Can’t Catch a Break: Intersectional Inequalities at Work

Abstract

The labor market is the site of longstanding and persistent inequalities across race and gender groups in hiring, compensation, and advancement. In this paper, we draw on data from 13,574 hourly service-sector workers to extend the study of intersectional labor market inequalities to workers’ experience on the job. In the service sector, where workers are regularly expected to be on their feet for long hours and to contend with workloads that are intense and unrelenting, regular break time is an essential component of job quality and general well-being. Yet, we find that Black women are less likely than their counterparts to get a break during their work shift. Although union membership and laws mandating work breaks are effective in increasing access to breaks for workers overall, they do not ameliorate the inequality Black women face in access to work breaks within the service sector. A sobering implication is that worker power and labor protections can raise the floor on working conditions but leave inequalities intact. Our findings also have implications for racial health inequalities, as the routine daily stress of service sector takes a disproportionate toll on the health of Black women.

1. Introduction

Service sector work is often physically and mentally taxing, with workers regularly expected to be on their feet for long hours and contending with a workload that is intense and unrelenting (Appelbaum, Bernhardt, and Murnane 2003; Jayaraman et al. 2011; Van Oort 2019; Vargas 2017). Regular break time is therefore an essential component of job quality and general well-being for these workers. Qualitative studies and press accounts have documented the consequences of working without adequate break time (Conti et al. 2006; Dababneh, Swanson, and Shell 2001; Mulholland and Stewart 2014; Schor 1992), but no large-scale data sources provide information on access to breaks. This lack of data has precluded an examination of access to break time and social inequalities in on-the-job respites.

Race and gender inequalities in the labor market are longstanding and entrenched. Prior research has documented gender and racial inequalities in hiring and promotion (Acker 2006; Pager, Bonikowski, and Western 2009; Castilla 2008; Gorman 2005; Quillian et al. 2017; Smith 2002), hourly wages (Mandel and Semyonov 2016; Tomaskovic-Devey 1993a), and other key dimensions of job quality (Storer, Schneider, and Harknett 2020; Tomaskovic-Devey 1993a). However, prior research on gender and racial inequality on-the-job is far more scarce. We address this gap by focusing on a component of the on-the-job experience that is essential for job quality and for workers’ mental and physical health: namely, getting a break during a work shift. Break access may be unequal across race and gender groups because of differential sorting into occupations and industries in which breaks are more accessible, or because of differences in worker power or exposure to protective regulatory environments. Further, the negotiated and discretionary nature of break time in many workplaces creates opportunities for inequalities to emerge in the interactional processes of giving and taking breaks. Women or workers of color may be differentially granted breaks due to bias or discrimination, and they may feel disproportionate pressure to work through breaks. While prior research on job quality has often looked at racial or gender inequalities separately, there are compelling reasons to consider how these two aspects of identity intersect to generate inequalities in working conditions. This paper addresses these gaps.

Drawing on survey data from 13,574 hourly workers employed in retail and food service collected by the Shift Project in 2019 and 2020, we document intersectional race and gender inequalities in access to work breaks. We consider whether inequalities are driven by occupational and industry sorting, differential worker power, and differential exposure to state laws that require work breaks. We then consider whether regulations that raise the floor for all workers may also serve to narrow inequalities in access to breaks.

We find that Black women are significantly less likely than their counterparts in other race/gender groups to get a break during their work shift. This difference persists after controlling for job tenure, occupational sorting, industry subsector, worker power, and exposure to state break laws. Further, even in contexts of greater worker power or with state laws in place to mandate breaks, these inequalities persist. These favorable service-sector environments do improve access to work breaks overall, but they do not appear to ameliorate intersectional racial and gender inequalities in this important dimension of job quality.

