Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 May 13.
Published in final edited form as: Am J Ind Med. 2018 Dec 19;62(2):131–144. doi: 10.1002/ajim.22935

Mental illness, drinking, and the social division and structure of labor in the United States: 2003-2015

Seth J Prins 1,2, Sarah McKetta 1, Jonathan Platt 1, Carles Muntaner 3, Katherine M Keyes 1,4, Lisa M Bates 1
PMCID: PMC6511991  NIHMSID: NIHMS1006684  PMID: 30565724

Abstract

Background:

We draw on a relational theoretical perspective to investigate how the social division and structure of labor are associated with serious and moderate mental illness and binge and heavy drinking.

Methods:

The Panel Study of Income Dynamics and the Occupational Information Network were linked to explore how occupation, the productivity-to-pay gap, unemployment, the gendered division of domestic labor, and factor-analytic and theory-derived dimensions of work are related to mental illness and drinking outcomes.

Results:

Occupations involving manual labor and customer interaction, entertainment, sales, or other service-oriented labor were associated with increased odds of mental illness and drinking outcomes. Looking for work, more hours of housework, and a higher productivity-to-pay gap were associated with increased odds of mental illness. Physical/risky work was associated with binge and heavy drinking and serious mental illness; technical/craft work and automation were associated with binge drinking. Work characterized by higher authority, autonomy, and expertise was associated with lower odds of mental illness and drinking outcomes.

Conclusions:

Situating work-related risk factors within their material context can help us better understand them as determinants of mental illness and identify appropriate targets for social change.

Keywords: division of labor, drinking, mental health, occupational health, occupations, social class, work

1 |. INTRODUCTION

How are mental illness and binge and heavy drinking patterned by the social division and structure of labor? There are long traditions of research in epidemiology and sociology on the relationships among occupations, the characteristics of work, and mental health outcomes. But research on work and mental health across the labor force is at least a decade old, or utilizes decades old data. The intervening period witnessed ongoing neoliberal transformations of the state and significant dislocations in the U.S. labor market, including continued wage stagnation, increasing economic inequality, declines in union membership and attendant reductions in protections for workers, increasing job insecurity, ongoing market deregulation, and a global economic crisis.16 Theory and evidence from past research, from. materialist and psychosocial perspectives, provide strong reasons to expect these processes and events to influence the social division and structure of labor in ways that have collateral consequences for population mental health.7,8

The present study thus has two primary aims: first, it describes the prevalence and odds of serious and moderate mental illness (“mental illness”) and binge and heavy drinking (“drinking outcomes”) across Census occupation categories in a contemporary, nationally representative sample, updating the literature on their occupational distribution. It also explores the relationship between these outcomes and other aspects of the social division and structure of labor (defined below), specifically, the gap between workers’ productivity and pay, unemployment, and the gendered division of domestic labor. Second, it investigates whether and how the structure of work in the early 21st century is associated with mental illness and drinking outcomes. Building on the research of Hadden et al9 and Alterman et al,10 the study makes novel use of the Department of Labor’s Occupational Information Network (O*NET), an under-utilized data source, and derives contemporary dimensions of the structure of work associated with mental illness and drinking outcomes.11

1.1 |. Work and mental health

The social division and structure of labor have implications for individual and population mental health. A person’s occupation, and the nature of their work, is a snapshot of dynamic power relations, between labor and capital, workers and bosses, the regulatory state and corporations, and even spouses or domestic partners. By virtue of being a snapshot, the effects of these social relations are difficult to detect and estimate; however, researchers have sought to operationalize the organization and character of work in ways that might reveal their traces.

Toward that end, this study adopts a relational approach12 to understanding social class and work as social determinants of health. It draws on relational theoretical perspectives that view the social division of labor into various occupations, and the structure or character of the work done in those occupations, as dynamic, often mutually antagonistic social relations in which individuals are situated relative to their roles in economic production.13 From this perspective, the social division of labor refers to the complex system that emerges in a commodity-producing society, in which individuals engage in qualitatively different forms of work.14 Different people engage in different trades, crafts, and professional specialties across numerous sectors and production processes. In capitalist economies, the social division of labor is driven “chaotically and anarchically by the market,”13 but can be tamed by regulations, social protections, and public investments in particular sectors through subsidies, grants, etc.

The early 21st century was characterized by ongoing changes in the social division of labor; namely, the continued decline in the goods-producing sector and growth of the service sector. Between 1999 and 2017, the percentage of the work force in the goods-producing sector dropped from 19% to 13.9%, while that of the service sector rose from 78% to 86.3%.15,16 This period was also characterized by ongoing wage stagnation coupled with an increase in work hours for all but the top 1% of wage earners, constituting a “productivity-to-pay gap.”17 For example, from 1973 to 2013, hourly compensation of a typical worker rose just 9%, while productivity increased 74%.17

The social division of labor is implicated in mental health: research using U.S. data from the early 1980s found that lawyers, secretaries, and other teachers and counselors had elevated odds of depression relative to people in other occupations.18 Alcohol consumption in particular, while not a health problem per se, is patterned by occupation1922 and causally implicated in at least 60 medical conditions, accounting for roughly the same amount of death and disability as tobacco and hypertension.23 Occupations are clearly implicated in population mental health: a recent study found that suicide rates differ appreciably by occupation, with the highest rates in farming, fishing, and forestry; construction and extraction; and installation, maintenance, and repair.24

The structure of labor refers to the organization of work and the character of work. Different jobs have different physical and psychosocial contexts, and are organized in different ways: the assembly line involves more repetition and fast pace; website design involves more problem-solving and creativity. These characteristics are not merely the aggregated individual attributes of the people who end up in particular jobs, but also reflect “the habituation of the worker to the mode of production.”13

For example, one aspect of the structure of labor is the separation of the work of production into distinct tasks performed by different workers. This subdivision of production is an innovation of managers and owners of capital, enabling them to extract as much value as possible from the labor they purchase.13,14 For example, a trend observed over the course of the 20th century was the deskilling of workers through scientific management and technologization. Through deskilling, managers and owners gained tighter control over the labor process, greater efficiency, and higher profits. First observed in manufacturing, deskilling has also occurred in once “white collar” occupations, for example clerical and administrative work.25,26 Conversely, some professional occupations have resisted deskilling and replacement by technology, such as jobs that require high levels of technical, professional, or creative expertise.27

Considerable research has explored how the organization of work creates stress that can manifest in mental illness and substance use problems. Imbalances in the demands of work (pace, complexity, emotional toll) relative to workers’ control and autonomy are associated with increased depression, alcohol abuse and dependence, and drug dependence.19,28,29 Imbalances between effort and reward are also associated with increases in common psychiatric disorders, including depression and anxiety.30,31 Alienation, self-estrangement, and powerlessness in work are associated with heavier drinking patterns and the development of problem drinking among workers with low occupational status.32,33 However, much of this research prioritizes psychosocial explanations over more relational approaches that situate psychosocial risk factors within the material context of workplace organization, a context designed to increase productivity and profit.

