Content uploaded by Scott Schieman
Author content
All content in this area was uploaded by Scott Schieman on Mar 18, 2017
Content may be subject to copyright.
The Status-Health Paradox:
Organizational Context, Stress Exposure, and Well-being in the Legal Profession
Jonathan Koltai1, Scott Schieman1, and Ronit Dinovitzer1
1 University of Toronto
NOTE: This article has been accepted for publication in the Journal of Health of Social
Behavior.
Corresponding Author: Jonathan Koltai, Department of Sociology, University of Toronto, 725
Spadina Ave, Toronto, ON M5S 2J4, Canada.
E-mail: jon.koltai@mail.utoronto.ca
Running head: THE STATUS-HEALTH PARADOX
1
Abstract
Prior research evaluates the health effects of higher status attainment by analyzing highly similar
individuals whose circumstances differ after some experience a “status boost.” Advancing that
research, we assess health differences across organizational contexts among two national
samples of lawyers who were admitted to the bar in the same year in their respective countries.
We find that higher status lawyers in large firms report more depression than lower status
lawyers, poorer health in the American survey, and no health advantage in Canada. Adjusting for
income exacerbates these patterns—were it not for their higher incomes, large firm lawyers
would have a greater health disadvantage. Lastly, we identify two stressors in the legal
profession—overwork and work-life conflict—which are more prevalent in the private sector
and increase with firm size. Adjusting for these stressors explains well-being differences across
organizational contexts. This study documents the role of countervailing mechanisms in health
inequality research.
Keywords
SES and health, work and stress, work-life conflict, stress of higher status, medical sociology
Forthcoming in the Journal of Health and Social Behavior
2
The Status-Health Paradox:
Organizational Context, Stress Exposure, and Well-being in the Legal Profession
Does higher status attainment always benefit health? Decades of research has established that
individuals positioned higher in a system of stratification report better health relative to those
lower in the social hierarchy (Marmot 2004). Yet, a debate lingers on why there is a ‘consistent’
effect of attained status markers on well-being beyond thresholds of material hardship—what
Adler and colleagues (1994) label the “challenge of the gradient.” This challenge obscures two
points that have received less attention but are nevertheless relevant for theories of social
causation and the social patterning of health: (1) attaining higher status does not enhance well-
being uniformly across all contexts; and (2) the health effects of status attainment may hinge on
the particular role arrangements and circumstances associated with the attained status (Link,
Carpiano, and Weden 2013).
These caveats have sparked calls for the investigation of contexts in which circumstances
surrounding status gain offset the health benefits of higher social standing (Lutfey and Freese
2005). The stress of higher status (SHS) perspective formalizes these ideas and articulates their
implications for social patterns in stress exposure and health (Schieman and Reid 2009). Our
study tests the SHS hypothesis by documenting the ways that status attainment is related to
rewards and what Link and his colleagues (2013) call ‘health-harmful circumstances.’ Focusing
on a relatively homogenous group of professionals working in the same occupation but different
organizational contexts, we ask: does higher status attainment filter individuals into more
stressful role conditions? And, do stress exposures in these higher status roles neutralize or even
reverse the health-enhancing effects of higher status attainment? In addressing these questions,
our study seeks to refine theories about the processes that produce health inequalities.
Running head: THE STATUS-HEALTH PARADOX
3
BACKGROUND
The Status Boost
One of the main challenges in the literature on the health effects of status attainment
involves the precise disentangling of the status gain. Marmot (2004:21) identified the following
concerns related to isolating the effects of status markers on health:
One cannot do experiments; this is real life. You cannot simply assign people to
different groups. Random assignment to high education or high income,
interesting as it might be, is not an option. What we would like to have for
example, is a group of people where everyone has a high income, and where
education matters little for success. If there were then health differences in status,
we could observe whether it matters for health.
In the pursuit of the ‘real life’ situations that naturally construct the scenarios described by
Marmot, researchers have investigated the health effects of winning versus losing in status
competitions—for example, an Academy Award or a Nobel Prize (Redelmeier and Singh 2001;
Rablen and Oswald 2008). In a compelling formulation of this approach, Link and colleagues
(2013) assess the longevity outcomes of Emmy Award winners, baseball players inducted into
the Baseball Hall of Fame, and winners of the U.S. presidential election. The authors argue these
scenarios are desirable for isolating the effects of higher status attainment insofar as winners and
losers are highly alike in their pre-competition attributes. Link and colleagues (2013:197)
observe: “Among these otherwise relatively similar individuals, a large gap in relative status is
created between winners and nominated losers due to the prestige of winning.” For individuals
who won their respective competitions—a ‘status boost’—some reaped benefits in terms of
longevity (Emmy winners), while others experienced no benefits (Hall of Fame inductees). By
comparison, presidents encountered a greater mortality risk than their losing competitors.
The question turns to why becoming the president of the United States—a quintessential
status boost—has detrimental effects. Drawing from the stress of the higher status perspective,
Forthcoming in the Journal of Health and Social Behavior
4
Link et al. (2013:209) emphasize that “presidents are exposed to a very stressful job…” and that
“the important aspect is the circumstances status brings...” When presidential candidates win,
they simultaneously lose: their status gain in the role transition to president is coupled with an
avalanche of acute and chronic health-harming stressors.
More Generalizable Conditions
Analyses of high-stakes status competitions described above provide provocative
findings, but the underlying dynamics might not be broadly applicable. We draw inspiration
from Pearlin’s (1989) emphasis on ‘everyday’ stressors to argue for greater analytic precision
toward more common role strains. According to Pearlin (1983:5):
The study of how people are affected by their jobs, for example, can also inform us of the
consequences of different organizational arrangements, of the function of occupation as a
source of social status, or of the values and goals engendered by occupational experience.
Roles are thus excellent vantage points from which, if we turn in one direction, we
observe the aspects of broader social organization and, if we turn in the other, we observe
the behaviour of individuals.
Our study design embraces these ideas while also isolating the effects of the status boost
in ways that echo the approach of Link et al. (2013). To do so, we analyze data from two samples
of the legal profession: 1) Wave 2 of the After the JD study (AJD2) —a nationally representative
cohort of lawyers who entered the American legal profession in 2000; and 2) the Law and
Beyond study (LAB)—a dataset comprised of a sample drawn from the entire population of
individuals admitted to the bar in 2010 in every jurisdiction across Canada. There are three
reasons why analyses of these data afford a unique evaluation of Link and colleagues’ ideas and
the SHS hypothesis. First, both datasets include identical or highly similar measures on all study
variables. We are therefore able to replicate all analyses to strengthen confidence in our results
by distinguishing sample-specific findings from those that are robust across contexts. Second, we
address comparability issues by studying individuals who are highly similar in education, degree,
Running head: THE STATUS-HEALTH PARADOX
5
occupation, and career stage; we also hold constant factors associated with ability and selection:
law school grades, law school rank, and education-related debt. These design features help
assuage concerns about sources of unobserved heterogeneity that influence selection processes,
status achievements, and health. Third, while study participants share similar statuses prior to
graduation, at the time of data collection they subsequently sorted into positions that differ on
core dimensions of the legal profession’s system of stratification: prestige and income. Further,
status markers are themselves functions of the organizational contexts in which lawyers are
embedded. Thus, unlike studies described above that rely on inference to explain results, we
leverage data about the uneven distribution of rewards and stressors across contexts to isolate the
effects of status attainment. This is crucial—as Link and colleagues (2013:197) argue: “if a
status boost brings salutary conditions, health should be enhanced; if it brings stressful or
otherwise harmful circumstances, health should suffer; and if it brings nothing of health
relevance at all, no health effects should be observed.” We now examine a key factor that
structures the health-relevant conditions most prominent among lawyers: organizational context.
Relative and Material Status in the Legal Profession
Sociologists have long noted that the legal profession is characterized by deep structural
divides, with the large law firm sitting at the pinnacle of the prestige ladder (Smigel 1964). Large
firms hire associates exclusively from the most elite schools, and its clients are primarily large
corporations (Dinovitzer and Garth 2007). Large firm lawyers are also considered more
prestigious because their corporate (higher status) clients present work that is “purely” legal, and
does not require the filtering out of irrelevant material and nonprofessional tasks (Sandefur
2001). Thus, studies of members of the bar demonstrate that large firm lawyers tend to work in
areas of law that are accorded the highest levels of prestige (Heinz et al. 2005).
Forthcoming in the Journal of Health and Social Behavior
6
While the private practice of law is grounded on selling lawyers’ services to clients and
time billed—regardless of size—the clients served by smaller firm lawyers are primarily
individuals. As a result, these lawyers tend to engage in professionally ‘unpure’ work, such as
trials (Sandefur 2001). The ranks of smaller firms are filled with lawyers with less prestigious
educational pedigrees, who come from less elite social backgrounds and rarely cross over to the
more prestigious corporate side of the bar (Dinovitzer and Garth 2007).