2. Prior Research

The Physical and Mental Demands of Retail Work and Importance of Breaks

The service sector is one of the largest growing sectors in the United States and has been shown to have particularly intense physical and mental demands (Appelbaum, Bernhardt, and Murnane 2003; Jayaraman et al. 2011; Van Oort 2019; Vargas 2017). Research has established that working without breaks is common in the service sector, where workers are often scheduled for shifts at the last minute and frequently experience short staffing during shifts (Bernhardt, Spiller, and Theodore 2013; Halpin 2015; Van Oort 2019; Vargas 2017). Working long, difficult shifts without breaks is likely harmful to these workers, as lacking breaks undermines worker well-being in other demanding work contexts. The rise of lean production techniques that demand higher productivity targets in shorter time periods has led to high-intensity work environments, with increased pace of work and limited rest breaks, in many industries (Lambert 2008; Moen and Kelly 2020). These environments have been shown to increase levels of stress, deteriorate physical health, and increase risk of injury in contexts such as food distribution warehouses (Conti et al. 2006) and car manufacturing factories (Mulholland and Stewart 2014). One study of meat-processing workers found that rest breaks during the work day, especially longer breaks, increased physical comfort (Dababneh, Swanson, and Shell 2001).

Research in the white-collar sector has also emphasized the importance of breaks for worker well-being. Like warehouse and factory workers, office workers are facing increasingly intense work environments where they are required to clock longer hours, be consistently available, and work through lunch breaks (Kalleberg 2013; Schor 1992). Although office workers do not suffer from the same type of physical strain as workers in warehouses and factories, breaks during the work day have significant impacts on work-life balance, well-being, and work productivity for these workers (Kalleberg 2013; Moen and Kelly 2020). Thus, across various types of sectors, breaks matter for health and well-being.

Intersectional Inequalities in Job Quality: Theory and Evidence

Because key aspects of job quality such as wages, fringe benefits, and work schedules have been found to be unequally distributed across race/ethnic and gendered lines (Tomaskovic-Devey 1993b; Blau and Kahn 2007; Storer, Schneider, and Harknett 2020) we expect access to work breaks to be another axis of inequality in the U.S. labor market. Prior research finds evidence that race and gender systems work in tandem to shape inequalities in the labor market (Browne and Misra 2003; Greenman and Xie 2008; Kristal, Cohen, and Navot 2018; Paul, Zaw, and Darity 2022), in line with intersectionality theory. This theory, rooted in Black feminism and developed over time by scholars like Patricia Hill Collins (1986; 1990), Kimberlé Crenshaw (1991), and Jennifer Nash (2008), makes several influential arguments, one of which is that systems of oppression are “interlocking,” creating distinct social experiences and power differentials depending a person’s position across these various systems (Collins 1986; 1990).

Interlocking social systems like race and gender could very well contribute to intersectional inequalities in many dimensions of job quality, including access to breaks. However, it is not clear how break access will be patterned across race-gender groups because the privileges and disadvantages associated with a given person’s multiple social positions can vary by the context and the outcome in question (Browne and Misra 2003).

Intersectional race-gender inequalities in access to breaks may be caused by occupational sorting, differential exposure to worker power, and differential exposure to regulatory protections, which all help shape the policies and norms that workers are exposed to at their jobs. We also expect to find residual intersectional inequalities in access to breaks based on discrimination and bias at work that cannot be explained by exposure to different policies or workplace sorting. Below, we draw on the literatures on occupational sorting, worker power, regulatory environments, and discrimination and bias—all of which may contribute to race-gender intersectional inequalities in breaks.

Occupational Sorting.

Access to breaks may differ across social groups in part through occupational sorting processes. Black and Hispanic men and women make up the majority of workers in the lowest-paid occupations in the United States, most of which are service sector jobs (Zhavoronkova, Khattar, and Brady 2022). Research has found that occupational segregation explains a large portion of the wage disadvantage for workers of color (Alonso-Villar and del Río 2013; del Río and Alonso-Villar 2015). Even within the service sector, workers of color, especially women of color, are disproportionately sorted into lower road employers with worse working conditions (Storer, Schneider, and Harknett 2020). Given this evidence, we would expect that workers of color, especially women of color, are concentrated in jobs with fewer breaks as well.