Shifts in power relations against workers and in favor of capital also saw one of the most significant changes in the social division and structure of labor since the turn of the century, the normalization of precarious labor and job insecurity.34,35 Often euphemistically described as “flexible work” or “the gig economy,” precarious labor can indeed be characterized by increased flexibility, but primarily of the sort that benefits employers. This flexibility makes it easier for employers to fire workers, use temporary laborers, and provide only contingent and part-time jobs, without benefits, through individual contracts and other forms of casualization.36 Contingent and casualized labor has also made its way into professional sectors such as journalism and academia.3739

Unemployment and job insecurity influence mental illness and drinking. Unemployment results in economic strains and psychosocial exposures that increase the risk of depression, psychological distress, and harmful drinking.4042 Workers in contingent or casualized jobs share characteristics with people who are unemployed (including periods of unemployment themselves), and share many of the same health risks.43,44 Furthermore, contingent and casualized workers typically have less autonomy, are less likely to raise concerns about workplace health and safety, and have worse health outcomes.4547 “Job churning,” or high rates of job loss during periods of low unemployment, is also associated with poorer health outcomes.48

Changes in the social division and structure of labor also intersect with racialized and gendered power relations, the latter of which we will briefly explore in the present study. (Doing justice to race, gender, and the social division and structure of labor, from a relational theoretical perspective, is beyond the scope of the present study.) Nonetheless, women are concentrated in occupations that are characteristically more stressful, with greater exposure to low decision latitude, high job insecurity, low wages, and lack of benefits.49 Furthermore, despite representing nearly half of the labor force, between 2003 and 2007, women spent 68% more hours on housework then men.50 Hours of housework are positively associated with depressive symptomology51 and inequity in household labor contributes substantially to gender differences in psychological distress.52

1.2 |. The present study

This study revisits the effects of the social division and structure of labor on mental illness and drinking outcomes, building on insights and evidence from research in 1990s and early 2000s. The objective is to conduct a broad, theory-informed empirical analysis of work and mental health, across a wide range of exposures related to the social division and structure of labor that track the major trends summarized above, to both update the literature and identify areas that warrant subsequent in-depth research. Specifically, this study sets out to answer three questions: First, how does the social division of labor pattern mental illness and drinking outcomes in the early 21st century? Second, is there an empirical structure of labor, and if so, what is its relationship to these outcomes? Third, can we theoretically derive dimensions of work that are also related to mental illness and drinking outcomes? To answer these questions, this study draws on the Panel Study of Income Dynamics, a nationally-representative, longitudinal survey of households in the United States, and the Occupational Information Network (O*NET), a comprehensive database of occupations and their characteristics. To our knowledge, this is the first time the latest iteration of the O*NET has been utilized to explore the relationship between the underlying structure of work and mental health across the labor market.

2 |. METHODS

2.1 |. Data and sample

We used data from the 2003–2015 waves of the Panel Study on Income Dynamics (PSID), the world’s longest-running household panel survey.53 PSID surveys are administered at the level of the household. Heads of household were the primary respondents of interest in 1968 when the study began; however, PSID also collects information (from heads of household) on spouses and children. We chose the years 2003–2015 because we are interested in updating the literature on work and mental health with data since the turn of the century, and because these years represented the best overlap in availability for mental illness and drinking measures, updated Census occupation codes, and the O*NET, which has been updated annually since 2003. We used the 2017 version of the O*NET database in our analyses.

We restricted the PSID sample to respondents who indicated that their total annual hours of work were greater than zero and selected a Census occupation category. We excluded individuals who had retired and those in the military (See Table 1 for details).

TABLE 1.

Description of study sample in 2003 (N = 12 423)

Variable N Percent
Sex
 Male 5664 45.6
 Female 6759 54.4
Racialized group membership
 White 7610 61.3
 Black 3401 27.4
 AI/AN 79 0.64
 Asian 210 1.7
 NH/PI 566 4.6
 Other 166 1.3
Employment status
 Working now 8354 90.07
 Temp layoff/leave 88 0.95
 Looking 454 4.89
 Disabled 53 0.57
 Keeping house 253 2.73
 Student 73 0.79
Household income
 <$35 K 4281 34.5
 $35–62.7 K 3500 28.1
 $62.7–100 K 2579 20.8
 >$100 K 2063 16.6
Industry
 Goods producing 2086 22.7
 Services 6578 71.6
Occupation
 Architecture/engineering 161 1.74
 Arts/entertainment 161 1.74
 Building/grounds maintenance 369 3.98
 Business operations specialists 149 1.61
 Community/soc services 153 1.65
 Computer/mathematical 181 1.95
 Construction/extraction 447 4.82
 Education, training, and library 564 6.08
 Farming/fishing/forestry 128 1.38
 Financial specialists 176 1.90
 Food prep/serving 379 4.09
 Healthcare practitioner/tech 409 4.41
 Healthcare support 288 3.11
 Installation/maintenance/repair 395 4.26
 Legal 80 0.86
 Life, phys, soc sciences 88 0.95
 Management 882 9.51
 Office/admin support 1415 15.26
 Personal care/service 329 3.55
 Production operations 809 8.72
 Protective services 240 2.59
 Sales 889 9.58
 Transportation/material moving 583 6.29
Health condition
 Moderate mental illness 2229 24
 Serious mental illness 260 2.8
 Heavy drinking (2005) 190 2
 Binge drinking (2005) 404 4.2

We linked PSID data with data from the O*NET.54 The O*NET is a comprehensive database of work and work characteristics developed under the sponsorship of the US Department of Labor. The O*NET database identifies, defines, and classifies occupations along six major domains: worker characteristics, worker requirements, experience requirements, occupation requirements, occupational characteristics, and occupation-specific information. Data are collected continuously based on a two-stage design in which (i) a random sample of businesses expected to employ workers in the targeted occupations are identified and (ii), a random sample of workers from those selected businesses are sampled. Data are collected by surveying job incumbents using standardized questionnaires. For occupations where it would be difficult to sample workers, O*NET data collectors identify and sample occupation experts from professional or trade association membership lists. Additional ratings are provided by trained occupation analysts. Responses from all three sources are used to provide complete information for each occupation.55 The O*NET provides information for roughly 1000 occupations, which are organized based on the Bureau of Labor Statistics Standard Occupational Classification (SOC) system. This allowed us to link them with Census occupation codes for heads of household and spouses in the PSID.

For all occupations, we obtained 102 items from the work activities (eg, monitor processes, materials, or surroundings; establish and maintain interpersonal relationships), work context (eg, responsibility for outcomes and results, cramped work space or awkward positions), and required education, training, and experience sections of the O*NET, the building blocks for our proxies of the structure of labor. We chose to focus on these aspects because we are interested in the structural qualities of occupations as social locations, rather than the attributes or skills of the individuals who occupy those locations. (Granted this is not a mutually exclusive distinction, especially for education, training, and experience.) Each SOC occupation code is given a “level” and “importance” score, which we summed for each of the 102 work activity and work context items. While the same work activity can be important for a variety of occupations, the amount or level of the activity needed in those occupations can differ dramatically.a We chose to sum, rather than average, level and importance scores to avoid making the assumption that level and importance were equally weighted. Education, training, and experience for each SOC occupation are each measured by an integer scale, which is administered to a sample of job incumbents and occupation experts. We took the weighted mean of these scores for each occupation. This left a dataset of roughly 1000 occupations with 102 work activity, work context, and education, training, and experience variables. To reduce collinearity, we explored a correlation matrix of the 102 items, and combined items that were correlated at greater than 0.8 by taking their mean. This reduced the number of items to 72. We chose 0.8 as the cutoff because exploratory data analysis suggested that this level of correlation sufficiently reduced the data while retaining theoretical interpretability. Finally, when there were multiple SOC occupation codes for a single Census occupation code, we took the mean of SOC code item values, in order to merge the O*NET data with the PSID data.

