to the effect of context on overall satisfaction, Hull finds few compelling differences in indications of general satisfaction, and she too stresses that “the paradox of the contented female lawyer” remains unresolved.
Shifting attention beyond work satisfaction to the role of depression or despondency opens up new possibilities in understanding the subjective orientation of women lawyers to their work. Unlike the literature on work satisfaction, research on emotional affect consistently indicates that women report more signs of depression than men (Reference Mirowsky and RossMirowsky & Ross 1989). This research also has stimulated interesting and largely unexplored suggestions about the meanings and consequences of expressions of depression in different occupational contexts.
Perhaps most significantly for our purposes, one line of thought observes that while men tend to externalize their frustrations through anger and aggression that can lead to domination, depression, or despondency alternatively represents an internalization of frustration that can advantage men by facilitating the acquiescence and subordination of women in structured hierarchical contexts of domination. Reference BatesonBateson (2000) articulates this view in suggesting that “… just as controlled aggression makes good soldiers, mild depression may make obedient wives and daughters, resigned to circumscribed roles.” Hagan and Kay extend this line of thought from the home to the workplace, and to the entry and exit of women from law, arguing that the legal profession belatedly opened its doors to women to facilitate an expansion of the client base with a new source of labor that was perceived as both able and compliant—that is, willing to work hard while being simultaneously compliant enough to accept reduced prospects of advancement and remuneration (Reference Hagan and KayHagan & Kay 1995; Reference Hagan and ZatzHagan, Zatz, et al. 1991). This compliance is consistent with both gender equality in expressions of work satisfaction and gender inequality in symptoms of depression. Again, from this perspective the frustrations of women lawyers with their work are expressed as internalized symptoms of despondency or depression, rather than as externalized expressions of dissatisfaction about working conditions. In this important sense, feelings of depression can be an emotional price paid for occupational acquiescence (see also Reference Bielby and BielbyBielby & Bielby 1989). To the extent that acquiescence breeds acceptance, we should not expect greater expressions of work dissatisfaction among women than among men.
The Toronto Lawyers Study
The data we analyze in this article come from a three-wave panel study that began with a stratified random sample of 1,051 women and men lawyers practicing in Toronto in 1985 (and re-contacted in 1991 and 1997). The Toronto lawyers study (see Reference Hagan and KayHagan & Kay 1995) was undertaken as a mail-back survey with a sample stratified by gender and type of practice to include equally men and women in larger firms (26 or more lawyers) and smaller firms (up to 25 lawyers) as well as nonfirm settings. Twice in the 1990s, respondents were re-contacted with response rates (corrected for mortality) of 79.3 (1991) and 70.5 percent (1997), respectively. The Toronto lawyers often provided extensive written comments on their surveys, and we also conducted 50 follow-up personal interviews. The analyses presented in this article are based on 436 men and women lawyers in legal practice in Toronto at the time of the 1997 survey.
About one-quarter of Canada's lawyers practice in Toronto (Reference ArthursArthurs et al. 1986). Toronto lawyers have experienced most if not all of the changes that have transformed legal work in other North American cities (Reference Galanter and PalayGalanter & Palay 1991), including the centralization and concentration of lawyers in large firms (Reference Heinz and LaumannHeinz & Laumann 1982; Reference HeinzHeinz et al. 2005; Reference Kay and HaganKay & Hagan 1995, Reference Kay and Hagan1998; Reference NelsonNelson 1988). As Reference Galanter, Dingwall and LewisGalanter (1983) observed more than 20 years ago, big firms in the United States became a model for firms in Canada, Australia, and England. Lawyering in Toronto is in many important respects similar if not identical to the conventional practice of law in other large cities on this continent.
The demographic statistics presented in Table 1 provide an overview of the sample. We focus in this analysis on the division between private firm and other forms of legal practice because this distinction proved more revealing as well as more parsimonious than more refined categorizations involving government and solo practice. Thus the only significant effect beyond those reported below that resulted from including dummy variables for government and solo practice was an association of the latter with our measure of occupational power. This association reflected the autonomy of solo practitioners but was not otherwise illuminating for our purposes, so we opted for the simpler binary division between private firm and other forms of practice.
Table 1. Descriptive Statistics and Variable Description
* p<.05;
** p<.01;
*** p<.001;
NS=non-significant.