Occupational sorting derives from various processes, including differences in education attainment (del Río and Alonso-Villar 2015) and hiring discrimination based on assumptions about people’s skills, competencies, and the type of work they are best suited for (Cech 2013; Charles and Bradley 2009; Eagly and Wood 2016). Hiring discrimination against both Black and Hispanic job applicants has been consistently documented in several audit studies (Quillian et al. 2017). Pager et al. (2009) find that Black and Latino men face a disadvantage in low-wage labor markets and are funneled into positions requiring less customer interaction and more manual labor, and the discrimination was greater for Black men. Though these sorting processes lead to disadvantages for workers of color with respect to wages, we lack systematic evidence for how occupational sorting impacts access to work breaks.

Worker Power through Union Membership.

Union membership provides workers with greater collective bargaining power, which may lead to not only higher wages, but also improved job conditions. Unions have been shown to improve some non-wage dimensions of job quality in the U.S., such as pension coverage and health insurance (Buchmueller, Dinardo, and Valletta 2002; Budd 2004). Many collective bargaining agreements address working time, such as establishing minimum hours for workers, converting part-time workers to full-time when demand increases, and requiring consistent scheduling (Bernhardt and Osterman 2017). Unions have also been shown to be associated with more stable work hours from week-to-week (Finnigan and Hale 2018), and greater satisfaction with work hours (Bryson and White 2016). The connection between unions and breaks, specifically, is less developed. Unions have long been interested in securing workers’ right to rest, including their historical efforts to limit the workday to eight hours (Whaples 1990). Unions could improve workers’ access to breaks by stipulating breaks in the contact or making workers aware of breaks offered by the employer or possibly the state. Unions such as UFCW-21 have been active in advocating for rest breaks (“Protecting Our Rights - Members Take Their Breaks” 2019). In 2019, UFCW-21 successfully included a rest break provision in an agreement between the union and Fred Meyer (“Agreement by and between UFCW 21 and Fred Meyer, Inc.” 2019). However, collective bargaining agreements may not always negotiate access to breaks, and we know little about whether unions have broadly improved access to breaks in the United States.

If union membership improves break access, as it does for other aspects of job quality, and is distributed unequally across race-gender groups, then union membership could contribute to inequalities in breaks. Evidence that Black men are unionized at higher rates than white men, followed by Black women and then white women, suggests that unionization may help to counteract racial inequality in the labor market (Rosenfeld and Kleykamp 2012).

Worker Power through Labor Market Strength.

The strength of local labor markets can affect job quality vis-à-vis workers’ bargaining power. There is evidence that slack labor markets (i.e., those with relatively higher local unemployment rates) are associated with a rise in “bad” jobs (Wallace and Kwak 2017) and could also be associated with worse job quality, like breaks. If workers are unequally exposed to slack labor markets, this could lead to intersectional inequalities in breaks. In fact, evidence suggests that due to racial segregation, Black workers disproportionately reside in areas with fewer jobs (Ihlanfeldt and Sjoquist 1998; Kain 1992; Stoll 2006; Weinberg 2000), as do Hispanic workers (Easley 2018).

Regulations Requiring Break Time.

The federal government does not guarantee rest or meal breaks for workers. However, 19 states have passed some form of break regulation (U.S. Department of Labor 2022), as shown in Figure 1. These break laws guarantee either a 20- or 30-minute unpaid break for shifts ranging between 5 and 8 hours, depending on the state. The most generous laws are in California, Colorado, Kentucky, New Hampshire, North Dakota, and Washington, where employers are required to give workers a 30-minute break during each 5-hour shift. Most state break laws allow breaks to be waived with mutual employer-employee consent. California is the only state that compensates for untaken breaks, giving workers 1 hour of pay for every untaken break.

Figure 1:

U.S. States with Break Laws

Note: States that have passed some form of break regulation are shown in blue. Data are from the U.S. Department of Labor (Wage and Hour Division) page on “Minimum Length of Meal Period Required under State Law for Adult Employees in Private Sector 1” as of 2022.

If break laws are effective at increasing access to breaks, then race/gender groups concentrated in states with regulations in place will enjoy disproportionate benefits. California is a populous state with a break law in place, and a sizeable share of the California population is Hispanic. Figure 1 shows that many Southern states with large shares of the population identifying as African American do not have break laws. These geographic and regulatory patterns may lead to racial disparities in access to breaks.

Gendered and Racialized Expectations at Work.