2.2 |. Measures

2.2.1 |. Occupation

Occupation, our proxy for the social division of labor, was classified by 3-digit Census occupation codes from the 2000 Census. We combined the 508 Census occupation codes appearing in the PSID into 23 standard occupation categories (Table 1).

2.2.2 |. Employment status

Employment status is indicated by comparing respondents whose current employment status is “looking for work/unemployed” to those who are currently employed.

2.2.3 |. Total housework hours

Our proxy for the gendered division of domestic labor, this measure represents respondents’ number of hours per week spent on cooking, cleaning, and other work around the house.

2.2.4 |. Index of productivity to pay

For full-time workers below the top 1% of wage earnings, we created a crude productivity-to-pay indicator, to operationalize the productivityto-pay gap discussed above. We defined this indicator as

ReceivedwagesTruewagesReceivedwages

where received wages are an individual’s hourly pay based on a 40-hour work week, and true wages are an individual’s hourly pay based on the number of hours they actually worked.1,4 This measure can be interpreted as the percentage of their received wage that individuals were hypothetically not paid for productive hours, assuming their output per hour was constant and they were paid at the same rate for every hour worked. This measure is a crude indicator of economic exploitation.

2.2.5 |. Factor-analytically derived structure of work

We hypothesized that the 72 work activity; context; and education, training, and experience items from the O*NET represented a smaller number of underlying latent factors. We conducted an exploratory factor analysis to determine an optimal factor structure, balancing model fit with the intelligibility of the resulting latent variables. Analysis was performed via weighted least squares and an oblimin rotation for oblique factors. Per convention, we chose loadings greater than 0.3 as the cut point for the strength of association between items and factors,5658 and created linear combinations of items by summing the product of item values and their factor loadings, for each occupation. Our preferred factor structure contained nine factors (Root Mean Square Error of Approximation = 0.08, Tucker Lewis Index = 0.79), which we labeled as follows. We also include the number of items and internal reliability for each scale:

  1. Physical and risky, 18 items (Cronbach’s α = 0.92) characterized by manual labor, exposure to environmental hazards, and other bodily risks.

  2. Managerial leadership, 18 items (Cronbach’s α = 0.95) characterized by developing objectives and strategies, monitoring and controlling resources, and coordinating, guiding, and training others.

  3. Analytic and administrative, 11 items (Cronbach’s α = 0.93) characterized by analyzing, interpreting, evaluating, and recording information.

  4. Biomedical exposures and care, 11 items (Cronbach’s α = 0.77) characterized by interacting closely with and caring for sick people.

  5. Service and sales, 12 items (Cronbach’s α = 0.89) characterized by dealing with customers, selling or influencing them, and working with the public.

  6. Team leadership, 9 items (Cronbach’s α = 0.82) characterized by coordinating and leading others and working with a group or team.

  7. Technical and craft, 6 items (Cronbach’s α = 0.88) characterized by working with and repairing technical devices, parts, and equipment.

  8. Assembly line, 9 items (Cronbach’s α = 0.74) characterized by repetition, exactness, and time pressure.

  9. Freedom/unstructured, 7 items (Cronbach’s α = 0.8) characterized by high levels of experience and education and the ability to determine tasks, priorities, and goals.

2.2.6 |. Theoretically derived structure of work

The relational theoretical perspective that informs this study emphasizes the objective, material dynamics that characterize the organization of work. This perspective stands in contrast to traditional stratificationist approaches, which tend to focus on individuals’ interactions within their immediate work environments, and take workplace organization as given, without considering how it is determined by external political, economic, and technological processes.59 For example, while a stratificationist approach might prioritize workers’ subjective psychosocial experiences of workplace demands60 such as task requirements, workload, time pressures, deadlines, and rate requirements, a relational perspective would argue that (i) such prioritization may conceal the objective function of workplace demands to increase productivity and profit, and (ii) these objective features may be more appropriate targets for activism and intervention.59 From these insights, particularly as they relate to deskilling,13 as well as workplace domination and managerial strategies to extract as much value from labor as possible,13,61 we developed four theoretically derived scales based on items selected from the 72 raw O*NET measures. These scales are an attempt to more explicitly operationalize the power relations that characterize the organization of work, and are represented by the sums of the following O*NET items. We also include the number of items, and internal reliability for each scale (Cronbach’s α for scales with more than three items, and the Spearman-Brown ρ, i.e., the standardized Cronbach’s α, for scales with three or fewer items62).

  1. Authority, 8 items (Cronbach’s α = 0.89): Coordinate, guide, lead, and develop others; responsibility for outcomes and results; frequency of decision making; impact of decisions on co-workers or company; monitor and control resources; scheduling work and activities and develop objectives and strategies; make decisions and solve problems; staff organizational units.

  2. Autonomy, 2 items (Spearman-Brown ρ = 0.8): Freedom to make decisions; structured versus unstructured work.

  3. Automation, items (Spearman-Brown ρ = 0.73): Degree of automation, importance of repeating same tasks, pace determined by speed of equipment.

  4. Expertise, 3 items (Spearman-Brown ρ = 0.71): On-site or in-plant training, related work experience, required level of education.

2.2.7 |. Health outcomes

In 2003 and 2007–2015, respondents were administered the Kessler-6 (K6), a six-question scale that was developed to estimate the prevalence of serious mental illness as defined by US Public Law (PL) 102–321, the Alcohol, Drug Abuse, and Mental Health Administration Reorganization Act.63 Serious mental illness is defined as at least one 12-month Diagnostic and Statistical Manual IV disorder (other than a substance use disorder) with serious impairment.64,65 Clinical validation studies have shown that the K6 reliably distinguishes between individuals with and without serious mental illness (Receiver Operating Characteristic Area Under the Curve [ROC AOC]: 0.86), using the cut point of 13 or greater on the scale.65 A cut point of 5 ≤ K6 < 13 (ROC AUC: 0.82) has also been shown to reliably identify respondents with moderate mental illnesses, defined as mental distress necessitating mental health treatment and causing impairments in functioning.66 Both outcomes are examined here.