Private sector lawyers are often contrasted with those in the public sector who are
government employees working within organizational settings characterized by bureaucratic
decision-making. Dixon and Seron (1995) observe that professional power is lower in the public
sector, though lawyers in this sector experience the advantages of more predictable hours. While
there is some variance depending on whether one works for federal or state/provincial/local
governments, public sector lawyers are often considered less prestigious and have lower incomes
relative to those in small private firms (Heinz et al. 2005).
That large firm lawyers work in the most prestigious setting is also buttressed by
stratification research that focuses on organizational size in the private sector, which strongly
shapes income and job-related resources. As Kalleberg and Van Buren (1996:62) assert, “the
maxim ‘bigger is better’ is true in the sense that employees of large organizations obtain higher
earnings and more fringe benefits and promotion opportunities than do employees in small
organizations.”
Collectively, the arguments and evidence presented above underscore the structural
divides in the legal profession and their links to status. Here, we wish to emphasize a major
point: while we are not able to explicitly measure prestige in our study, we will provide evidence
of an income gradient in the legal profession, with large law firm lawyers earning substantially
Running head: THE STATUS-HEALTH PARADOX
7
more income, followed by lawyers in smaller firms and with public sector lawyers earning the
least. While these incomes gaps between settings are wider in the US, descriptive research
demonstrates the same basic distribution of income in the Canadian context (Dinovitzer 2015).
The question then becomes: do these dimensions of stratification and status in the legal
profession shape the distribution of health-relevant conditions—and, ultimately, health
outcomes?
The Status-Health Paradox
While the legal profession’s organizational contexts distribute earnings and prestige
unequally, what they imply for the patterning of health remains underexplored. Given that
lawyers working in large private firms experience the highest prestige and financial remuneration
in the profession, it seems reasonable to expect more favourable health returns within this
context. But some research documents lower levels of well-being in these settings (Dinovitzer et
al. 2004)—potentially indicative of a status-health paradox within the legal profession. By
status-health paradox, we refer to a status-based distribution of health outcomes within a given
subgroup that does not conform to probabilistic statements about between-group comparisons in
the population. The identification of a status-health paradox does not undermine the premises
and established evidence that undergirds social stress theory—as Schwartz and Meyer
(2010:1112) explain, hypotheses derived from stress paradigm state that “…on average,
disadvantaged group members will fare worse than advantaged group members in health
outcomes. It is important to recognize that some subgroups of the disadvantaged group may fare
as well or even better than some subgroups of the advantaged group, and vice versa.” Following
that logic, some of this heterogeneity is likely to be non-random and thus systematically
explicable. Link and colleagues’ (2013) finding that presidential winners are confronted with a
Forthcoming in the Journal of Health and Social Behavior
8
greater mortality risk highlights this methodological issue, and also typifies a status-health
paradox.
If a status-health paradox is discovered, the next step is to identify the mechanisms that
produce it. Prominent themes in the lawyer-specific literature that emphasize extreme work hours
(overwork) and work-life conflict (WLC) suggest that the SHS hypothesis may help explain any
apparent status-health paradox among these professionals. Galanter and Palay (1994) invoke the
“tournament of lawyers” metaphor to underscore the ways that firm size reflects a structural
source of stress. Larger firms are organized around a promotion-to-partnership system: firms
hold a tournament in which the associates of a particular ‘entering class’ compete, with the prize
of partnership awarded to some fixed percentage of ‘top’ contestants. This competitive
compensation structure incentivizes large firm lawyers to exhibit maximum effort and
exceptional productivity. Because quality and quantity of productivity are difficult for partners to
measure, long hours represent a common proxy. Employees might overwork in an effort to win
the status competition and secure the economic rewards that presumably follow (Cha and
Weeden 2014).
Two concepts central to Blair-Loy’s (2003) research are also germane for our predictions:
lawyers are expected to embrace the work devotion schema and embody ideal worker norms that
compel total commitment and unrelenting availability (Wallace 1999). In a qualitative study of
lawyers, Epstein and colleagues (1999) observed that logging extreme hours (especially billable
hours) has gained a mythical status—characterized as ‘heroic’ by some partners. But there are
consequences for such heroism: overwork limits the time, energy, and attention available for
other important roles. These stressors are unevenly distributed across practice settings—they are
Running head: THE STATUS-HEALTH PARADOX
9
disproportionately experienced by lawyers in the private sector, and tend to increase with firm
size (Dinovitzer et al. 2004).
The distribution of overwork and WLC in the legal profession corresponds with
research guided by the SHS perspective that documents a positive association between these
chronic stressors and indicators of status (Schieman and Reid 2009). Chronic stressors such as
overwork and WLC may suppress the health benefits of higher social standing (Schieman and
Reid 2009), but they do not reverse the direction of the SES-health relationship in the
population because lower status individuals still face a greater cumulative and operant burden
(Turner et al. 1995). However, in the high-pressure context of the legal profession, these
stressors might overwhelm the resources they are paired with, generating a status-health
paradox. As patterns of time-related work demands across practice settings are similar in the
US and Canada (Dinovitzer 2015), adjusting for the uneven distribution of these stress
exposures should explain any status-health paradox in both national contexts.
Summary of Hypotheses
Based on the ideas and evidence outlined above, we propose the following:
The ‘Higher Status is Better for Health’ Hypothesis 1a: Lawyers in large firms
will report the lowest levels of depression and lowest risk of poor health.
Hypothesis 1b: The ‘Resources of Higher Status’ Hypothesis: Lawyers
in large firms are paid more—and higher earnings should be protective
against depression and poor health. Adjusting for the higher income of
lawyers in large firms should therefore account for their lower levels of
depression and a lower risk of poor health.
The ‘Status-Health Paradox’ Hypothesis 2a: Lawyers in large firms should
report the highest levels of depression and greatest risk of poor health.
Hypothesis 2b: The ‘Financial Consolation’ Hypothesis: Higher
earnings are protective such that were it not for their higher earnings,
lawyers in larger firms would report even higher levels of depression and a
higher risk of poor health.
Forthcoming in the Journal of Health and Social Behavior
10
Hypothesis 2c: The ‘Stress of Higher Status’ Hypothesis: (a) Lawyers
in large private firms should experience the most overwork and WLC; and
(b) Greater exposure to overwork and WLC should explain the overall
sector and firm size differences predicted in the ‘status-health paradox’
hypothesis.
DATA AND METHODS
Sample
The After the JD (AJD) study (Dinovitzer et al. 2004) is a longitudinal survey of a
nationally representative cohort of lawyers admitted to the bar in 2000. The first wave was
launched in 2002. A total of 4538 sample members responded, which is 71% of those located
and who met criteria for inclusion. The second wave (AJD2) was launched in 2007, and the
sample was comprised of Wave 1 respondents and nonrespondents. We focus on Wave 2 in the
present study because Wave 1 did not include several of our focal measures. The survey yielded
3,705 respondents meting eligibility criteria. This included 70.4% of first wave respondents and
26.9% of those who did not respond in wave 1. The overall response rate for the second wave
was 50.6% of eligible members.
The After the JD study was based on a two stage sampling design. The US was first
divided into eighteen strata determined by region and size of the new lawyer population. Within
each stratum, one primary sampling unit (PSU) was selected. These PSUs were comprised of 1)
the four “major” markets with more than 2,000 new lawyers each (Chicago, Los Angeles, New
York, and Washington, DC); 2) five “large” markets, each with between 750 and 2,000 new
lawyers (Boston, Atlanta, Houston, Minneapolis, and San Francisco); and 3) nine smaller
markets (CT, NJ remainder, FL remainder, TN, OK, IN, St Louis, UT, and OR).
Running head: THE STATUS-HEALTH PARADOX
11
The second stage sampled lawyers randomly in each of the PSUs proportionately to the
national population of US lawyers. The sampled lawyers are representative of lawyers entering
the US legal profession who graduated from law school between June 1998 and July 2000 and
gained admission to the bar in 2000. For unbiased estimates for underrepresented groups, the
survey design included an oversample of 1,465 new lawyers from minority groups (Black,
Hispanic, and Asian American). Sample weights are applied to all analyses in combination with
the “svy” command in Stata 14.1 to adjust for clustering due to the complex study design.
Next, we draw on the 2010 Law and Beyond study (LAB), which surveyed the population
of individuals admitted to the bar in 2010 in every jurisdiction in Canada. Because LAB focused
on early-lawyer careers, the design included individuals who graduated from law school after
2007 and were admitted to the bar in 2010. The LAB study was launched in September 2012
with 1,099 complete and eligible responses. After adjusting for eligibility, the final response rate
is 46%, while the cooperation rate is 79%. Analyses are weighted to better approximate the
distribution of the eligible population of 2010 admittees.