There may be residual intersectional inequalities that cannot be explained by occupational sorting or by differential exposures to worker power or break laws. These residual inequalities in access to breaks may come about if women or workers of color experience differential treatment on the job or feel disproportionate pressure to work through breaks to avoid race- or gender-based sanctioning. Unlike wages, breaks tend to be negotiated on a day-to-day basis, similar to work schedules (Halpin 2015). In these everyday interactions at work, managers may hold different expectations of appropriate break-taking depending on the race and gender of the subordinate (Acker 2006), and workers’ expectations of the appropriateness of their own break-taking may vary depending on their racial and gender identity.

Anti-black racial stereotypes and bias could lead to discriminatory treatment of Black workers. The schema that Black people have a worse work ethic has long been used to maintain power—especially labor power—over Black workers (Minoff 2020) and could contribute to managers granting Black workers less rest. The nature of this discrimination may be gendered. For example, research suggests that some employers hold stereotypes of working-class Black women as single mothers—even those who are not parents at all—and thus consider them unreliable workers (Kennelly 1999). In contrast, there is some evidence that employers associate working-class Black women with motivation and commitment due to the assumption that they are the main economic providers for their families (Kennelly 1999; Chris Tilly and Moss 2003). Similarly, some employers view Black women as more educated, more compliant, and more skilled, especially in soft skills like interacting with customers, than Black men (Shih 2002; Chris Tilly and Moss 2003).

Black workers, particularly Black women, may also be less likely to request breaks due to the unequal structures and cultures in which they operate. If they face discriminatory treatment in other areas of work, like raises and promotional cycles (Giuliano, Levine, and Leonard 2011), they may feel pressured to work through breaks to achieve the same rewards as their white counterparts (Kirshenman and Neckerman 1991). While women regardless of race often face backlash for agentic behavior (Eagly and Karau 2002; Rudman 2008), Black women are particularly likely to be perceived as angry or pushy (Donovan 2011; Hicks 2022; Toosi et al. 2019; Weitz and Gordon 1993; C. M. West 1995), which may inhibit their break-taking. To overcome managers’ conscious or unconscious biases, Black women and men may need to work harder and longer, requesting and/or taking fewer breaks.

It is uncertain where Latino workers will fall relative to their Black counterparts in their access to breaks. Some studies have found that certain low-wage employers perceive Hispanic workers to have superior work ethic, especially with regards to doing undesirable jobs, though there is ambiguity on the role of immigration status (Shih 2002; Tilly and Moss 2003). This could mean that Latino workers are sought out for demanding, low-paid jobs because they are seen as being more easily exploited and pressured into working through breaks. On the other hand, this positive perception of Latino workers could mean they are given the benefit of doubt and can more easily take breaks during shifts.

Finally, white workers may gain access to the “resources” or benefits of whiteness in workplaces (Lewis 2004; Reskin 2002; Roediger and Roediger 1999). This may occur via managers through in-group favoritism (Reskin 2002; Tilly 1999) or through the assumed competence of white workers, given that the prototype of the ideal employee was constructed in the image of white men (Nkomo and Al Ariss 2014). Although most service sector workers likely experience pressure to work through their breaks, white men may be the least likely to do so because the concept of rest does not have negative associations based on their race or gender.

However, not all white individuals have access to the full benefits of whiteness in organizations, since these benefits are “simultaneously gendered, heteronormative, classed, and biased toward the able-bodied” (Nkomo and Al Ariss 2014, 397). White women may contend with stereotypes that women workers are less competent (Ridgeway 2001) or less reliable and committed if they are mothers (Correll, Benard, and Paik 2007), which could make them less likely to request and/or be granted breaks. Also, in line with the conception of gender as something that is “done” or “practiced” in relation to others (West and Zimmerman 1987; Martin 2003), women of any race may “do gender” by not taking breaks, since subordination and compliance are expectations of femininity.

Raising the Floor and Inequalities in Working Conditions

Regulations and worker power are two avenues for improving conditions for workers overall; however, it is an open question whether these avenues will also narrow inequalities in job conditions across groups. Regulations and worker power may mitigate inequalities in job quality by lifting up lower-status workers for whom there is more room for improvement or may improve conditions overall but leave inequalities unchanged if all groups benefit equally.