In the 2005–2015 survey waves, participants were asked about their usual or average quantity and frequency of alcohol consumption. Consistent with guidelines from the National Institute on Alcohol Abuse and Alcoholism,67 we defined “heavy drinking” as usually consuming four or more (women) or five or more (men) drinks in the same occasion, 60 or more times in the past year (or roughly five or more times per month). We defined “binge drinking” as consuming four or more (women) or five or more (men) drinks in the same occasion, at least once in the past month. Because these questions refer to the usual or average number of drinks, they likely underestimate the number of people who have binge drinking episodes or are heavy drinkers. While binge and heavy drinking are not health problems per se, they are associated with increased risk of alcohol dependence and alcohol use disorders, and attendant health problems.68

2.2.8 |. Control variables

Prevalence estimates and odds ratios are adjusted for the following variables unless otherwise noted: self-identified racialized group membership was recorded as black, white, American Indian/Alaska Native, Asian, Native Hawaiian/Pacific Islander, and other. Age was recorded in years. Employment status was categorized as described in Section 2.2.2. Relation to head of household was categorized as head, legal wife, or female cohabitator (no households were identified as headed by a woman with a male spouse present).

2.3 |. Analysis

All analyses were conducted in R version 3.5.0. First we present odds ratios for the relationship between occupations and mental illness and drinking outcomes, utilizing PSID’s longitudinal sampling weights to account for unequal selection probabilities, differential non-response, and differential attrition. These odds ratios and 95% confidence intervals were estimated using dummy variables for each occupation. In each of these models, the reference group is all occupations other than the focal occupation. Model-based prevalences estimates for mental illness and drinking outcomes are presented in the Appendix. We then examine the relationships among our index of productivity to pay, housework hours, and mental illness and drinking outcomes. Next, we present odds ratios and 95% confidence intervals for the relationship between our factor analytic- and theory-derived scales of the structure of labor and mental illness and drinking outcomes. Odds ratios for factors and scales are estimated with continuous variables; the contrast is for a standard deviation unit increase in factor or scale. Estimates control for age, sex, racialized group membership, year, and relation to head of household unless otherwise noted. Complex survey design-weighted logistic models with design-based standard errors were fit with R package “survey.”69 Figures were generated in R package “ggplot2.”70 Figures show results for significant relationships; non-significant findings are excluded for readability.

2.3.1 |. Sensitivity analysis

In our main analysis, we are interested in the total effect of work on mental illness and drinking outcomes; therefore, we do not control for income, as it is an outcome of a person’s occupation and mediates the relationship between work and mental illness and drinking outcomes.71 However, in sensitivity analyses, we explore whether direct effects of work remain after blocking the pathway through a measure of household income. For these analyses, we categorized total annual household income as less than $35 000, greater than $35 000 and less than or equal to $62 700, greater than $62 700 and less than or equal to $100 000, and greater than $100 000.

3 |. RESULTS

Table 1 provides demographic information on the analytic sample, which included 12 423 individuals in 2003. Approximately 72% of the sample worked in service-providing industries (cf. 83% of US adults16), with only 22.7% in goods-producing industries (cf. 17% of US adults16). The largest proportion of the sample worked in office and administrative support occupations (approximately 12%, cf. 17.7% of US adults16); followed by sales (7.5%, cf. 10.6% of US adults16) and management occupations (7.4%, cf., 5% of US adults16). The prevalence of moderate mental illness was 24% (cf. 27.9% of California adults66), binge drinking 4.2% (cf. approximately 14.3% US adults), serious mental illness 2.8% (cf. 6.1% of US adults72), and heavy drinking 2% (cf., approximately 7% of US adults73).

3.1 |. Social division of labor

3.1.1 |. Occupations

Figure 1 shows significant odds ratios (ORs) and 95% confidence intervals (CIs) for mental illness and drinking outcomes comparing each occupation to all other occupations. Occupations characterized by manual labor (“blue collar” jobs) were associated with increased odds of drinking outcomes. Individuals in building and grounds maintenance; construction and extraction; farming, fishing, and forestry; food preparation and serving; production operations; and transportation and material moving occupations had higher odds of binge drinking relative to all other occupations. Eight occupations were associated with lower odds of binge drinking. Individuals in construction and extraction; food preparation and serving; installation, maintenance, and repair; and protective services occupations had higher odds of heavy drinking. A combination of occupations involving manual labor and those involving customer interaction, entertainment, sales or other service-oriented work (“pink collar” jobs) were associated with mental illness. For moderate mental illness, individuals in arts and entertainment; building and grounds maintenance; healthcare support; legal; and sales occupations had higher odds. Three occupations were associated with lower odds of moderate mental illness. Individuals in building and grounds maintenance; healthcare support; and personal care/services occupations had higher odds of serious mental illness. Four occupations were associated with lower odds of serious mental illness. See Supplementary material Table S1 for OR and 95%CI values. Supplementary material Table S2 shows the model-based prevalence estimates and standard errors for mental illness and drinking outcomes by occupation.

FIGURE 1.

FIGURE 1

Odds ratios for mental illness and drinking outcomes for each occupation relative to all other occupations. Note. Serious mental illness is defined as a score of 13 or greater on the K6. Moderate mental illness is defined as score greater than or equal to 5 and less than 13 on the K6

3.1.2 |. Employment status

Looking for work/being unemployed was associated with increased odds of serious (3.7, 2.84–4.82) and moderate (1.86, 1.63–2.12) mental illness. Drinking outcomes were not significantly associated with employment status.

3.1.3 |. Housework hours

Figure 2 shows the predicted probability of serious and moderate mental illness by weekly housework hours (our proxy for the gendered division of domestic labor), stratified by sex. Figure 2 shows that for both men and women, an increase in housework hours is significantly associated with increased moderate and serious mental illness; however, the frequency distribution of housework hours shows that more women engage in housework at all levels of weekly hours. Drinking outcomes were not significantly associated with housework hours.

FIGURE 2.

FIGURE 2

Predicted probability of serious and moderate mental illness relative to increases in housework hours. Note. Serious mental illness is defined as a score of 13 or greater on the K6. Moderate mental illness is defined as score greater than or equal to 5 and less than 13 on the K6

3.2 |. Structure of labor

3.2.1 |. Productivity-to-pay indicator

Figure 3 shows the predicted probability of serious and moderate mental illness, for full time workers below the top 1% of earners, by the percentage of received wages that they were hypothetically not paid for productive hours. Figure 3 shows that the “percentage not paid” to workers is positively associated with moderate mental illness and serious mental illness for both men and women. Controlling for wage type, that is, salaried versus hourly pay, had no effect on this finding (results available upon request). Drinking outcomes were not associated with the productivity-to-pay indicator.

FIGURE 3.

FIGURE 3

Predicted probability of serious and moderate mental illness relative to increases in the productivity to pay gap. Note. Each Yaxis has its own scale, in order to make contrasts visible. Serious mental illness is defined as a score of 13 or greater on the K6. Moderate mental illness is defined as score greater than or equal to 5 and less than 13 on the K6

3.2.2 |. O*NET factors

Figure 4 shows significant odds ratios for mental illness and drinking outcomes for standard deviation increases in O*NET factors. A standard deviation increase in physical and risky work was associated with increased odds of binge drinking (1.25, 1.12–1.4), heavy drinking (1.25, 1.09–1.44), and serious mental illness (1.14, 1.02–1.28). A standard deviation increase in technical and craft work was associated with increased odds of binge drinking (1.15, 1.04–1.27). Standard deviation increases in analytic and administrative work, freedom/unstructured work, and managerial leadership work were associated with lower odds of all four mental illness and drinking outcomes. Standard deviation increases in services and sales work and team leadership work were associated with lower odds of binge drinking, heavy drinking, and serious mental illness. A standard deviation increase in work characterized by biomedical exposures and care was associated with lower odds of heavy drinking.