Analyses were restricted to individuals who indicated that they are a practicing lawyer in
their primary job and employed at the time of the survey; we also excluded individuals who
indicated that they were employed by a business that was not a law firm because those cases lack
information about firm size in the LAB study. We used multiple imputation by chained equations
to handle missing values on all predictor variables (White, Royston, and Wood 2011). We
included our dependent variables in the imputation models, but then employed the multiple
imputation then deletion (MID) strategy recommended by von Hippel (2007) to exclude cases
missing on the dependent variable from all analyses.1 This procedure yielded 2,576 and 877
cases for self-rated health in the U.S. and Canada respectively in our final analytic samples, and
Forthcoming in the Journal of Health and Social Behavior
12
2,488 and 863 cases for depression. We generated m=50 complete datasets using MICE. To
estimate models, each m completed dataset was analyzed separately, and then results were
combined to minimize uncertainty associated with missing information (Little and Rubin 2002).
Focal Measures
Unless otherwise noted, we use identical measures from the AJD2 and LAB datasets.
Depressive Symptoms. Depressive symptoms are measured with a 7–item version of the
Center for Epidemiological Studies Depression (CES-D) scale (Mirowsky and Ross 2003).
Respondents were asked: “How many days during the past week (0–7) have you: 1) felt you just
couldn't get going; 2) felt sad; 3) had trouble getting to sleep or staying asleep; 4) felt that
everything was an effort; 5) felt lonely; 6) felt you couldn't shake the blues; and 7) had trouble
keeping your mind on what you were doing?” We averaged the responses; higher scores indicate
more depression [α = .85 (AJD2); α = .89 (LAB)].
Self-Rated Health. This item asks: “Compared to most people your age, how would you
rate your health? Would you say your overall health is 1) Much better than most people your
own age, 2) Somewhat better, 3) About the same as most people your own age, 4) Somewhat
worse, 5) Much worse than most people your own age.” We coded responses as follows: “much
better” and “somewhat better” is coded 1; “about the same as most people your own age” is
coded 2; and “somewhat worse” and “much worse than most people your own age” is coded 3.
Practice Setting. Respondents indicated the type of organization they work for. Those in
law firms could select solo practice or private law firm. They were also asked to indicate how
many lawyers work in their firm; these were coded as: 2–20, 21–100, 101–250 and 251–plus.
The public sector includes federal government, state/provincial or local government, legal
services, duty counsel, public interest organization, other non-profit organization, or an
Running head: THE STATUS-HEALTH PARADOX
13
educational institution. We compare public sector with workers in three categories: (1)
solo/small firm (2–20); (2) medium firm (21–100); and (3) large firm (100–plus).
Income. In the AJD2, respondents were asked, “Approximately what was your total
compensation (pre-taxes) from your primary employer for the calendar year 2006 in each of the
following categories?” One item also measures personal income in the LAB data: “What was
your total compensation (pre-taxes) from your primary employer for the calendar year 2011?” In
both surveys, prompts were used for earnings from salary, bonuses, profit sharing/equity
distribution, and the present value of stock options. Answers were combined to measure
respondents’ total income. We coded income into quintiles to make group comparisons and
evaluate nonlinearities.
Overwork. One item asks: “In the last week, how many hours did you spend in each of
the following activities (if you were on vacation or sick leave last week use last week that you
worked.)” Prompts asked for hours related to working in the office or firm, working from home
on weekdays, working on the weekend, going to networking functions, participating in
recreational activities for networking purposes with other lawyers or clients. Answers were
combined to produce total hours. We coded those who worked 60 hours or more per week as 1
and all others 0.
Work-Life Conflict (WLC). Two items measure WLC. Respondents were asked, “How
often does your job interfere with each of the following?” Two statements followed: a) “Your
home or family life,” and b) Your social or leisure activities.” In the AJD2, responses were
recorded on a 4–point Likert scale ranging from 1 = “never” to 4 = “frequently.” In the LAB
data, responses were recorded on a 7–point Likert scale ranging from 1 = “never” to 7 = “very
frequently.” We averaged the items; higher scores reflect more WLC [α = .89 (AJD2); α = .92
Forthcoming in the Journal of Health and Social Behavior
14
(LAB)]. We standardized the indexes for comparability of descriptive statistics and coefficients
across contexts.
Control variables. All regression models adjust for gender, age, race, marital status, and
whether the respondent has a child at home. In addition, we also adjust for variables that may
influence practice setting and health. First, law school rank is an important predictor of future
practice setting, with those from the most prestigious law schools disproportionately entering
large law firms after graduation. If the law school experience at elite institutions entails high
competition and pressure, this stress exposure may influence lawyers’ health prior to workforce
entry. Second, we adjust for cumulative law school grade. Higher grades influence the
attainment of higher status, as prestigious law firms actively seek out high performing graduates.
On the other hand, the pressure to attain higher grades may represent a stressor that impacts
health. We adjust for law school debt because Dinovitzer (2015) observes that participants
working for larger private firms rate debt as a salient concern and a factor in their choice of
practice setting—and debt is linked with distress (Drentea and Reynolds 2014).
Plan of Analyses
We first provide descriptive patterns that represent the foundational elements of our
hypotheses: the distributions of income, overwork, and WLC across practice settings, as well as
the distribution of those two stressors across income groups. Then, in multivariate analyses, we
use OLS regression to evaluate depressive symptoms as the dependent variable. For health, we
use ordinal logistic regression. In Model 1, we establish the patterns across practice settings. This
step tests the ‘higher status is better’ versus the ‘status-health paradox’ hypotheses. Public sector
workers represent the reference category primarily because income is our focal measured
Running head: THE STATUS-HEALTH PARADOX
15
dimension of SES. This status marker is lowest in the public sector, and increases with firm size
across private sector settings. This addresses core issues relating to SES and health: if more
income leads to better health, then we should see better health in the private sector, particularly
among large firm lawyers.
In Model 2, we adjust for income to evaluate the ‘resources of higher status’ and
‘financial consolation’ hypotheses. Models 3 and 4 add overwork and WLC respectively to test
the SHS hypothesis. These progressive adjustments allow us to examine changes in coefficient
sizes across models (Mirowsky 2013). Comparing logit coefficients across models with different
covariates can be problematic because the scale of these coefficients changes with differences in
the latent residual variance across models (see Mood 2010), so we present average marginal
effects in all analyses of self-rated health, and provide two panels that predict (a) good health and
(b) poor health. All multivariate models adjust for the control variables. To preserve space,
descriptive statistics for all study variables are presented in the online supplement.2
RESULTS
Key Descriptive Patterns
Before presenting multivariate analyses, we provide a descriptive profile of our focal
variables. In the top half of Table 1, the first column demonstrates a clear income gradient across
practice settings: individuals in the public sector earn less than those in all private settings. In the
private sector, the average level of earnings rises sharply with firm size (see Panels A and B of
Figure 1).
Alongside those greater financial rewards, the patterns of stress exposure across practice
settings also reflect a gradient. As columns 2 and 3 in Table 1 indicate, overwork and WLC is
lowest in the public sector and rises sharply across private practice settings (see Panel C and D of
Forthcoming in the Journal of Health and Social Behavior
16
Figure 1). Collectively, these patterns present a mixed story about the “higher status is better”
narrative. If ‘better’ is measured by earnings, then individuals working in large private firms are
indeed better off. By contrast, if ‘better’ is measured by less overwork and WLC, then
individuals working in larger private firms are worse off.
[INSERT TABLE 1 AND FIGURES 1 AND 2 ABOUT HERE]
The bottom half of Table 1 reports the distribution of overwork and WLC across income
groups—both stressors rise sharply across income quintiles (see Panels A and B of Figure 2).
Taken together, these bivariate patterns demonstrate an intimate pairing of higher role stress and
higher status (more prestige, better earnings), and these epitomize the countervailing dynamics
articulated by the SHS perspective.
Findings for Depression
Model 1 in Tables 2 and 3 shows that, with the exception of lawyers in mid-sized firms in
the AJD2, individuals in each private sector settings report significantly higher levels of
depressive symptoms compared to those in the public sector. The inclusion of income in Model 2
amplifies these differences, revealing a suppression effect: were it not for the higher incomes in
larger firms, the difference in depression between lawyers in large firms and those in the public
sector would be even greater. Large firm lawyers are more likely to have incomes in the highest
income quintile—and those individuals, in turn, report the fewest depressive symptoms. The
same dynamic reveals a suppression effect for lawyers in mid-sized firms in the AJD2: after
accounting for income in Model 2, the coefficient for these individuals becomes significant.
The differences in depression between job settings are reduced with the inclusion of
overwork (Model 3) in both datasets. In Model 4, differences in depression across settings are
reduced to statistical non-significance when we adjust for WLC for Canadian lawyers, while
Running head: THE STATUS-HEALTH PARADOX
17
solo/small firm lawyers are the only group that remains significantly different from the public
sector in the AJD2. Both of these stressors are associated positively with depressive symptoms.