Union Membership.

Union membership could reduce inequality in access to breaks if they equalize access to breaks for workers and thereby improve breaks more for disadvantaged workers. Although the relationship between unionization and inequality in breaks has not been examined, prior research on whether unions reduce gender and racial pay gaps is sizeable but mixed. Some studies do not find different union wage gains by race-gender groups (Rosenfeld and Kleykamp 2012) or find a higher gain for men overall but not by race overall (Budd and Na 2000). However, other studies have found higher union wage gains for non-white groups and for women groups (Farber et al. 2021; Kerrissey and Meyers 2021). Given the mixed findings for income, it remains ambiguous of whether unions reduce race-gender inequality in access to breaks.

Worker Power in Strong Labor Markets.

If strong local labor markets improve job quality for the most disadvantaged workers, then access to breaks could be more equal in strong local labor markets (i.e., in regions with a lower unemployment rate). One report finds that Black workers’ wages are more responsive to changes in the unemployment rate than white workers’ wages: when unemployment goes down, Black workers’ hourly and annual earnings increase more than whites’ (Bivens 2021). Thus, strong local labor markets may reduce inequality in access to breaks by giving a bigger “premium” to Black and possibly Hispanic workers.

Regulations.

Regulations can reduce racial and gender inequality by expanding protections and benefits to previously excluded workers (Davis and Gould 2015; Derenoncourt and Montialoux 2021; Robbins and Vogtman 2015). Implementing a state-level break law could reduce inequality by limiting opportunities for discrimination and mandating breaks in low-quality jobs, which are disproportionately held by women and people of color (Dozier 2010; Huffman 2004; Kristal, Cohen, and Navot 2018; Storer, Schneider, and Harknett 2020). Research has found that raising the minimum wage reduced gender inequality in wages (Davis and Gould 2015; Robbins and Vogtman 2015) and racial inequality in earnings and income (Derenoncourt and Montialoux 2021). Beyond wages, there has been limited literature on the equalizing effects of labor regulations, and no study has attempted to compare the effects of break laws and access to breaks on racial inequalities.

Gaps and Contributions

By assessing access to, and inequalities in, work breaks in the service sector, our study aims to provide insights on a dimension of job quality that plays a critically important role in workers’ lives yet has thus far been understudied. The existing literature on workplace inequalities provides reasons to expect that access to work breaks may be unequally distributed across race/gender groups; however, the nature of these inequalities is uncertain. Because work breaks are discretionary and negotiated in some geographic locations and mandated by law in others, we have an opportunity to consider whether variations in regulations and worker power contribute to inequalities in break access. We therefore make three contributions: (1) We document access to breaks in the service sector by race and gender groups, (2) We examine the role of individual-level attributes, occupational sorting, worker power, and break regulations in contributing to inequalities in access to breaks, and (3) We demonstrate whether conditions that facilitate access to work breaks narrow inequalities across groups in access to breaks.

3. Data and Methods

The data for this paper comes from The Shift Project survey of service sector workers. The Shift Project has collected survey data from workers employed by large retail and food service firms starting in Fall 2016. The data for this paper comes from the survey waves collected in Fall 2019 and Spring 2020, the waves when the questions about breaks during work shifts were included. We analyze data from 13,574 surveyed workers who were paid by the hour and employed at one of 113 large retail or food service employers. The companies included in the data collection include big box stores such as Walmart, Target, and Home Depot; grocery stores such as Safeway, Whole Foods, and Trader Joes; fast food restaurants and cafes such as McDonald’s, Burger King, Starbucks, and Dunkin’; as well as retail apparel, pharmacies, electronics, furniture, and other miscellaneous retailers.

The Shift Project data collection approach uses advertisements on Facebook and Instagram that are targeted to employees of large, named retail and food service firms. The advertisements are targeted to adults 18 years of age or older, residing in the U.S., and who are employees at one of the top retailers or casual dining establishments by revenue. The firms targeted in the Shift Project survey employ over 50% of the retail and food service workforce (author tabulations from the Reference USA database).