FIGURE 4.

FIGURE 4

Odds ratios for mental illness and drinking outcomes for standard deviation increases in each factor-analytically derived O*NET dimension of work. Note. Serious mental illness is defined as a score of 13 or greater on the K6. Moderate mental illness is defined as score greater than or equal to 5 and less than 13 on the K6. “Heavy drinking” is defined as usually consuming 4 or more (women) or 5 or more (men) drinks in the same occasion, 60 or more times in the past year (or roughly 5 or more times per month). “Binge drinking” is defined as consuming 4 or more (women) or 5 or more (men) drinks in the same occasion, at least once in the past month

3.2.3 |. Theoretically derived O*NET scales

Figure 5 shows significant odds ratios for mental illness and drinking outcomes by theoretically derived O*NET scales. A standard deviation increase in work characterized by automation was associated with increased odds of binge drinking (1.11, 1.05–1.17). Standard deviation increases in authority, autonomy, and expertise were all significantly associated with lower odds of heavy drinking, moderate mental illness, and serious mental illness.

FIGURE 5.

FIGURE 5

Odds ratios for mental illness and drinking outcomes for standard deviation increases in each theoretically derived O*NET dimension of work. Note. Serious mental illness is defined as a score of 13 or greater on the K6. Moderate mental illness is defined as score greater than or equal to 5 and less than 13 on the K6. “Heavy drinking” is defined as usually consuming 4 or more (women) or 5 or more (men) drinks in the same occasion, 60 or more times in the past year (or roughly 5 or more times per month). “Binge drinking” is defined as consuming 4 or more (women) or 5 or more (men) drinks in the same occasion, at least once in the past month

3.2.4 |. Sensitivity analyses

We repeated the above analyses, controlling for total annual household income, to determine whether a direct effect of our measures of the social division and structure of labor remained. The results of these analyses are available upon request. Controlling for total annual household income did not appreciably alter the above findings; in particular, results for employment status, housework hours, and the productivity to pay gap remained intact. However, additional occupations, O*NET factors, and theoretically-derived scales became associated with mental illness and drinking outcomes for men and women.

4 |. DISCUSSION

Using a large, contemporary, nationally representative sample, this study investigated (i) how the social division of labor patterns mental illness and drinking outcomes in the early 21st century; (ii) whether there is there an empirical structure to the organization and character of work, and if so, whether it is associated with these outcomes; and (iii) whether we could derive, based on theory, dimensions of the organization and character of work that are also associated with these outcomes. This study analyzed a wide range of exposures related to the social division and structure of labor, to both update and extend knowledge on the relationship between work and mental health, and identify areas that warrant subsequent in-depth research. We found that the social division and structure of labor were associated with mental illness and drinking outcomes in several general and specific ways.

The finding that occupation is associated with mental illness and drinking outcomes is largely descriptive, but illustrates that the social division of labor may shape the distribution of these outcomes. Consistent with prior research,18,32,33,74 we find that “blue collar” and lower-status occupations are associated higher odds of drinking outcomes, and that lower-status and public-facing service, sales, and care jobs are associated with higher odds of mental illness. This finding is also consistent with data from the 2003 Canadian Community Health Survey, which found that individuals in sales and service; trades, transportation, and equipment operators; and processing, manufacturing, and utilities occupations had higher risk of poor mental health.75 Prior studies largely used now-outdated versions of census occupation categories or used different measures of mental illness; yet, the results of the present study overlap with and extend this work, as we used current Census occupation categories, a well-validated measure of mental illness, and contemporary, standard definitions of binge and heavy drinking.

Given that individuals’ occupations and mental illness and drinking outcomes were ascertained in the same survey waves, it is also likely that some portion of the association between occupations and these outcomes reflects selection effects; that is, individuals with mental illness and those who engage in binge and heavy drinking may be more likely to obtain employment in particular occupations. In either case, results of sensitivity analyses revealed that these findings were not entirely explained by income.

This study found that looking for work/being unemployed was associated with increased odds of serious and moderate mental illness. These findings are consistent with prior research, which finds that unemployment and job insecurity are associated with increased mental illness, andthat this relationship is not completely explained by selection effects.76 In addition to the direct material and psychosocial consequences of unemployment and job insecurity (although our findings held after adjusting for income), these statuses may also have an indirect influence on mental illness and drinking; workers frequently looking for employment or in insecure jobs may be less inclined to organize collectively against harmful working conditions, since they are less secure in taking on their employers.36

Consistent with prior research, more hours spent on housework are significantly associated with higher odds of moderate and serious mental illness for both men and women. Given that the average number of hours spent on housework in our sample was 7.8 for men and 13.3 for women, our findings suggest that inequity in the division of domestic labor places women at risk of moderate and serious mental illness. This disparity represents persistent processes of oppression and economic exploitation, given that women’s gains in workforce participation by the turn of the century did not reduce their disproportionate burden of unpaid domestic work.77

Regarding the productivity-to-pay gap, for full-time workers below the top 1% of wage earners, every hour that an individual was hypothetically not paid for productive labor was associated with higher odds of moderate and serious mental illness. Sensitivity analyses show that this was the case regardless of individuals’ income and regardless of whether the worker was salaried or paid hourly. This finding suggests that wage stagnation since the turn of the century has consequences for population mental health. More generally, the index of productivity to pay can be conceptualized as a crude measure of economic exploitation, which researchers have identified as an understudied social determinant of mental illness.78,79 Future research should develop more sophisticated operationalizations of individual-and group-level economic exploitation that account for employers’ capital outlays for fixed assets, non-wage forms of worker compensation, and unequal distributions of income-generating assets, and link these measures to population health data. Focusing explicitly on the degree of exploitation in particular occupations or that particular individuals or groups experience, rather than proxies such as income inequality or socioeconomic status, shifts attention to the structural processes that generate economic inequality and which may be more appropriate explanatory mechanisms for mental illness.

This study identified a 9-factor empirical structure to the organization and character of work, and found that physical and risky and technical and craft work were associated with binge and heavy drinking. These findings are consistent with Alterman and colleagues,10 who found that physical labor and hazardous work exposures were associated with heavy drinking. However, beyond physical and risky work, our factor structure differs from Alterman and colleagues’ because we limited items for factor analysis to the characteristics of occupations and excluded characteristics of individuals who hold particular occupations. We did this because our theoretical orientation emphasizes the structure of work as a social relation rather than a set of individual attributes. We found that work characterized as analytic and administrative, freedom/unstructured, managerial leadership, service and sales, and team leadership were associated with lower odds of mental illness and drinking outcomes. Nonetheless, like Alterman and colleagues, our findings differ from traditional job strain and demand/ control models of the character of labor by focusing on objective features of workplace organization rather than subjective assessments of the psychosocial work environment.59 However, as Alterman and colleagues note, there is indirect evidence for substantial overlap between these operationalizations, to the extent that the objective features of workplace organization are correlated with subjective psychosocial assessments of the workplace environment. We argue that the traditional focus on the latter at the expense of the former prioritizes individual-level interventions that leave current structural arrangements intact, and are less likely to be effective than efforts that challenge and alter status quo power relations.