Taken together, adjustments for overwork and WLC reduce differences in depressive symptoms
across practice setting to statistical insignificance with the exception of lawyers in small practice
settings in the U.S. Thus, if overwork and WLC were distributed evenly across practice settings,
there would be almost no differences in depressive symptoms across these contexts. But, as we
have demonstrated, overwork and WLC are not evenly distributed; they are more prevalent in the
private sector and increase with firm size. Additionally, the significant association between
overwork and depression is fully accounted for with the inclusion of WLC in Model 4 in the
LAB survey, and reduced by roughly 38% in the AJD2 study. This indicates that lawyers who
overwork report more depression because they also tend to experience more WLC. Post hoc tests
(not shown) revealed no significant differences in depression between private sector settings
across all models in both national contexts.
The findings for income are also compelling. Model 2 shows that those in the top income
quintile in Canada, and those in the top three income quintiles in the U.S., report significantly
fewer depressive symptoms relative to their peers in the lowest income group. The difference
between these groups and the lowest earners widens after adjustment for overwork in Model 3.
Finally, with the inclusion of WLC in Model 4, the absolute size of the coefficient for the
comparison between the 4th income quintile and the bottom income group in Canada increases
from –.303 (Model 3) to –.486 (Model 4), and becomes statistically significant (p < .01).
Likewise, the size of the coefficient for the comparison between the 2nd income quintile and the
bottom income group in the AJD2 increases from –.155 in Model 3 to –.199, and becomes
significant (p < .05) in Model 4. These suppression effects indicate that higher overwork and
Forthcoming in the Journal of Health and Social Behavior
18
WLC among higher earning groups conceal what would be greater disparities in depression
across the income spectrum.
[INSERT TABLE 2 ABOUT HERE]
Findings for Self-Rated Health
Panels A (predicting good health) and B (predicting poor health) in Tables 4 and 5
present findings from ordered logistic regression models where the displayed coefficients
represent the average marginal effects (AMEs). In Panel A, Model 1 shows that the average
estimated probability of reporting good health is roughly 7% lower in large firms relative to the
public sector in the AJD2, and 11.5% lower in mid-sized firms relative to the public sector in the
LAB study. While these differences are significant (p < .05), all other settings do not differ from
the public sector. Additionally, post hoc tests (not shown) revealed that the probability of
reporting good health is roughly 6% lower in large firms relative to solo/small firms in the AJD2
(p < .05), while no other differences were observed between private practice settings in either
dataset across all models.
The magnitude of the AMEs across practice settings are amplified in Model 2. After
adjusting for income, the estimated probability of reporting good health for those in American
large firms is 11.3% lower compared to public sector lawyers, and 8.7% lower (post hoc test not
shown) relative to solo/small firm lawyers (p < .01). In Model 2 in the LAB data, the predicted
probability of reporting good health is roughly 15% lower for individuals in large firms
compared to the public sector (p < .01).
Although overwork (Model 3) is not related to good health, Model 4 shows that an
increase in WLC significantly decreases the probability of reporting good health in the U.S. and
Canada respectively (p < .001). The inclusion of WLC reduces the differences in the probability
Running head: THE STATUS-HEALTH PARADOX
19
of good health across all practice settings to non-significance in both datasets. These reductions
are due to the fact that WLC decreases the probability of good health and that WLC is more
prevalent among lawyers in these settings.
We observe similar patterns for poor health in Panel B. One dissimilarity is that we
observe no significant differences among practice settings in Model 1 among Canadian lawyers.
However, mirroring the patterns above, the inclusion of income in Model 2 amplifies differences
in poor health between settings, while the addition of WLC in Model 4 reduces these differences
to non-significance in both datasets. Post hoc tests (not shown) revealed that lawyers in large
firms have a greater probability of poor health compared to those in the solo/small firms; this
difference is magnified with the inclusion of income in Model 2, and then reduced to non-
significance after the addition of overwork and WLC in Models 3 and 4. No other differences in
good or poor health were observed between private practice settings.
[INSERT TABLE 3 ABOUT HERE]
Turning to income in Panel A of Tables 4 and 5, Model 2 demonstrates that being in the
top income quintile increases the probability of reporting good health by about 13% and 21%
relative to those in the lowest group in the U.S. and Canada respectively; those in the 2nd quintile
in the U.S. also have a higher probability of good health compared to the lowest earners. The
inclusion of WLC in Model 4 reveals a suppression effect: the probabilities of reporting good
health for individuals in the 3rd and 4th income quintiles in the AJD2, and in the 4th income
quintile in the LAB study, become significantly higher compared to the lowest earners. These
suppression patterns are evident because WLC—which is more prevalent among higher
earners—decreases the probability of reporting good health (p < .001).
Forthcoming in the Journal of Health and Social Behavior
20
In Panel B of Tables 4 and 5 the income patterns mirror those above. Model 2
demonstrates that being in the top income quintile reduces the probability of reporting poor
health by 4.5% and 10.2% relative to those in the public sector in the U.S. and Canada
respectively. The inclusion of overwork in Model 3 has no effect, but adjustment for WLC in
Model 4 reveals a suppression effect—the probabilities of reporting poor health for individuals
in the 3rd income quintile in the AJD2, and lawyers in the 4th quintile in the LAB study, become
significantly lower compared to the bottom earners. These results suggest that the health benefits
of higher income are concealed by higher levels of WLC among the highest earning lawyers.
[INSERT TABLES 4 AND 5 ABOUT HERE]
DISCUSSION
A relatively straightforward narrative characterizes much prior theorizing and research on
social stress: individuals positioned higher in a social hierarchy should experience better health.
We demonstrate the ways that this narrative gets complicated when status-based resources are
offset by stress exposures across two national contexts. First, we observe a status-health paradox:
higher status lawyers have a mental health disadvantage relative to their peers in the public
sector, and are no better off in terms of health. Second, the interplay between practice setting and
income adds complexity: adjusting for income, health disparities across practice settings become
even greater—were it not for higher incomes in larger firms, the health gap between the public
sector and larger firms would be even larger. Taken together, these patterns support the status-
health paradox and the financial consolation hypotheses.
While the ‘higher status is better for health’ hypothesis is not supported, we observed that
higher earners have fewer depressive symptoms and a lower risk of poor health. As individuals
in larger firms tend to have higher earnings, the ‘resources of higher status’ claim is supported
Running head: THE STATUS-HEALTH PARADOX
21
with respect to practice setting and income. Collectively, these cross-cutting patterns produce a
suppression effect: lawyers in large firms tend to earn more; earnings are protective for health;
holding earnings constant reveals greater health differences.
We also contribute to evaluations of the SHS hypothesis. The hypothesis predicts and our
observations confirm three interrelated findings: (1) Lawyers in the private sector and in
progressively larger firms experience higher levels of overwork and WLC; (2) Overwork and
WLC are associated positively with depressive symptoms and risk of poor health; and (3) When
our models include these stressors (especially WLC), we fully explain differences in depression
and health across organizational contexts.
If health outcomes are unevenly distributed across organizational contexts among
otherwise similar individuals, what is it about these settings that generate such patterns?
Informed by the sociological study of stress, our focus was drawn to conditions most relevant to
role experiences of lawyers. The nature of the work role and concomitant responsibilities reflect
the organizational settings in which lawyers are embedded—and this has important implications
for stress exposure and health. These discoveries resonate with what Wheaton and Gotlieb
(1997) call the “power of the ordinary,” and the “cumulative effects of structural givens in daily
life.” Structural givens in the legal profession—overwork and WLC—vary systematically across
practice settings. The ‘power of the ordinary’ characterizes the potency of chronic stressors in
everyday role arrangements. WLC is ordinary for lawyers in large private firms—and this is why
it is also powerful for the social patterning of depression and health.
While these observations describe individual-level experiences, they also reflect meso
and macro-level concerns at the intersection of social stratification, medical sociology, and
occupational health psychology. As Fenwick and Tausig (1994:268) explain: “Stressful jobs are
Forthcoming in the Journal of Health and Social Behavior
22
not randomly distributed throughout the economy; rather, they are products of macroeconomic
structures and forces such as the economic sector and organizational structure of firms in which
the jobs are located…” Our findings reinforce the claim that organizational contexts shape not
only material and relative status, but also the job conditions to which individuals are exposed,
and are thus logically prior to demands, resources, and health (Tausig and Fenwick 2011). These
processes make the study of stress a quintessentially sociological endeavour (Pearlin 1989).
Our findings have broader implications for social stress theory and health inequality
research. While on the surface our results might seem to challenge established theory and
evidence based on the stress process paradigm, we underscore that our findings do not
undermine observations about the status-health relationship in the population because, on
average, lower status individuals still disproportionately experience higher rates of physical and
mental illness. Lutfey and Freese (2005) articulate this methodological issue and its implications
for fundamental cause theory. Describing ‘countervailing mechanisms’ that may put higher
status individuals at greater risk for poor health, Lutfey and Freese (2005:1365) observe that
“[f]undamental relationships do not require that all of the pathways between X and Y support the
relationship. Countervailing mechanisms may work in the other direction; indeed, the only
requirement is that the effects of such mechanisms are cumulatively smaller than the
mechanisms producing the fundamental relationship.”