Recruitment advertisements include a picture tailored to resemble the appearance of employees of the targeted firm (e.g., matching color of uniform) and their workplace setting (e.g., a grocery store or other appropriate backdrop). The text of the ad reads, “Working at [company name]? Take a Survey and Tell Us about Your Job!” The advertisement also states that those who complete the survey will be entered into a lottery for an iPad or a $500 gift card. Those who click on the ad are directed to a Qualtrics survey, which can be completed using a smart phone, tablet, or computer and takes between 15 and 20 minutes to complete.

The Shift Project survey sample is a non-probability sample, raising concerns about sample selectivity. These concerns are at least partially alleviated by validation of the data in benchmark comparisons with two probability data sources: the National Longitudinal Survey of Youth and the Current Population Survey (Schneider and Harknett 2022). The study team has also conducted tests of bias on unobservables by recruiting respondents using paired recruitment messages designed to recruit respondents at two ends of a continuum of potentially omitted variables: those working too many versus too few hours, those who hate versus love their job, and those who get along with versus don’t get along with their managers. In each of these bias tests, the researchers found that estimates of the relationships of interest between job conditions and well-being outcomes were not significantly different in the paired samples (for further details, see Schneider and Harknett 2022).

The Shift Project has constructed survey weights to align the sample attributes with non-managerial service sector workers in the American Community Survey on race, gender, age, and industry. The sample descriptives we present apply these survey weights. Because all of our regression analyses are stratified by race and gender, we present unweighted regression results. In separate analyses we find that the application of survey weights does not noticeably alter the regression results.

We restricted our sample to workers who reported work shifts that were at least 6 hours long and who provided survey responses about access to work breaks, their gender and race identities, and other covariates. We focus on shifts of at least 6 hours because shorter shifts are not covered by break laws. We also restrict our sample to six intersectional race and gender groups that included at least 100 workers each: white, non-Hispanic; Black, non-Hispanic; and Hispanic women and men. To establish a complete dataset, only observations containing data for all of our demographic, confounder, and exposure variables are used in the analyses. Data and code for replication are available at Harvard’s Dataverse repository, https://doi.org/10.7910/DVN/NJABRM.

Dependent variables.

The main dependent variable in our analysis is a measure of whether the employee has a break during their longest work shift. All surveyed workers were asked the length of their longest and shortest shift, then follow-up questions about the minutes of break time during a shift of each length. We focus on access to a break during a worker’s longest shift, because breaks are arguably especially important when shifts are lengthy. The analysis of breaks is limited to shifts 6 hours or greater as this is the standard for break laws in most break law states. Workers were asked: “At your [employer name] workplace, how many minutes of break time do you get during a shift that is [reported length of longest shift] hours long?” Our main measure of work breaks is a dichotomous measure in which workers who answered “zero minutes” are coded as 0 (having no work break), and those who answered with a positive value were coded as 1 (having a work break). In supplemental analyses shown in the Appendix, we substitute a continuous measure of break minutes in place of the dichotomous measure and find a pattern of results that is largely consistent with those presented in our main analysis.

Race and gender identity.

The main independent variable is a categorical measure of intersectional race and gender identity. This variable includes six possible categories: white, non-Hispanic women; Black, non-Hispanic women; Hispanic women; white, non-Hispanic men; Black, non-Hispanic men; and Hispanic men. We focus our analysis on those six intersectional identities because other categories had small cell sizes (including Asian) or were too heterogeneous to be interpretable (including “other” race and multi-racial).

Control variables.

We control for workers’ reported length of longest shift in all of our models. We also include controls for a set of demographics, human capital, and economic resources. The demographic controls include the respondents’ age, parental status (has dependent children or not), and marital status (married, cohabiting, not living with a partner). Human capital variables include educational attainment (less than high school, high school, some college, associate’s, bachelor’s, advanced degree), and years worked at their current employer (less than 1 year, 1, 2, 3, 4, 5, 6 or more years). Economic resource variables include hourly wage rates and household income bracket.

These background attributes are potential confounders, for instance, if parental demands create pressure to work through breaks to be able to get home to children and if parental status varies across race/gender groups. Another example is that economic resources may provide an economic imperative to work through unpaid breaks and economic resources are likely to vary by race/gender.