The stratification of the sample by gender results in a near equal balance of men and women, with only a slightly larger number of men (51 percent men/49 percent women). Although men are significantly older (47.8/44.4, p<0.001), the majority of the sample is over 40 and thus advancing into middle age. The women are significantly more likely to be in private practice (45/32 percent, p<0.01), while overall more than one-third of the sample is practicing privately (mean=0.38). A substantial proportion of our sample has moved out of private practice as they have grown older. Nearly two-thirds of the sample is working in government or in the private sector. More than 90 percent of the sample is married or cohabiting, with women significantly more likely to be divorced, widowed, or separated (13/3 percent, p<0.001). More than 80 percent of the sampled lawyers have children, although this is significantly more often the case for men (89/76 percent, p<0.001). The marriage and children variables are methodologically as well as theoretically important for our analysis because they are expected to be correlated with feelings of despondency or depression (Reference Mirowsky and RossMirowsky & Ross 1989), but not with job satisfaction. This allows us to estimate the potential reciprocal relationship between depression and job satisfaction. Meanwhile, this sample of lawyers overall seems rather conventionally middle-aged in character. Since job satisfaction is known to increase with age (Reference Alwin and KrosnickAlwin & Krosnick 1991), it is important to hold variation in this feature of our sample constant.
Two potentially important and more job-specific variables in our analysis are a structural measure of occupational power in the workplace and a social-psychological measure of perceived powerlessness. The structural measure of occupational power is based on the class analyses of Reference WrightWright (1996; Reference WrightWright et al. 1982) and asks about participation in policymaking decisions involving such issues as budgets, investments, or resources and various other operational issues. The responses rank from one to five and include no participation, an advisory role, direct participation in at least one area, some areas, and all or most areas of decisionmaking. In this middle-aged sample of lawyers, the mean level of occupational power involves participation in one or more areas of decisionmaking (mean average=3.84), with the men holding significantly more of this occupational power than the women (4.07 vs. 3.62, p<0.01).
The social psychological measure of perceived powerlessness is drawn from Reference Mirowsky and RossMirowsky and Ross's (see 1989) studies of gender and depression. They reason that respondents who feel they have less power and control over events around them will be more depressed. They ask respondents to agree or disagree with the following eight statements: I am responsible for my own success; I can do just about anything I really set my mind to; my misfortunes are a result of mistakes I have made; I am responsible for my failures; the really good things that happen to me are mostly luck; there's no sense planning a lot—if something good is going to happen it will; most of my problems are due to bad breaks; (and) I have little control over the bad things that happen to me. As would be predicted from the objective measure of occupational power, women lawyers score significantly higher on this scale of perceived powerlessness than men lawyers (17.6 vs. 16.5, p<0.01).
Much has been written about the role of childbearing in the determination of the occupational success of women lawyers (see Reference RhodeRhode 2001). We give this concern special attention in our analysis by incorporating a scale of the respondents' perceptions of the workplace consequences of having children. Respondents were asked to rank on a scale from one to five if they believed having children had caused them: loss of seniority, delay in promotion, loss of office space, pressure to return to work during parental leave, pressure to work while on parental leave, an unreasonable work load following parental leave, to test their commitment to work, loss of clients, difficulty in obtaining leave, difficulty in obtaining flexible hours or part-time work, and loss of income. There is a significant tendency of women lawyers to perceive these consequences as more pervasive than men lawyers (21.8 vs. 15.3, p<0.001).
A great deal of work has gone into the creation of survey measures of depression. The most widely used survey measure was created and concentrated into five items by the Center for Epidemiologic Studies of the National Institute of Mental Health (see Reference RadloffRadloff 1977). The items ask respondents to self-report how many days over the past week you have felt: you just couldn't get going, sad, everything was an effort, lonely, (and) you couldn't shake the blues. The average lawyer in the sample reports one of these signs of depression in the last week, while the tendency of women to report more of these symptoms than men is highly significant (1.2 vs. 0.8, p<0.001). Efforts to determine if the tendency of women to indicate more of these symptoms than men is an artifact of inclinations of women to simply report such feelings more freely than men have yielded no convincing evidence that the gender difference in scores is a methodological artifact (Reference Mirowsky and RossMirowsky & Ross 1995). We use the term despondency along with depression in this article to reflect the range from milder to more serious reporting of symptoms on this scaled measure.