As long the stressors of higher status do not outweigh the total stress exposure
encountered by lower status individuals, observations concerning the status-based distribution of
health outcomes in the population should remain consistent. Yet, social stress researchers should
be cognizant of shifts in the nature of work and the structure of the workforce over time, and the
implications such changes have for patterns of stress exposure and the status-health gradient. In
Running head: THE STATUS-HEALTH PARADOX
23
her review of new forms of work, Smith (1997) observed trends toward destabilization and work
intensification for professional and managerial workers in recent decades. Destabilization is
reflected in rising job insecurity among these groups (Kalleberg 2009), and work intensification
is manifest in research demonstrating that higher SES workers disproportionately encounter
chronic stressors like job pressure, overwork, role blurring, and work-life conflict (Schieman and
Koltai 2016). When coupled with an increasing proportion of professionals and managers in the
labour force (Esping-Andersen 2002), these temporal trends and current patterns may influence
the shape of the SES-health gradient—especially through the middle and upper range of the
status spectrum. Future research should thus probe for similar status-health paradoxes in other
high-commitment professions.
Our findings also have several practical implications. First, the alleviation of health
disparities represents a core and unifying objective in research concerned with the social
determinants of health. If this goal is to be applied to the legal profession in North America, then
our analyses suggest that policy or organizational interventions should focus on the mitigation of
WLC. 4 That is, were it not for the uneven distribution of WLC in the legal profession, there
would be no differences in distress or poor health across practice settings. Next, consideration of
lawyer well-being would benefit law firms themselves. In its early incarnation, law firms’
promotion to partner tournament system required an “up or out” process in which lawyers who
were not promoted to partner were let go (Galanter and Palay 1994). Attrition was necessary—if
lawyers left the firm because of stress, this created the turnover firms required without any action
on their part. Yet the traditional model has changed rapidly in recent years. Law firms continue
to promote a small number of associates to partnership, but many are now employed in non-
partnership track positions (Galanter and Henderson 2008). Simultaneously, there is increased
Forthcoming in the Journal of Health and Social Behavior
24
demand and competition for these same lawyers from new legal services providers who are
offering more flexible workplaces and options to minimize WLC (Williams, Platt and Lee 2015).
Thus, if law firms seek to remain competitive in attracting and retaining associates, they will
increasingly need to design work environments that reduce inter-role conflict.
One final point of reflection involves comparisons across settings, and how these square
with our hypotheses. We did not observe statistically significant differences between private
practice settings for depression in either dataset. That is, while public sector lawyers tend to
report less depression relative to their higher status peers, no differences emerged between large,
medium, and solo/small firms. Given the gradient in chronic stress exposure across practice
settings, we find this result surprising. But we do not believe that this lack of differences
undermines our general conclusions—we still observe a status-health paradox in all analyses
insofar as higher status lawyers are either worse off or no better off in terms depression relative
to those at the bottom of the status spectrum. Where differences in depression are observed
between public sector lawyers and those in higher status contexts, these differences can be
explained by SHS dynamics. That said, we do observe a distribution of health between practice
settings in the U.S. that better approximates a gradient—large firm lawyers report a lower
probability of good health and a higher probability of poor health relative to those in the public
sector and those in solo practices and small firms.
Study Limitations
We acknowledge several potential limitations of our study. First, the cross-sectional
nature limits claims about causality. We cannot rule out the possibility that individuals with
higher pre-existing levels of poor health selected into larger, higher status firms. Alternatively,
individuals with poor health might select less stressful work settings to manage pre-existing
Running head: THE STATUS-HEALTH PARADOX
25
conditions. Existing evidence, however, has shown that poorer physical and mental health either
impair higher status attainment or cause individuals to ‘drift’ downward (not upward) in the
social hierarchy (Haas 2006).
Next, while our design and logic echoes that of Link et al. (2013), we recognize that the
status boost in our study cannot be assumed to be completely random; it is possible that
individuals with certain traits or personality factors—such as ability or neuroticism—select into
higher status firms. Selection bias may therefore complicate a causal interpretation of our
findings if these pre-existing differences are the underlying cause of both higher status
attainment and poor health outcomes. While certainly plausible, the pathways required for such
traits to represent true confounders are not necessarily established empirically. For example,
neurotic individuals experience more distress (Bienvenu et al. 2004), but a meta-analysis
documents that neuroticism is negatively correlated with salary and promotions (Ng et al. 2005).
In a similar vein, cognitive ability is positively related to higher status attainment, but ability is
generally related to better health outcomes (Singh-Manoux et al. 2005)—although Link et al.
(2008) found no significant association between cognitive abilities and health net of SES. Our
models also control for two key indicators of ability and practice setting selection: law school
grades and law school rank. Nevertheless, we cannot definitively rule out that selection factors in
the present study.
We also recognize the potential for alternative mechanisms that may explain our
observations. While large organizations provide higher wages, benefits, and prestige, they tend to
offer less autonomy (Kalleberg and Vanburen 1996). Control over the labour process, in turn, is
inversely related to distress and poor health (Tausig and Fenwick 2011). Social justice work
tends to be lower paying (Dinovitzer et al. 2004), and social responsibility and meaningful
Forthcoming in the Journal of Health and Social Behavior
26
activity are thought to be core elements of positive human health (Ryff and Singer 1998). Thus,
lawyers in less prestigious settings might find their work more rewarding, meaningful, or
autonomous—and these factors might contribute to their lower distress and poor health. From a
statistical standpoint, however, the existence of an alternative mechanism does not necessarily
negate the explanatory role of WLC and overwork. To do so, the putative alternative mechanism
would need to 1) be affected by the exposure (practice setting), and 2) confound the association
between the intervening variable and the outcome (VanderWeele 2015).
While there are remarkable similarities across national contexts—especially in the
distribution of income and stress exposure—subtle differences between Canada and the US
remain unexplained. For example, we observe a health disadvantage for large firm lawyers
relative to the public sector and those in solo/small firms in the US but not in Canada. We offer
two points of speculation: first, these differences could reflect period effects—the US data were
collected during the beginning of the Great Recession, and research has documented that this
was a tumultuous time for large firms (Wald 2010). The greater health disadvantage for large
firm lawyers in the US might therefore reflect the added stress of job insecurity. Second, because
lawyers in the U.S. sample have been practicing for a longer period, the health differences may
be attributable to cohort effects—extreme hours and WLC may lead to greater depression in the
short term, but it may take longer for these stressors to initiate diverging health trajectories.
CONCLUSION
Our inquiry began through our interest in understanding why a material or relative status
boost improves the health of some while it harms the health of others. Previous studies focused
on winners versus losers in various ‘status competitions,’ and produced mixed findings. Link et
al. (2013) argued that the effects of a status boost depend on the conditions in the newly acquired
Running head: THE STATUS-HEALTH PARADOX
27
role. We advanced prior studies by analyzing data that mimic such rare circumstances. Unlike
prior studies, our data contain information about a cohort of individuals whose work lives are
generalizable to other high-commitment professional groups, as well as information about
stressors embedded in their work environments. These design characteristics enabled this study
to offer a significant contribution to theories surrounding social causation: we identify a set of
patterns that run contrary to our traditional expectations about the relationship between SES and
health, and we provide an explanation for this deviation.
Acknowledgements: Previous versions of this manuscript were presented at the 2016
International Conference on Social Stress Research in San Diego, CA, and the 2016 American
Sociological Association annual meeting in Seattle, WA. The authors would like to thank Blair
Wheaton, Markus Schafer, Geoffrey Wodtke, participants in the Toronto Inequality Workshop,
and several anonymous reviewers for their incredibly helpful feedback on earlier versions of this
paper. This research was supported by the Social Sciences and Humanities Research Council of
Canada (Grant No. 410-2009-2778), the American Bar Foundation, National Science Foundation
(Grant No. SES0115521 and SES0550605), Access Group, Law School Admission Council,
National Association for Law Placement, National Conference of Bar Examiners, and the Open
Society Institute. The views and conclusions stated herein are those of the authors and do not
necessarily reflect the views of individuals or organizations associated with the After the JD
study.
Forthcoming in the Journal of Health and Social Behavior
28
ENDNOTES
1. This procedure resulted in a rate of missingness between roughly 20% and 32%
depending on the response variable and dataset. We offer several points of consideration
about these rates. First, simulation studies have shown that multiple imputation (MI)
strategies perform well even under circumstances with substantially higher proportions of
missingness (Jansen et al. 2010), and employing multiple imputation without deletion (ie.
not excluding imputed values on the dependent variables) yielded the same substantive
findings in this study. We employ the multiple imputation then deletion (MID) approach
because, as von Hippel (2007: 83) explains, “[w]hen there is something wrong with the
imputed Y values, MID protects the estimates from the problematic imputations. And
when the imputed Y values are acceptable, MID usually offers somewhat more efficient
estimates than an ordinary MI strategy.” Next, MI assumes that missing values are
missing at random (MAR), and not missing completely at random (MCAR). This is an
important distinction because the MAR assumption states that missing data can be
considered random conditional on other respondent characteristics that determined their
missingness (Donders et al. 2006). As Schafer (1997: 27) explains “The crucial
assumption here “is not that the propensity to respond is completely unrelated to the
missing data, but that this relationship can be explained by data that are observed.”