Sorting variables.

Our analysis includes two separate categorical variables indicating the workers’ occupation and industry. The occupation measure includes 15 job titles and a separate dichotomous measure of managerial status. The industry subsector variable includes 11 industry subsector categories such as grocery store or pharmacies.

Worker power.

We operationalize worker power using self-reported union membership and the strength of the local labor market, operationalized as the county unemployment rate in the year prior to the survey from the Bureau of Labor Statistics. Because county-level unemployment data was available annually and not monthly, we calculated a weighted average of the unemployment rates for the survey year and the prior year using the month the survey was taken, so that we could then merge unemployment and survey data by county and calendar month.

Regulations.

Finally, we include a measure of the presence of state laws that entitle workers to break time, as previously shown in Figure 1. The stipulations of the laws in place are quite uniform, with all requiring between a 20 to 30 minute unpaid break for a 6 to 8 hour shift.

Table 1 displays descriptive information for key variables included in our models with the survey weights applied. Of note, 20% of workers report that they do not get any break time during a shift of at least 6 hours in length. The average longest shift is 9 hours.

Table 1:

Descriptives for Key Variables

	Mean or Proportion	Std Dev
Has a work break during shift	0.80
Length of shift (hours)	9.0	2.2
Women, white	0.33
Women, Black	0.06
Women, Hispanic	0.10
Men, white	0.31
Men, Black	0.07
Men, Hispanic	0.13
In a union	0.09
County unemployment rate for year prior	4.7	1.3
Works in a state with a break law	0.34
Observations	12,969

Note: Weighted to align with demographic attributes of non-managerial workers employed in the service sector in the American Community Survey. n=605 observations are omitted because of missing values for survey weights.

The distribution of surveyed workers across race and gender categories is 33% white women and 31% white men. Just 6% of the sample are Black women and 7% are Black men. Hispanic women comprise 10% of the sample, and Hispanic men are 13%.

Among surveyed workers, 9% reported being a member of a union, and 34% were employed in a state covered by a mandatory break law. The average county unemployment rate was 4.7 percent.

Appendix Table 1 contains additional descriptive information on the sample. This table shows that the average age of the sample is 37 years old, and almost half of surveyed workers have children. About half are married or living with a partner. 39% of workers have a high school diploma or have not completed high school, 38% have some college, and the rest have at least an associate’s or bachelor’s degree. Almost 30% of workers have 1 year or less than 1 year of tenure in their current job, 36% have worked 2–5 years, and the other third have worked for 6 or more years in their current job. The average hourly wage reported by workers is $13.70.

Appendix Table 2 shows the distribution of occupations and industry subsectors. This table reveals that about 17% or respondents are cashiers, 6% work in sales, 8% work in customer service, 13% work in some form of food preparation, 4% are stockers, and 24% were managers who were paid by the hour. The remaining one quarter of the sample is spread across waitstaff, delivery, drivers, and other occupations. Forty-four percent of workers are employed in restaurants or food service, 23% in a grocery store, and 7% in general merchandise or department stores. The remaining one-quarter of workers are employed in hardware, auto parts, clothing, furniture, electronics, pharmacies, sporting goods, and miscellaneous retail sectors.

Analytic Approach.

In Tables 2 and 3, we estimate a series of nested linear probability regressions to assess race- and gender-based inequalities in access to work breaks. White women are treated as the reference cell because this is the largest demographic group in our sample. To test the significance of differences across the full set of race/gender groups, we conduct post-estimation F-tests of the equality of coefficients corresponding to each group. We generate predicted values of access to breaks for each of our six race/gender groups using the Stata margins command.