Finally, we operationalized theoretical insights on deskilling and “the habituation of the worker to the mode of production”13 along four dimension: authority, autonomy, automation, and expertise. We found that work characterized by high automation was associated with higher odds of binge drinking, and that higher authority, autonomy, and expertise were associated with lower odds of mental illness and drinking outcomes. These findings are intuitive, but also consistent with contemporary theory that certain occupations may place workers in strategic and privileged positions within labor markets and economic production.27 Control over advanced knowledge and expertise renders these workers’ labor effort difficult to monitor and control. Possession of delegated ownership authority also places workers in strategic and privileged positions within the apparatus of domination that ensures adequate effort from other workers.27 By being relatively protected from domination and exploitation, work characterized by authority, autonomy, and expertise may offer protection from psychosocial risk factors for mental illness and drinking outcomes. Future research should develop explicit, objective measures of economic domination and investigate their relationship with mental health outcomes.

The social division and structure of labor have direct and indirect material consequences, such as work-related physical exertion or the ability to afford healthcare, that likely influence mental illness and drinking patterns. But the impact of features of work such as autonomy or automation on health outcomes ultimately requires diathesis-stress models that account for how these material phenomena “get under the skin.”43,80 A large body of research has identified potential mechanisms for this process, including various forms of alienation and anomie (eg, powerlessness, locus of control, self-estrangement), and has linked these mechanisms to stress and psychological distress.81,82 What remains empirically unclear, however, are the pathways connecting the objective features of the social division and structure of labor to these psychosocial mechanisms. For example, future research might link objective measures of exploitation and domination, O*NET factors and scales, and individual-level assessments of alienation and anomie to explore whether union membership or more democratic workplaces buffers the effect of authority, autonomy, automation, and expertise on mental illness and binge and heavy drinking. Similarly, this approach might be used to determine whether worker ownership of firms or establishments is associated with lower rates of mental illness and drinking outcomes.

Our findings should be understood in light of the following limitations. First, as noted, because our exposures and outcomes are measured in the same study waves over time, associations likely include some effect of social selection into occupations and jobs with particular characteristics. However, prior research suggests that findings such as ours cannot be fully accounted for by selection effects.28 Second, PSID data are collected at the level of the household, and the head of the household is the primary respondent. Virtually all men in the sample were heads of household, versus 40% of women, and those women were likely to be widowed or divorced. The remaining 60% of women were spouses. Thus, women who are or were married are overrepresented in the PSID relative to other large surveys. Although we controlled for relation to head of household in all analyses, findings may not be fully generalizable. Third, there are other important social relations, in particular racialization, racism, gendering, and sexism, which select individuals into occupations and increase domination and exploitation at work8387; these factors warrant more in-depth analysis beyond being treated as control variables.

Drawing on relational social theory regarding the social division and structure of labor, this study updated and extended knowledge about occupations, work, and the prevalence and odds of mental illness and binge and heavy drinking. Work characterized by physical and risky labor, technical and craft labor, and automation were detrimental for mental illness and drinking outcomes. Work characterized as analytic and administrative, freedom/unstructured, managerial leadership, team leadership, and biomedical exposures/care, in addition to work characterized by authority, autonomy, and expertise, were associated with lower odds of mental illness and drinking outcomes. Our findings suggest that explicit engagement with the objective features of occupations and the organization of work is necessary to situate psychosocial risk factors within their material context, to better understand the social division and structure of labor as determinants of mental health, and identify appropriate targets for social change.

Supplementary Material

2

ACKNOWLEDGMENTS

The collection of data used in this study was partly supported by the National Institutes of Health under grant number R01 HD069609 and R01 AG040213, and the National Science Foundation under award numbers SES 1157698 and 1623684.

FUNDING

National Institute on Drug Abuse, T32-DA-37801; National Institute of Mental Health, T32-MH-13043.

Footnotes

DISCLOSURE (AUTHORS)

The authors declare no conflicts of interest.

ETHICS APPROVAL AND INFORMED CONSENT

The research in this manuscript is a secondary analysis of publicly available, de-identified data. As such, it was exempt from IRB review at Columbia University. However, this study falls under a broader program of research that was approved by the Columbia University Institutional Review Board.

DISCLOSURE BY AJIM EDITOR OF RECORD

Paul Landsbergis declares that he has no conflict of interest in the review and publication decision regarding this article.

DISCLAIMER

None.

a

To understand the difference between level and importance, consider the skill “speaking”: possession of this skill is equally important for paralegals and trial lawyers (compared to say, a landscaper), but paralegals do not require as high a level of this skill as trial lawyers, who might argue before the Supreme Court.

SUPPORTING INFORMATION

Additional supporting information may be found online in the Supporting Information section at the end of the article.