Additionally, while the MAR assumption is unlikely to be precisely satisfied in most
cases (Dong and Peng 2013), departures from MAR are not large enough to invalidate the
results of an MAR-based analysis in many realistic applications (Collins et al. 2001;
Schafer and Graham 2002).Thus, because we are dealing with a relative homogenous sub
group, and because a comprehensive range of known predictors of both outcomes are
included in our imputation model, we are reasonably confident that violations of the
MAR assumption are unlikely to produce serious bias in our findings.
2. We tested interactions between gender and practice setting in all models in both surveys,
and found no evidence that the effects of practice setting on health or depression differ by
gender—these results are therefore not presented.
3. We are mindful of the discrepancy between this approach and what is recommended by
fundamental cause theory. A traditional fundamental cause approach looks to upstream
factors (ie. inequalities in SES) in order to alleviate disparities in morbidity and mortality
in the population. The theory argues that policy and interventions that overemphasize
intervening mechanisms are flawed insofar as these mechanisms will be replaced with
others that reproduce the inequality. These important insights lead to an obvious
conundrum when applied to our findings: alleviating health disparities in the legal
profession through policy focused on upstream factors might lead to the conclusion that
large firm lawyers should be paid even more in order to offset the harmful effects of
Running head: THE STATUS-HEALTH PARADOX
29
overwork and WLC. We are thus left with policy recommendations that would
exacerbate social inequality. On the other hand, our recommendations may be seen as
consistent with the fundamental cause approach insofar as we have contextualized health-
harmful mechanisms and identified the social conditions (ie. organizational contexts) that
put individuals in the legal profession “at risk of risk” (Link and Phelan 1995). Given
these points, we view a focus on intervening mechanisms to be reasonable in the context
of the legal profession.
REFERENCES
Adler, Nancy E., Thomas Boyce, Margaret A. Chesney, Sheldon Cohen, Susan Folkman, Robert
L. Kahn, and S. Leonard Syme. 1994. “Socioeconomic Status and Health: The Challenge
of the Gradient.” American Psychologist 49(1):15-24.
Blair-Loy, Mary. 2003. Competing Devotions. Cambridge, MA: Harvard University Press.
Bienvenu, O. Joseph, Jack F. Samuels, Paul T. Costa, Irving M. Reti, William W. Eaton, and
Gerald Nestadt. 2004. “Anxiety and Depressive Disorders and the Five Factor Model of
Personality: A Higher and Lower Order Personality Trait Investigation in a Community
Sample.” Depression and Anxiety 20(2):92-97.
Cha, Youngjoo and Kim A. Weeden. 2014. “Overwork and the Slow Convergence in the Gender
Gap in Wages.” American Sociological Review 79(3):457-484.
Collins, Linda M., Joseph L. Schafer, and Chi-Ming Kam. 2001. “A Comparison of
Inclusive and Restrictive Strategies in Modern Missing Data Procedures.” Psychological
Methods 6(4): 330.
Dinovitzer, Ronit. 2015. Law and Beyond: A National Study of Canadian Law Graduates.
Rochester, NY: Social Science Research Network. Retrieved March 5, 2016
http://papers.ssrn.com/abstract=2615062.
Dinovitzer, Ronit, Garth, B. G., Sander, R., Sterling, J., and Wilder, G. Z. 2004. After the J.D.:
First Results of a National Study of Legal Careers. Chicago, IL:NALP
Dinovitzer, Ronit and Bryant G. Garth. 2007. “Lawyer Satisfaction in the Process of Structuring
Legal Careers.” Law & Society Review 41(1):1–50.
Donders, A. Rogier T., Geert JMG van der Heijden, Theo Stijnen, and Karel GM Moons.
“Review: a gentle introduction to imputation of missing values.” 2006. Journal of
Clinical Epidemiology 59(10): 1087-1091.
Forthcoming in the Journal of Health and Social Behavior
30
Dong, Yiran, and Chao-Ying Joanne Peng. 2013. “Principled missing data methods for
Researchers.” SpringerPlus 2(1).
Drentea, Patricia, and John R. Reynolds. 2014. “Where Does Debt Fit in the Stress Process
Model?" Society and Mental Health 5(1):16-32
Dixon, Jo and Carroll Seron. 1995. “Stratification in the Legal Profession: Sex, Sector, and
Salary.” Law and Society Review 29(3):381-412.
Epstein, Cynthia Fuchs, Carroll Seron, Bonnie Oglensky, and Robert Saute. 1999. The Part-Time
Paradox. New York: Routledge.
Fenwick, Rudy and Mark Tausig. 1994. “The Macroeconomic Context of Job Stress.” Journal of
Health and Social Behavior 35(3):266–282.
Galanter, Marc and William Henderson. 2008. “The Elastic Tournament: A Second
Transformation of the Big Law Firm.” Stanford Law Review 60(6):1867–1929.
Galanter, M. and Palay, T., 1994. Tournament of Lawyers: The Transformation of the Big Law
Firm. University of Chicago Press.
Haas, S.A. 2006. “Health Selection and the Process of Social Stratification: The Effect of
Childhood Health on Socioeconomic Attainment.” Journal of Health and Social
Behavior 47(4):339-354.
Heinz, J. P., Nelson, R. L., Sandefur, R. L., & Laumann, E. O. 2005. Urban lawyers: The New
Social Structure of the Bar. Chicago: University of Chicago Press
Janssen, Kristel JM, A. Rogier T. Donders, Frank E. Harrell, Yvonne Vergouwe, Qingxia
Chen, Diederick E. Grobbee, and Karel GM Moons. 2010. “Missing Covariate Data in
Medical research: to Impute is Better than to Ignore.” Journal of Clinical Epidemiology
63(7): 721-727.
Kalleberg, Arne L. and Mark E. Van Buren. 1996. “Is Bigger Better? Explaining the
Relationship between Organization Size and Job Rewards.” American Sociological
Review 61(1):47–66.
Kalleberg, A.L., 2009. Precarious Work, Insecure Workers: Employment Relations in
Transition.” American Sociological Review 74(1):1-22.
Link, Bruce G., and Jo Phelan. 1995. “Social conditions as fundamental causes of disease.”
Journal of Health and Social Behavior 35(Extra Issue): 80-94.
Link, Bruce G., Richard M. Carpiano, and Margaret M. Weden. 2013. “Can Honorific Awards
Give Us Clues about the Connection between Socioeconomic Status and Mortality?”
American Sociological Review 78(2):192–212.
Link, Bruce G., Jo C. Phelan, Richard Miech, and Emily Leckman Westin. 2008. “The
Resources that Matter: Fundamental Social Causes of Health Disparities and the
Challenge of Intelligence.” Journal of Health and Social Behavior 49(1):72-91.
Running head: THE STATUS-HEALTH PARADOX
31
Little, Roderick J. A. and Donald B. Rubin. 2002. Statistical Analysis with Missing Data.
Hoboken, NJ: Wiley-Interscience.
Lutfey, Karen and Jeremy Freese. 2005. “Toward Some Fundamentals of Fundamental
Causality: Socioeconomic Status and Health in the Routine Clinic Visit for diabetes1.”
American Journal of Sociology 110(5):1326–72.
Mirowsky, John. 2013. “Analyzing Associations between Mental Health and Social
Circumstances.” Pp. 143–65 in Handbook of the Sociology of Mental Health, edited by C.
S. Aneshensel, J. C. Phelan, and A. Bierman. New York: Springer.
Mirowsky, John and Catherine E. Ross. 2003. Social Causes of Psychological Distress.
Piscataway, NJ: Transaction Publishers.
Mood, Carina. 2010 “Logistic regression: Why We Cannot Do What We Think We Can Do, and
What We Can Do About It.” European Sociological Review 26(1):67-82.
Ng, Thomas WH, Lillian T. Eby, Kelly L. Sorensen, and Daniel C. Feldman. 2005. “Predictors
of Objective and Subjective Career Success: A Meta‐Analysis.” Personnel Psychology
58(2):367-408.
Pearlin, L. I. (1983). Role strains and personal stress. In H. B. Kaplan (Ed.), Psychosocial stress:
Trends in Theory and Research (pp. 3-32). New York: Academic Press.
Pearlin, Leonard I. 1989. “The Sociological Study of Stress.” Journal of Health and Social
Behavior 30(3):241–256.