Finally, three standard measures of job satisfaction (see Reference HodsonHodson 1989) are included in this analysis. They ask respondents to report on scales from one to five their overall job satisfaction, inclination to recommend the job to a friend, and likelihood of choosing the same job again. The scores were summed and divided by the number of items. Respondents on average score nearly four on this five-point scale, indicating the familiar high level of general satisfaction, with no significant difference by gender (3.93 vs. 3.89).
The Modeling Strategy
Four of the endogenous constructs introduced above that are the core of this analysis—powerlessness, child consequences, depression, and job satisfaction—all involve multiple measures that can be incorporated within a LISREL structural equation framework to minimize problems of measurement error and test assumptions about causal relationships. LISREL is a statistical program developed by Joreskog and Sorbom (see Reference HaydukHayduk 1996) for the estimation of what are more generally called covariance structure models. Such models attempt to explain relationships among sets of manifest or observed variables in terms of a typically smaller number of latent or unobserved concepts. The unobserved concepts can be thought of as a synthesis of the observed variables. The synthesis is achieved first by linking the observed variables to the unobserved concepts through the estimation of measurement or factor analytic models common in psychometrics, and second by estimating the causal relationships among the latent concepts through structural equation models associated with econometrics. A further feature of LISREL that is important for our purposes is its capacity to assess how a model that is specified in theoretical terms fits data that are observed in more than one context, for example, among women and men lawyers.
We begin by considering whether the indicators we have identified of powerlessness, child consequences, depression, and job satisfaction are more meaningfully identified as one single factor or in terms of the four distinct factors or constructs with which we have associated them. We then consider whether the latter four constructs are associated with their indicators differently among men and women. The preceding measurement issues are addressed in the Appendix (Figure A1). The results overall indicate that the four concepts are somewhat empirically distinct from one another in the full sample and when men and women are considered in separate subsamples. However, we conclude that identical measures are best used with men and women lawyers alike for all concepts in our models. We include tests for gender differences in the structural effects of these concepts in the models estimated below (see Table 2).
Table 2. Development of LISREL Model of Depression and Job Satisfaction Among Toronto Lawyers (N=436)
* p<.05;
** p<.01;
*** p<.001;
NS=non-significant.
We turn now to the structural models. We started this structural phase of our analysis by specifying what we call the pure mediation model, in which the exogenous variables—gender, age, position in private practice, being divorced/widowed/separated, and having children—are assumed to exercise their influence successively through the structural measure of occupational power, the subjective measure of perceived powerlessness, the perceived consequences of having children, recent feelings of depression, and self-reported job satisfaction. In the pure mediation model, only direct effects are specified between the successive endogenous variables that follow the influences of the exogenous variables on occupational power. That is, no “looping” or “bypassing” indirect effects are initially specified that skip around intervening variables in the interior part of the pure mediation model comprising the above ordered variables. For example, the fully mediated model allows sex, age, practice setting, marital status, and presence of children (variables to the left of occupational power on Figure 1) to impact on occupational power, but not for these factors to simultaneously impact on perceptions of power, consequences of having children, or feelings of depression or satisfaction (variables flowing to the right of occupational power in Figure 1). Modification indices and comparisons of chi-squares are next used to test both the statistical significance of adding possible indirect effects or deleting the initially specified direct effects between latent constructs in the model.
Figure 1. Structural Model of Child Consequences, Depression, and Job Satisfaction*
The structural model is estimated using LISREL, with the development of the model summarized in Table 2. A figure portraying the full measurement and structural results is presented in Figure A2 of the Appendix while a simplified structural model is presented in Figure 1. The modeling process reveals that occupational power, subjective feelings of powerlessness, and perceived consequences of having children are all involved in shaping women lawyers' feelings. The effects of gender on depression are indirect in this model. While depression in turn is linked to job satisfaction, a number of intervening links between gender and job satisfaction structure the way in which the relationship between depression and job satisfaction is gendered. A different way of saying this is that a web of relationships is involved in women lawyers internalizing feelings of depression that absorb the effects that being a woman lawyer might otherwise have on job satisfaction.
To better understand this web of relationships and intermediary processes, we begin by estimating the pure mediation model of job satisfaction. This initial model provides a baseline against which to assess the significance of adding paths of indirect influence. The pure or fully mediated model (A) actually fits the observed data fairly well (χ2=1075.15, df=471), with an adjusted goodness of fit index of 0.85, suggesting that being a woman and other variables operate through (reduced) occupational power, perceived powerlessness, perceived consequences of having children, and depressed feelings in increasing job (dis)satisfaction. However, the remaining models we examine indicate that this explanatory framework can be modified to better reflect the processes involved.