REFERENCES

  • 1.Sherman M A Short History of Financial Deregulation in the United States. Washington, DC: Center for Economic Policy Research; 2009. [Google Scholar]
  • 2.Kalleberg AL, Wachter von TM. The U.S. labor market during and after the great recession: continuities and transformations. RSF. 2017;3: 1–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Crotty J Structural causes of the global financial crisis: a critical assessment of the “new financial architecture.” Cambridge J Econ. 2009;33:563–580. [Google Scholar]
  • 4.Milkman R, Luce S. Labor unions and the great recession. RSF. 2017;3:122–145. [Google Scholar]
  • 5.Dunn M, Walker J. Union Membership in the United States. Washington, DC: U.S. Bureau of Labor Statistics; 2016:1–13. [Google Scholar]
  • 6.Gould E 2014. Continues a 35-Year Trend of Broad-Based Wage Stagnation. Washington, DC: Economic Policy Institute; 2015:1–16. [Google Scholar]
  • 7.Navarro V Neoliberalism, Globalization, and Inequalities. Navarro V, eds. Amityville, New York: Baywood Publishing Company; 2007. [Google Scholar]
  • 8.Navarro V, Muntaner C. The Financial and Economic Crises and Their Impact on Health and Social Well-Being. Navarro V Muntaner C, eds. New York: Routledge; 2016. [Google Scholar]
  • 9.Hadden WC, Kravets N, Muntaner C. Descriptive dimensions of US occupations with data from the O*NET. Soc Sci Res. 2004;33:64–78. [Google Scholar]
  • 10.Alterman T, Grosch J, Chen X, et al. Examining associations between job characteristics and health: linking data from the Occupational Information Network (O*NET) to two U.S. national health surveys. J Occup Environ Med. 2008;50:1401–1413. [DOI] [PubMed] [Google Scholar]
  • 11.Muntaner C, Anthony JC, Crum RM, Eaton WW. Psychosocial dimensions of work and the risk of drug dependence among adults. Am J Epidemiol. 1995;142:183–190. [DOI] [PubMed] [Google Scholar]
  • 12.Emirbayer M Manifesto for a relational sociology. Am J Sociol. 1997;103:281–317. [Google Scholar]
  • 13.Braverman H. Labor and Monopoly Capital: The Degradation of Work in the Twentieth Century. New York: Monthly Review Press; 1974. [Google Scholar]
  • 14.Marx K. Capital, Volume 1 London: Penguin Classics; 1990. [1876]. [Google Scholar]
  • 15.Fisk DM. American Labor in the 20th Century. 2001.
  • 16.Bureau of Labor Statistics. BLS Data Finder. Available at: https://beta.bls.gov/dataQuery/search. Published 2018. Accessed June 5, 2018.
  • 17.Mishel L, Gould E, Bivens J. Wage Stagnation in Nine Charts. Washington, DC: Economic Policy Institute; 2015:1–13. [Google Scholar]
  • 18.Eaton WW, Anthony JC, Mandel W, Garrison R. Occupations and the prevalence of major depressive disorder. J Occup Med. 1990;32:1079–1087. [DOI] [PubMed] [Google Scholar]
  • 19.Crum RM, Muntaner C, Eaton WW, Anthony JC. Occupational stress and the risk of alcohol abuse and dependence. Alcohol Clin Exp Res. 1995;19:647–655. [DOI] [PubMed] [Google Scholar]
  • 20.Diala CC, Muntaner C, Walrath C. Gender, occupational, and socioeconomic correlates of alcohol and drug abuse among U.S. rural, metropolitan, and urban residents. Am J Drug Alcohol Abuse. 2004;30: 409–428. [DOI] [PubMed] [Google Scholar]
  • 21.Shaikh RA, Sikora A, Siahpush M, Singh GK. Occupational variations in obesity, smoking, heavy drinking, and non-adherence to physical activity recommendations: findings from the 2010 National Health Interview Survey. Am J Ind Med. 2014;58:77–87. [DOI] [PubMed] [Google Scholar]
  • 22.Barnes AJ, Zimmerman FJ. Associations of occupational attributes and excessive drinking. Soc Sci Med. 2013;92:35–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Room R, Babor T, Rehm J. Alcohol and public health. Lancet. 2005;365:519–530. [DOI] [PubMed] [Google Scholar]
  • 24.McIntosh WL, Spies E, Stone DM, Lokey CN, Trudeau A-RT, Bartholow B. Suicide rates by occupational group—17 states, 2012. MMWR Morb Mortal Wkly Rep. 2016;65:641–645. [DOI] [PubMed] [Google Scholar]
  • 25.Glenn EN, Feldberg RL. Degraded and deskilled: the proletarianization of clerical work. Soc Probl. 1977;25:52–64. [Google Scholar]
  • 26.Derber C Managing professionals. Theor Soc. 1983;12:309–341. [Google Scholar]
  • 27.Wright EO. Class Counts: Comparative Studies in Class Analysis. Cambridge, United Kingdom: Cambridge University Press; 1997. [Google Scholar]
  • 28.Link BG, Lennon MC, Dohrenwend BP. Socioeconomic status and depression: the role of occupations involving direction, control, and planning. Am J Sociol. 1993;98:1351–1387. [Google Scholar]
  • 29.Reed PL, Storr CL, Anthony JC. Drug dependence enviromics: job strain in the work environment and risk of becoming drug-dependent. Am J Epidemiol. 2006;163:404–411. [DOI] [PubMed] [Google Scholar]
  • 30.Stansfeld SA, Fuhrer R, Shipley MJ, Marmot MG. Work characteristics predict psychiatric disorder: prospective results from the Whitehall II Study. Occup Environ Med. 1999;56:302–307. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Godin I, Kittel F, Coppieters Y, Siegrist J. A prospective study of cumulative job stress in relation to mental health. BMC Public Health. 2005;5:67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Yang MJ, Yang MS, Kawachi I. Work experience and drinking behavior: alienation, occupational status, workplace drinking subculture and problem drinking. Public Health 2001;115:265–271. [DOI] [PubMed] [Google Scholar]
  • 33.Seeman M, Anderson CS. Alienation and alcohol: the role of work, mastery, and community in drinking behavior. Am Sociol Rev. 1983;48: 60–77. [PubMed] [Google Scholar]
  • 34.Vosko LF. Managing the Margins. Oxford: Oxford University Press; 2010. [Google Scholar]
  • 35.Weil D The Fissured Workplace. Cambridge, MA: Harvard University Press; 2014. [Google Scholar]
  • 36.Standing G The Precariat. New York: Bloomsbury Academic; 2011. [Google Scholar]
  • 37.Dobbie D, Robinson I. Reorganizing higher education in the United States and Canada. Labor Studies J. 2008;33:117–140. [Google Scholar]
  • 38.Bradley G Contingent faculty and the new academic labor system. Academe. 2004;90:28–31. [Google Scholar]
  • 39.Gollmitzer M Precariously employed watchdogs? Journalism Prac. 2014;8:826–841. [Google Scholar]
  • 40.Dooley D, Fielding J, Levi L. Health and unemployment. Annu Rev Public Health. 1996;17:449–465. [DOI] [PubMed] [Google Scholar]
  • 41.Goeij MCM, Bruggink J-W, Otten F, Kunst AE. Harmful drinking after job loss: a stronger association during the post-2008 economic crisis? Int J Public Health. 2017;62:563–572. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Boden JM, Lee JO, Horwood LJ, Grest CV, McLeod GFH. Modelling possible causality in the associations between unemployment, cannabis use, and alcohol misuse. Soc Sci Med. 2017;175:127–134. [DOI] [PubMed] [Google Scholar]
  • 43.Landsbergis PA, Grzywacz JG. Work organization, job insecurity, and occupational health disparities. Am J Ind Med. 2014;57:495–515. [DOI] [PubMed] [Google Scholar]
  • 44.Kim TJ, Knesebeck Von Dem O. Is an insecure job better for health than having no job at all? A systematic review of studies investigating the health-related risks of both job insecurity and unemployment. BMC Public Health. 2015;15:985. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Benavides FG, Benach J, Diez Roux AV, Roman C. How do types of employment relate to health indicators? Findings from the second European survey on working conditions. J Epidemiol Community Health. 2000;54:494–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Aronsson G Contingent workers and health and safety. Work Employment Soc. 1999;13:439–459. [Google Scholar]