Rablen, Matthew D. and Andrew J. Oswald. 2008. “Mortality and Immortality: The Nobel Prize
as an Experiment into the Effect of Status upon Longevity.” Journal of Health Economics
27(6):1462–1471.
Redelmeier, Donald A. and Sheldon M. Singh. 2001. “Survival in Academy Award–winning
Actors and Actresses.” Annals of Internal Medicine 134(10):955–62.
Ryff, Carol D., and Burton Singer. 1998. “The contours of positive human health.”
Psychological Inquiry 9(1):1-28.
Sandefur, Rebecca L. 2001. “Work and Honor in the Law: Prestige and the Division of Lawyers’
Labor.” American Sociological Review 66(3): 382–403.
Schafer, Joseph L. 1997. Analysis of Incomplete Multivariate Data. London: Chapman & Hall.
Schafer, Joseph L., and John W. Graham. 2002. “Missing Data: Our view of the State of the
Art.” Psychological Methods 7(2): 147.
Schieman, Scott, and Koltai, Jonathan. 2016. “Discovering Pockets of Complexity:
Socioeconomic Status, Stress Exposure, and the Nuances of the Health Gradient. Social
Science Research.
Forthcoming in the Journal of Health and Social Behavior
32
Schieman, Scott and Sarah Reid. 2009. “Job Authority and Health: Unraveling the Competing
Suppression and Explanatory Influences.” Social Science and Medicine 69(11):1616–
1624.
Schwartz, S. and Meyer, I.H., 2010. “Mental Health Disparities Research: The Impact of Within
and Between Group Analyses on Tests of Social Stress Hypotheses.” Social Science and
Medicine 70(8):1111-1118.
Smith, V., 1997. New Forms of Work Organization. Annual Review of Sociology 23(1):315-
339.
Tausig, Mark and Rudy Fenwick. 2011. Work and Mental Health in Social Context. New York:
Springer.
VanderWeele, Tyler. 2015. Explanation in Causal Inference: Methods for Mediation and
Interaction. Oxford University Press.
von Hippel, Paul T. 2007. “Regression with Missing Ys: An Improved Strategy for Analyzing
Multiply Imputed Data.” Sociological Methodology 37:83–117.
Wald, Eli. 2010. “The Great Recession and the Legal Profession” Fordham Law Review
78(5):2051-2066.
Wallace, Jean E. 1999. “Work-to-Nonwork Conflict among Married Male and Female Lawyers.”
Journal of Organizational Behavior 20(6):797–816.
Wheaton, Blair. and I.H. Gotlib. 1997. “Trajectories and Turning Points over the Life Course:
Concepts and Themes.” Pp. 1–25 in Stress and Adversity over the Life Course, edited by
I.H. Gotlib and Blair Wheaton. Cambridge, England: Cambridge University Press.
White, Ian R., Patrick Royston, and Angela M. Wood. 2011. “Multiple Imputation Using
Chained Equations: Issues and Guidance for Practice.” Statistics in Medicine 30(4):377-
399.
Williams JC, Platt A, Lee J. 2015. “Disruptive Innovation: New Models of Legal Practice.” San
Francisco: UC Hastings College Law, Centre for WorkLife Law.
http://worklifelaw.org/new-models-report/.
Running head: THE STATUS-HEALTH PARADOX
33
Author Biographies
Jonathan Koltai is a doctoral candidate in the department of sociology at the University of
Toronto. His research interests broadly focus on the physical and mental health consequences of
social inequality. His dissertation examines the ways that social stratification and organizational
contexts shape the relationship between workplace conditions and individual well-being, and
how these processes change over time.
Scott Schieman is Professor and Canada Research Chair in the Department of Sociology at the
University of Toronto. His research focuses on the social psychology of inequality and its
relationship to health outcomes. He is the lead investigator of the Canadian Work, Stress, and
Health study (CANWSH), a national longitudinal study of workers.
Ronit Dinovitzer is Associate Professor of Sociology at the University of Toronto. She is also a
Faculty Fellow at the American Bar Foundation in Chicago, where she is Co-Director of the
Research Group on Legal Diversity, and Affiliated Faculty in Harvard's Program on the Legal
Profession. Ronit publishes and teaches widely on the social structure of the legal profession and
on professional ethics, and has been a lead investigator on studies of lawyer careers in the United
States and Canada.
Forthcoming in the Journal of Health and Social Behavior
34
FIGURE 1. Economic Rewards and Stressors across Practice Setting
Panel A. Average Levels of Income across Practice Setting in the U.S.
Panel B. Average Levels of Income across Practice Setting in Canada
40000
60000
80000
100000
120000
140000
160000
180000
Public Sector Solo/Small Firm Mid-Size Firm Large Firm
Average Total Income
Practice Setting
50000
60000
70000
80000
90000
100000
110000
Public Sector Solo/Small Firm Mid-Sized Firm Large Firm
Average Total Income
Practice Setting
Running head: THE STATUS-HEALTH PARADOX
35
Panel C. Overwork (Proportion Working 60+ hours) across Practice Setting
Panel D. Average Level of Work-life Conflict across Practice Setting
-0.05
0.05
0.15
0.25
0.35
0.45
0.55
Public Sector Solo/Small
Firm
Mid-Size Firm Large Firm
Overwork
Practice Setting
US
Canada
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Public Sector Solo/Small
Firm
Mid-Size Firm Large Firm
Work-Life Conflict
Practice Setting
US
Canada
Forthcoming in the Journal of Health and Social Behavior
36
FIGURE 2. Levels of Stress Exposure across Income Level
Panel A. Overwork (Proportion Working 60+ hours) across Income Levels
Panel B. Average Level of Work-life Conflict across Income Levels
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
12345
Work-Life Conflict
Income Quintile
US
Canada
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
12345
Overwork
Income Quintile
US
Canada
Running head: THE STATUS-HEALTH PARADOX
37
TABLE 1. Descriptive Patterns among Focal Variables
Note: Means and proportions are weighted combined estimates from fifty multiple imputation datasets.
Income
(mean)
Overwork
(proportion)
WLC
(mean)
US
CAN
US
CAN
US
CAN
Practice Setting
Public Sector
$78,402
$70,392
.210
.166
-.385
-.542
Solo Practice / Small firm
$102,826
$71,791
.326
.297
-.029
-.065
Mid-size Firm
$128,430
$83,989
.364
.317
.285
.135
Large Firm
$180,193
$101,968
.487
.480
.590
.588
Income
Income Quintile 1
$49,170
$44,412
.262
.200
-.325
-.373
Income Quintile 2
$76,735
$64,818
.256
.257
-.161
-.226
Income Quintile 3
$98,531
$77,577
.314
.311
.043
-.033
Income Quintile 4
$ 133,771
$91,781
.374
.272
.245
.199
Income Quintile 5
$240,230
$115,164
.482
.474
.506
.401
Forthcoming in the Journal of Health and Social Behavior
38
TABLE 2. Depression Regressed on Practice Setting, Income, and Stressors in the U.S. (N=2,488)
Notes: Coefficients and standard errors (in parentheses) are weighted combined estimates from fifty multiple
imputation datasets. Models include all control variables.
a Compared to public sector lawyers
b Compared to income quintile 1
* p < .05; ** p < .01; *** p < .001 (two–tailed test).
Model 1
Model 2
Model 3
Model 4
Job Setting
Solo / Small Firm a
.242*
(.09)
.311**
(.10)
.279**
(.10)
.199*
(.09)
Mid-size Firm a
.210
(.11)
.340**
(.12)
.294**
(.12)
.143
(.12)
Large Firm a
.230**
(.09)
.404***
(.10)
.320**
(.10)
.102
(.11)
Income
Quintile 2 b
—
—
-.168
(.10)
-.155
(.10)
-.199*
(.09)
Quintile 3 b
—
—
-.232*
(.12)
-.231*
(.11)
-.303**
(.11)
Quintile 4 b
—
—
-.406 ***
(.11)
-.412***
(.11)
-.504***
(.10)
Quintile 5 b
—
—
-.343**
(.13)
-.363**
(.12)
-.477***
(.12)
Time Demands
Overwork
—
—
—
—
.365***
(.07)
.228**
(.07)
WLC
—
—
—
—
—
—
.343***
(.04)
Constant
1.234**
(.44)
1.250**
(.45)
1.154***
(.44)
1.681***
(.43)
Running head: THE STATUS-HEALTH PARADOX
39
TABLE 3. Depression Regressed on Practice Setting, Income, and Stressors in Canada (N=863)
Notes: Coefficients and standard errors (in parentheses) are weighted combined estimates from fifty multiple
imputation datasets. Models include all control variables.
a Compared to public sector lawyers
b Compared to income quintile 1
* p < .05; ** p < .01; *** p < .001 (two–tailed test).