For example, a number of nonsignificant paths can be removed from the pure mediation model without significantly reducing its overall fit. Thus, when we remove the paths from age, divorce, and having children to occupational power, and the path from perceived powerlessness to perceived child consequences, the resulting model (B, χ2=1079.88, df=475) is not significantly diminished in its fit (−4.73, df=4, p>0.05) with the observed data. However, a number of indirect effects can be added to the pure mediation model to significantly improve its fit.
The first of these additions are the effects of being a woman and age, which together increase feelings of perceived powerlessness and significantly improve the fit of the model (C, χ2=1066.86, df=473, Change=13.02, df=2, p<0.01). In addition, being younger, being a woman, and having children notably increase the perceived consequences of having children (D, χ2=933.7, df=470, Change=133.15, df=3, p<0.001). In turn, being divorced and having less occupational power increase feelings of depression (Model E, χ2=908.8, df=468, Change=24.91, df=2, p<0.001). Last, being in private practice decreases job satisfaction, while perceived powerlessness also yields dissatisfaction (Model F, χ2=883.43, df=466, Change=25.37, df=2, p<0.001).
The final two models in the top panel of Table 2 test the possibility that the relationship between depression and job satisfaction is reciprocal or that the relationship is reversed, with job (dis)satisfaction leading to depression. Adding the reciprocal path in Model G (χ2=883.06, df=465, Change=0.37, df=1, p>0.10) does not significantly improve the fit of the model, while including only the opposite effect (χ2=899.56, df=466, Change=−16.13, p<0.001) significantly worsens the degree of fit. By this evidence, the causal sequence seems clearly to move from feelings of depression or despondency to job dissatisfaction. Still, there remains a further set of possibilities to be examined: namely, that the fit of our model will operate differently for women than men lawyers.
The bottom panel of Table 2 utilizes the “stacking” or “grouping” feature of LISREL that allows testing whether specific paths in a model operate differently among women and men. This second panel in Table 2 begins with a baseline “unconstrained” Model I (χ2=1642.60, df=884) that allows all the paths in our model to be estimated separately for men and women. Model J then constrains all of the paths to be equal for men and women. As might be expected, the result (χ2=1703.21, df=894, Change=−60.61, df=10, p<0.001) is a significant decline in fit with the observed data. We then examine the two paths that seem most likely from our theoretical perspective to vary in their influence between women and men. If this is so, allowing these paths to vary by gender in their effects should significantly improve the fit with the data.
The first of these paths involves the perceived consequences of having children on feelings of depression. Allowing this effect to vary between women and men significantly improves the fit of Model K (χ2=1698.46, df=893, Change=4.75, df=1, p<0.05). As expected, the perceived negative consequences of having children on feelings of depression is significantly greater among women (β=0.26, p<0.001) than men (β=0.19, p<0.05). The next model (L, χ2=1658.17, df=892, Change=40.29, df=1, p<0.001) is significantly improved by allowing a second path to vary: the effect of having children on the perceived negative consequences of having children. This effect was quite strong among women (β=0.45, P<0.001) and nonsignificant among men (β=0.12, p>0.10). We estimate a final model to determine whether the effect of depression on job satisfaction may vary between women and men (M, χ2=1657.38, df=891, Change=0.79, df=1, p>0.10), but allowing this parameter to vary for men and women does not significantly improve the model fit.
There are three major causal chains in our final model that begin with gender and reveal its indirect influence operating through occupational power, subjective (or perceived) powerlessness, and perceived consequences of having children on depressed feelings, and this, in turn, on job (dis)satisfaction (see Figure 1). The first of these chains (whose indirect effects are the product of the involved path coefficients) indicates that women have less occupational power and that reduced occupational power results in higher feelings of depression (−0.15 ×−0.18). Lower occupational power also produces feelings of powerlessness (β=−0.16), which in turn leads to reduced job satisfaction (β=−0.11). Meanwhile, a second causal chain indicates that above and apart from gender differences in occupational power, women in this sample are more likely than men to feel powerless (β=0.18), which is a further indirect path to reduced satisfaction (β=−0.11). The third causal chain indicates that women are more fearful of the occupational consequences of having children and that these perceived consequences increase feelings of depression (0.41 × 0.25), with depression in turn decreasing self-reported job satisfaction. Consistent with the above, we also find that the impact of having children on the perceived negative career consequences is unique to women (β=0.45, p<0.05; compared to men: β=0.12, p>0.05), and that the impact of these perceived consequences on feelings of depression is greater for women as well (β=0.26).