  • 47.European Foundation for the Improvement of Living and Working Conditions. Working conditions in atypical work. Luxembourg: Office for Official Publications of the European Communities, 2001. [Google Scholar]
  • 48.Strully KW. Job loss and health in the U.S. labor market. Demography. 2009;46:221–246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Tausig M The sociology of work and well-Being In: Aneshensel CS, Phelan JC, Bierman A, eds. Handbook of the Sociology of Mental Health. Handbook of the Sociology of Mental Health. Dordrecht: Springer Netherlands; 2013:433–455. [Google Scholar]
  • 50.Krantz-Kent R Measuring time spent in unpaid household work: results from the American Time Use Survey. Monthly Lab Rev. 2009;46–59. [Google Scholar]
  • 51.Glass J, Fujimoto T. Housework, paid work, and depression among husbands and wives. J Health Soc Behav. 1994;35:179–191. [PubMed] [Google Scholar]
  • 52.Bird CE. Gender, household labor, and psychological distress: the impact of the amount and division of housework. J Health Soc Behav. 1999;40:32–45. [PubMed] [Google Scholar]
  • 53.McGonagle KA, Schoeni RF, Sastry N, Freedman VA. The panel study of income dynamics: overview, recent innovations, and potential for life course research. Long it Life Course Stud. 2012;3:268–284. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.National Center for ONET Development. Available at: https://www.onetcenter.org/db_releases.html. Published 2017. Accessed January 1, 2018.
  • 55.National Center for ONET Development. Available at: https://onet.rti.org/index.cfm. Published 2018. Accessed January 1, 2018.
  • 56.Distefano C, Zhu M, Mîndrilă D. Understanding and using factor scores: Considerations for the applied researcher. Pract. Assess. Res. Eval 2009;14:1–11. [Google Scholar]
  • 57.Tabachnick BG, Fidell LS. Using Multivariate Statistics, 5th ed. Boston: Pearson/Allyn & Bacon; 2007. [Google Scholar]
  • 58.Kline P An Easy Guide to Factor Analysis. London, New York: Routledge: 1994. [Google Scholar]
  • 59.Muntaner C, O’Campo PJ. A critical appraisal of the demand/control model of the psychosocial work environment: epistemological, social, behavioral and class considerations. Soc Sci Med. 1993;36: 1509–1517. [DOI] [PubMed] [Google Scholar]
  • 60.Karasek R, Theorell T. Healthy Work. New York: Basic Books; 1990. [Google Scholar]
  • 61.Wright EO. Understanding class: towards an integrated analytical approach. New Left Rev. 2009;60:101–116. [Google Scholar]
  • 62.Eisinga R, Grotenhuis MT, Pelzer B. The reliability of a two-item scale: pearson, Cronbach, or Spearman-Brown? Int J Public Health. 2012;58: 637–642. [DOI] [PubMed] [Google Scholar]
  • 63.Kessler RC, Green JG, Gruber MJ, et al. Screening for serious mental illness in the general population with the K6 screening scale: results from the WHO World Mental Health (WMH) survey initiative. Colpe LJ, ed. Int J Method Psychiat Res. 2010; 19:4–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Kessler RC, Andrews G, Colpe LJ, et al. Short screening scales to monitor population prevalences and trends in non-specific psychological distress. Psychol Med. 2002;32:959–976. [DOI] [PubMed] [Google Scholar]
  • 65.Kessler RC, Barker PR, Colpe LJ, et al. Screening for serious mental illness in the general population. Arch Gen Psychiatry. 2003;60: 184–189. [DOI] [PubMed] [Google Scholar]
  • 66.Prochaska JJ, Sung H-Y, Max W, Shi Y, Ong M. Validity study of the K6 scale as a measure of moderate mental distress based on mental health treatment need and utilization. Int J Method Psychiat Res. 2012;21: 88–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.National Institute on Alcohol Abuse and Alcoholism. Available at: https://www.niaaa.nih.gov/alcohol-health/overview-alcohol-consumption/moderate-binge-drinking. Published 2018. Accessed May 25, 2018.
  • 68.Centers for Disease Control and Prevention. Available at: www.cdc.gov/alcohol/fact-sheets/binge-drinking.htm. Published 2018. Accessed May 21, 2018.
  • 69.Lumley T “Survey: analysis of complex survey samples”. R package version 3.34 2016. [Google Scholar]
  • 70.Wickham H ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York, 2016. [Google Scholar]
  • 71.Muntaner C, Borrell C, Chung H. Class relations, economic inequality and mental health: why social class matters to the sociology of mental health In Mental Health, Social Mirror. Boston, MA: Springer US; 2007: 127–141. [Google Scholar]
  • 72.Kessler RC, Berglund PA, Bruce ML, et al. The prevalence and correlates of untreated serious mental illness. Health Serv Res. 2001;36:987–1007. [PMC free article] [PubMed] [Google Scholar]
  • 73.SAMHSA. 2015 National Survey on Drug Use and Health (NSDUH). 2015. Available at: https://www.samhsa.gov/data/sites/default/files/NSDUH-DetTabs-2015/NSDUH-DetTabs-2015/NSDUH-DetTabs-2015.htm#tab2-46b.
  • 74.Clougherty JE, Souza K, Cullen MR. Work and its role in shaping thesocial gradient in health. Ann NY Acad Sci. 2010;1186:102–124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 75.Marchand A Mental health in Canada: are there any risky occupations and industries? Int J Law Psychiat. 2007;30:272–283. [DOI] [PubMed] [Google Scholar]
  • 76.Kessler RC, Turner JB, House JS. Unemployment, reemployment, and emotional functioning in a community sample. Am Sociol Rev. 1989;54:648–657. [Google Scholar]
  • 77.Swidler E Invisible exploitation: how capital extracts value beyondwage labor. Mon Rev. 2018;69:29–29. [Google Scholar]
  • 78.Prins SJ, Bates LM, Keyes KM, Muntaner C. Anxious? Depressed? You might be suffering from capitalism: contradictory class locations and the prevalence of depression and anxiety in the USA. Sociol Health Illn. 2015;37:1352–1372. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 79.Muntaner C, Ng E, Prins SJ, Bones-Rocha K, Espelt A, Chung H. Social class and mental health: testing exploitation as a relational determinant of depression. Int J Health Serv. 2015;45:265–284. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 80.Monroe SM, Simons AD. Diathesis-stress theories in the context of life stress research: implications for the depressive disorders. Psychol Bull. 1991;110:406–425. [DOI] [PubMed] [Google Scholar]
  • 81.Seeman M, Lewis S. Powerlessness, health and mortality: a longitudinal study of older men and mature women. Soc Sci Med 1995;41:517–525. [DOI] [PubMed] [Google Scholar]
  • 82.Gecas V The social psychology of self-Efficacy. Annu Rev Sociol. 1989;15:291–316. [Google Scholar]
  • 83.Chung Bridges K, Muntaner C, Fleming LE, et al. Occupational segregation as a determinant of US worker health. Am J Ind Med. 2008;51:555–567. [DOI] [PubMed] [Google Scholar]
  • 84.Darity WA Jr, Mason PL. Evidence on discrimination in employment: codes of color, codes of gender. J Econ Perspect. 1998;12:63–90. [Google Scholar]
  • 85.Reskin B, Cassirer N. Occupational segregation by gender, race and ethnicity. Sociol Focus. 1996;29:231–243. [Google Scholar]
  • 86.Bergmann BR. Occupational segregation, wages and profits when employers discriminate by race or sex. Eastern Econ J. 1974;1: 103–110. [Google Scholar]
  • 87.King MC. Occupational segregation by race and sex, 1940–88. Monthly Lab Rev. 1992;115:30–37. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

2

RESOURCES