Model 1
Model 2
Model 3
Model 4
Job Setting
Solo / Small Firm a
.360**
(.12)
.374**
(.13)
.335**
(.13)
.134
(.13)
Mid-size Firm a
.407*
(.16)
.473**
(.17)
.433**
(.17)
.180
(.16)
Large Firm a
.428**
(.16)
.625**
(.18)
.557**
(.19)
.146
(.19)
Income
Quintile 2 b
—
—
-.097
(.17)
-.111
(.17)
-.156
(.17)
Quintile 3 b
—
—
-.150
(.18)
-.177
(.18)
-.282
(.17)
Quintile 4 b
—
—
-.289
(.19)
-.303
(.18)
-.486**
(.18)
Quintile 5 b
—
—
-.499**
(.18)
-.547**
(.18)
-.683***
(.17)
Time Demands
Overwork
—
—
—
—
.297*
(.12)
.050
(.12)
WLC
—
—
—
—
—
—
.460***
(.05)
Constant
.642
(.48)
.649
(.50)
.653
(.50)
.963
(.49)
Forthcoming in the Journal of Health and Social Behavior
40
TABLE 4. Self-Rated Health Regressed on Practice Setting, Income, and Stressors in the U.S. (2,576)
Panel A. Predicting Probability of Good Health
Panel B. Predicting Probability of Poor Health
Note: Coefficients (AMEs) and standard errors (in parentheses) are weighted combined estimates from fifty
multiple imputation datasets. Models include all control variables.
a Compared to public sector lawyers
b Compared to income quintile 1
Model 1
Model 2
Model 3
Model 4
Job Setting
Solo / Small Firm a
-.007
(.03)
-.026
(.03)
-.022
(.03)
.001
(.03)
Mid-size Firm a
-.016
(.05)
-.048
(.05)
-.043
(.04)
-.015
(.06)
Large Firm a
-.067*
(.03)
-.113**
(.04)
-.103*
(.04)
-.064
(.04)
Income
Quintile 2 b
—
—
.084*
(.04)
.083*
(.04)
.089*
( (.04)
Quintile 3 b
—
—
.077
(.04)
.077
(.04)
.089*
( (.04)
Quintile 4 b
—
—
.065
(.04)
.066
(.04)
.082*
( (.04)
Quintile 5 b
—
—
.129**
(.05)
.131**
(.05)
.150**
( (.05)
Time Demands
Overwork
—
—
—
—
-.045
(.02)
-.022
(.03)
WLC
—
—
—
—
—
—
-.060***
(.01)
Model 1
Model 2
Model 3
Model 4
Job Setting
Solo / Small Firm a
.002
(.01)
.008
(.01)
.007
(.01)
.003
(.00)
Mid-size Firm a
.001
(.02)
.015
(.02)
.014
(.01)
.001
(.01)
Large Firm a
.024*
(.01)
.040**
(.01)
.037*
(.02)
.023
(.02)
Income
Quintile 2 b
—
—
-.031*
(.02)
-.031*
(.02)
-.035*
( (.02)
Quintile 3 b
—
—
-.029
(.02)
-.029
(.02)
-.034*
( (.02)
Quintile 4 b
—
—
-.025
(.02)
-.025
(.02)
-.032
( (.02)
Quintile 5 b
—
—
-.045**
(.02)
-.045**
(.02)
-.053**
( (.02)
Time Demands
Overwork
—
—
—
—
.015
(.01)
-.008
(.01)
WLC
—
—
—
—
—
—
.021***
(.01)
Running head: THE STATUS-HEALTH PARADOX
41
TABLE 5. Self-Rated Health Regressed on Practice Setting, Income, and Stressors in Canada (877)
Panel A. Predicting Probability of Good Health
Panel B. Predicting Probability of Poor Health
Note: Coefficients (AMEs) and standard errors (in parentheses) are weighted combined estimates from fifty
multiple imputation datasets. Models include all control variables.
a Compared to public sector lawyers
b Compared to income quintile 1
Model 1
Model 2
Model 3
Model 4
Job Setting
Solo / Small Firm a
-.068
(.04)
-.076
(.04)
-.071
(.04)
-.027
(.04)
Mid-size Firm a
-.115*
(.05)
-.143**
(.05)
-.137*
(.06)
-.085
(.06)
Large Firm a
-.068
(.05)
-.153**
(.05)
-.144**
(.05)
-.064
(.06)
Income
Quintile 2 b
—
—
.022
(.05)
.024
(.05)
.032
( (.04)
Quintile 3 b
—
—
.064
(.05)
.067
(.05)
.089
( (.05)
Quintile 4 b
—
—
.104
(.06)
.104
(.06)
.139*
( (.06)
Quintile 5 b
—
—
.207***
(.06)
.213***
(.06)
.235***
( (.06)
Time Demands
Overwork
—
—
—
—
-.041
(.04)
.004
(.04)
WLC
—
—
—
—
—
—
-.057***
(.01)
Model 1
Model 2
Model 3
Model 4
Job Setting
Solo / Small Firm a
.032
(.02)
.035
(.02)
.032
(.02)
.013
(.02)
Mid-size Firm a
.059
(.03)
.072*
(.03)
.070*
(.03)
.046
(.03)
Large Firm a
.033
(.02)
.079**
(.03)
.074*
(.03)
.033
(.03)
Income
Quintile 2 b
—
—
-.014
(.03)
-.016
(.03)
-.023
( (.03)
Quintile 3 b
—
—
-.039
(.03)
-.042
(.03)
-.057
( (.03)
Quintile 4 b
—
—
-.060
(.03)
-.061
(.03)
-.082*
( (.04)
Quintile 5 b
—
—
-.102**
(.03)
-.106**
(.03)
-.121***
( (.03)
Time Demands
Overwork
—
—
—
—
.022
(.02)
-.002
(.02)
WLC
—
—
—
—
—
—
.026***
(.01)
Forthcoming in the Journal of Health and Social Behavior
42
APPENDIX A. Descriptive Statistics for Study Variables AJD2 Study
Note: Values represent weighted estimates derived from the first imputation dataset
(dependent variables do not contain imputed values). Proportions may not sum to 1 due to rounding.
Variables
Mean / Proportions
SD
Range
Focal Variables
Depression
1.187
1.338
0 - 7
Health
Poor Health
.092
—
—
Average Health
.399
—
—
Good Health
.510
—
—
Income
Bottom Quintile
.233
—
—
2nd Quintile
.181
—
—
3rd Quintile
.191
—
—
4th Quintile
.219
—
—
Top Quintile
.176
—
—
Overwork
.335
—
—
Work-Life Conflict
.048
.985
-2.361 – 1.584
Organizational Context
Public Sector
.267
—
—
Solo Practice/Small Firm
.390
—
—
Mid-Size Firm
.124
—
—
Large Firm
.219
—
—
Controls
Age
37.660
5.801
30 - 73
Race / Ethnicity
Non-white
.201
—
—
White
.799
—
—
Gender
Female
.422
—
—
Male
.578
—
—
Marital Status
Single
.232
—
—
Married
.736
—
—
Cohabiting
.032
—
—
Child at Home
Yes
.568
—
—
No
.432
—
—
Debt
57,875
43,116
0 - 300,000
Cumulative Law School Grade
3.253
.342
2 - 5
Law School Rank
4th Tier
.141
—
—
3rd Tier
.179
—
—
Top 21–100
.511
—
—
Top 11–20
.076
—
—
Top 10
.094
—
—
Running head: THE STATUS-HEALTH PARADOX
43
APPENDIX B. Descriptive Statistics for Study Variables in the LAB Study
Note: Values represent weighted estimates derived from the first imputation dataset
(dependent variables do not contain imputed values). Proportions may not sum to 1 due to rounding.
Variables
Mean / Proportions
SD
Range
Focal Variables
Depression
1.411
1.487
0 - 7
Health
Poor Health
.148
—
—
Average Health
.393
—
—
Good Health
.458
—
—
Income
Bottom Quintile
.187
—
—
2nd Quintile
.217
—
—
3rd Quintile
.197
—
—
4th Quintile
.162
—
—
Top Quintile
.237
—
—
Overwork
.311
—
—
Work-Life Conflict
-.001
.996
-2.008 - 1.574
Organizational Context
Public Sector
.218
—
—
Solo Practice/Small Firm
.428
—
—
Mid-Size Firm
.135
—
—
Large Firm
.219
—
—
Controls
Age
32.329
5.080
26 - 66
Race / Ethnicity
Non-white
.213
—
—
White
.784
—
—
Gender
Female
.543
—
—
Male
.457
—
—
Marital Status
Single
.379
—
—
Married
.410
—
—
Cohabiting
.211
—
—
Child at Home
Yes
.180
—
—
No
.820
—
—
Debt
42,914
37,837
0 - 185,000
Cumulative Law School Grade
4.919
1.271
1 - 8
Law School Rank
Civil Law
.135
—
—
4th Quarter
.099
—
—
3rd Quarter
.289
—
—
2nd Quarter
.208
—
—
1st Quarter
.268
—
—