Discussion and Conclusions
The findings of this study indicate that despite the fact that women and men lawyers report reasonably similar levels of satisfaction with their work, women lawyers are substantially more likely to report feelings of depression or despondency about their lives. These feelings are significantly influenced by the circumstances in which they do their work, especially their concerns about the consequences of having children on their advancement in the practice of law. We infer from this initial finding that women are more likely to internalize feelings of despondency deriving from their work than they are to externalize these feelings through expressions of job dissatisfaction. In the still-relevant language of Reference MillsMills (1956), they are more likely to privatize than publicize their professional troubles. It seems likely that the temporal conjuncture of childbearing and partnership decisions in law firms makes this an especially salient problem for women lawyers, although this problem obviously manifests itself more broadly.
Indeed, aspects of the gendered division of internalization/externalization tendencies manifest themselves in the study of conduct disorder and aggression among disadvantaged children and adolescents. Here Reference Rosenfield, Horowitz and ScheidRosenfield (1999:212) draws on a power-control theory of gender and delinquency (Reference Hagan and SimpsonHagan, Simpson, et al. 1987) to help explain gender differences in nonnormative behavior in childhood and adolescence, noting that in traditional families daughters are controlled more than sons by their parents and that this leads girls to more restrictively internalize their expressions of distress and boys to more freely externalize their feelings of frustration. Rosenfield then summarizes a large body of research that leads to pervasive gender differences in expressions of distress:
There is evidence that boys and girls receive messages from the adult world that become increasingly divergent during adolescence. These messages correspond to splits in basic assumptions: Over time, conceptions about self-worth, control in the world, autonomy, and importance relative to others heighten for boys and decline for girls. This divergence contributes to explaining the emergence of sex differences in internalizing and externalizing problems (1999:220).
Recent work on gender differences in nonnormative behavior during the transition from adolescence to adulthood indicates that these internalizing and externalizing responses may coalesce into trajectories that can lead through emerging adulthood to longer-term adult disadvantage (Reference Hagan and FosterHagan & Foster 2003).
It is perhaps no less likely that gender differences in early socialization and in the transition from adolescence to adulthood also establish gendered patterns of response to the stressful circumstances of advantaged occupations such as lawyering. We have seen specific evidence in our findings that women lawyers respond to their lower levels of occupational power in their work with greater feelings of despondency. We have seen even greater evidence that women lawyers respond to the consequences of having children with increased feelings of despondency. Again, however, the point is that the response to these differences—in reduced occupational power and concern about the consequences of childbearing—are often internalized in the form of personal feelings of despondency rather than in public expressions of job dissatisfaction.
Overall, we have emphasized that there are at least three causal pathways through which gender indirectly is connected to job dissatisfaction and feelings of despondency. Women, controlling for other influences, report that they have less occupational power than men, and this perception is directly related to greater despondency among women than men. Perceived powerlessness, in turn, directly dampens job satisfaction. Women are also more likely to experience concerns about the consequences for their careers of having children, and these concerns result in feelings of despondency. The combined picture that results illustrates the necessity to include measures of depressed affect in studies of dissatisfaction in legal as well as other kinds of work.
Studies of satisfaction with legal practice that do not include measures of despondency are thus open to concerns about model misspecification. Beyond this, the kind of model we have developed makes the further point that the causal web that leads to dissatisfaction in lawyering is dense in its structure. An even more densely elaborated model would doubtless make the point that there are also sources of satisfaction in legal practice that help offset sources of dissatisfaction. The financial advantages that flow from legal practice, even when they involve gendered disparities, are a likely source of satisfaction for women as well as men. On the one hand, the density of this causal web—of the satisfactions and dissatisfactions of lawyering—can help us understand how and why the profession remains attractive to women as well as men. On the other hand, the sources of despondency identified in this kind of modeling approach can also help us understand how and why the profession needs to change to counteract the feelings of despondency it produces among women.
Much in our results helps explain why the legal profession has not changed nearly as much as many would have hoped. Benefits as well as costs of practice for women have produced an overall ambivalence in expressions of satisfaction and dissatisfaction that makes it easier to forestall structural change. Women lawyers' feelings of despondency likely absorb voices of distress that might otherwise give expression to protestations of work dissatisfaction. The downside of this situation is that it has allowed the profession to reap the benefits of an enlarged labor pool without accommodating the needs of its reconstituted membership. However, it should also be acknowledged that not all kinds of legal practice present all of the problems we have identified, and this has encouraged women more than men to seek careers in areas of practice, such as public interest law, that offer alternative rewards to both practitioners and the public. In this unintended way, the public may have benefited from some of the profession's gendered disparities in career opportunities and emotional rewards.
Appendix: Measurement Models
The first panel of Table A1 presents the single- and four-factor models in the combined sample of men and women lawyers. The unadjusted and adjusted goodness of fit indices and ratios of chi-square to degrees of freedom for the single and four factor models respectively both indicate that the four-factor model fits (0.81 and 0.78, χ 2/df=4.27) the observed data much better than the single factor model (0.58 and 0.51, χ 2/df=10.33). This is also true when we separate the sample into men and women lawyers, where there is a further indication that the four-factor model fits the observed data somewhat better among women (0.79 and 0.75, χ 2/df=2.71) than among men (0.71 and 0.67, χ 2/df=3.46), with the former fit among women close but not quite as good as in the combined sample.
To further explore gender differences in the fit of the measurement models for our endogenous variables, we present the measurement coefficients that result from estimating these models separately among the men and women lawyers in Appendix Figure A1. These coefficients look rather similar for the women and men in the latent construction of job satisfaction, depression, and powerlessness, while there are some more notable differences in the coefficients for child consequences. Measures of three feared consequences in particular have notably larger coefficients for the women than men lawyers, respectively: fears of losing seniority (0.68 vs. 0.15), delay in promotions (0.19 vs. 0.72), and tests of commitment (0.22 vs. 0.78). As we note further below, all of these differences are larger than 0.5 and reflect the kinds of particular concerns one may well imagine women lawyers would have about taking time away from practice during and after bearing children. The coefficients for the remaining eight measures are much closer for men and women in the profession, which suggests, perhaps surprisingly, that both men and women are concerned about how family responsibilities conflict with most demands and expectations at work.
Table A1. Development of LISREL Measurement Model of Four Latent Constructs, Child Consequences, Powerlessness, Job Satisfaction, and Depression (N=436)
Figure A1. Measurement Model of Child Consequences, Powerlessness, Depression, and Job Satisfaction
Table A2. Development of Measurement Model of Child Consequences (N=436)
Table A2 further enumerates the significance of the differences noted above in the male and female coefficients for three observed indicators of the child consequences construct. The comparison of the χ2 s for Models A (811.98) and B (1019.45) indicates (df=10, χ2=207.47, p<0.001) that constraining all measurement coefficients in the model to be equal across genders produces a significant reduction in goodness of fit. Separate comparisons of Model A with Models C, D, and E indicate that each of the following indicators—loss of seniority, delayed promotion, and test of commitment—is more strongly connected to fears of the occupational consequences of having children for women than for men lawyers. Comparison of Model B and F indicates that collectively freeing the coefficients by gender for all three indicators significantly improves the fit of the measurement of child consequences (df=3, χ2=153.85, p<0.001). Still, eight of the 11 observed indicators of child consequences are rather similar in their measurement of child consequences for men and women lawyers, and for purposes of our structural modeling below, we therefore use all 11 indicators for both women and men lawyers. We do, however, further include tests for gender differences in structural effects in the models estimated below.
Modification indices were used to selectively correlate errors among indicators in the fully elaborated structural and measurement model displayed in Figure A2. Although it is inappropriate to atheoretically correlate such errors for the sole purpose of improving model fit, we follow the a priori expectation that errors between similarly worded measures should be correlated (see also Reference Hagan and WheatonHagan & Wheaton 1993:964). Errors between indicators (i.e., see Table 1) with identical keywords (i.e., loss, pressure, difficulty) or closely overlapping meanings (e.g., seniority/promotion, couldn't get going/everything is an effort) are correlated in Figure A2. This figure constitutes the full structural and measurement model that underwrites the simplified Figure A1 with structural coefficients presented in the text.
Figure A2. Linear Structural Equation Model of Child Consequences, Depression, and Job Satisfaction*