Table 2:

Access to Work Breaks Regressed on Race and Gender Identity and Covariates

	(1) Baseline	(2) Demographics	(3) Human Capital	(4) Resources
Women, Black	−0.09***	−0.08***	−0.08***	−0.08***
	(0.02)	(0.02)	(0.02)	(0.02)
Women, Hispanic	0.02	0.04***	0.04***	0.02
	(0.01)	(0.01)	(0.01)	(0.01)
Men, white	0.02**	0.02*	0.02*	−0.01
	(0.01)	(0.01)	(0.01)	(0.01)
Men, Black	0.02	0.03	0.03	0.02
	(0.03)	(0.03)	(0.03)	(0.03)
Men, Hispanic	0.05**	0.05***	0.05***	0.02
	(0.01)	(0.01)	(0.01)	(0.01)
Observations	13574	13574	13574	13574

Note: Coefficients from linear probability models and (robust) SEs shown. All models control for shift duration and year fixed effects. Model 2 adds age, parental status, and marital status. Model 3 adds education and job tenure. Model 4 adds household income and hourly wage. Coefficients on covariates are not displayed for parsimony.

^***= p < .001

^**= p < .01

^*= p < .05

Table 3:

Access to Work Breaks Regressed on Race and Gender Identity and Explanatory Variables

	(1) Occupation	(2) Industry	(3) Union	(4) Unemp rate	(5) Regulation
Women, Black	−0.08***	−0.08***	−0.08***	−0.08***	−0.07***
	(0.02)	(0.02)	(0.02)	(0.02)	(0.02)
Women, Hispanic	0.01	0.01	0.01	0.01	0.01
	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
Men, white	−0.00	−0.00	−0.00	−0.00	−0.00
	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
Men, Black	0.02	0.02	0.02	0.02	0.03
	(0.03)	(0.03)	(0.03)	(0.03)	(0.03)
Men, Hispanic	0.02	0.02	0.02	0.02	0.02
	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
Union			0.08*** (0.01)	0.07*** (0.01)	0.07*** (0.01)
Unemployment rate				0.00	0.00
				(0.00)	(0.00)
Break law					0.06*** (0.01)
Observations	13574	13574	13574	13574	13574

Note: Coefficients from linear probability models and (robust) SEs shown. All models control for shift duration, year fixed effects, age, parental status, marital status, education, job tenure, household income, and hourly wage. Coefficients on occupation and industry and control variables are not displayed for parsimony.

^***= p < .001

^**= p < .01

^*= p < .05

In Table 4, we summarize the results from regressions that interact race and gender identity with measures of worker power and regulations requiring work breaks. The table reconfigures the regression estimates to show the differences that worker power or regulations make within race/gender groups, as well as the “difference in differences” between groups. We use post-estimation F-tests to test the significance of differences within groups and of difference in differences between groups.

Table 4:

Differences in Access to Breaks by Unionization and Break Laws

	Women		Men
	within-group difference (p.p)	between-group difference-in difference (p.p)	within-group difference (p.p.)	between-group difference-in difference (p.p)
Union
White	6.5***	(ref.)	10.3***	(ref.)
Black	7.4	1.0 n.s.	3.8	−6.4 n.s.
Hispanic	4.7	−1.8 n.s.	4.9	−5.4 n.s.
Break law
White	6.5***	(ref.)	6.9***	(ref.)
Black	6.4	−0.0 n.s.	15.9***	9.0 n.s.
Hispanic	7.5***	1.0 n.s.	4.0	−2.9 n.s.

Note: “Within-group difference” is the percentage-point difference in access to breaks by union membership or working in a state with a break law within each of 6 race/gender groups.

“Between-group difference” is the percentage-point difference in Black or Hispanic within-group differences compared with the same-gender white within-group difference.

All models control for shift duration, year fixed effects, age, parental status, marital status, education, job tenure, household income, hourly wage, occupation, and industry.

^***= p < .001

^**= p < .01

^*= p < .05

n.s. = p ≥ .05

4. Results

Table 2 presents results from nested regression models predicting access to work breaks. All models control for the length of shift and year fixed effects. Model 1 shows that Black women are significantly less likely than their white women counterparts to have access to work breaks. Compared with white women, Black women are 9 percentage points less likely to have access to a break from work during a shift at least 6 hours in length. Model 1 also shows that white and Hispanic men are significantly more likely to have a work break compared with white women, by 2 and 5 percentage points, respectively. Model 2 adds controls for parenthood and marital status. A statistically significant 8 percentage-point gap between white and Black women persists. After controlling for demographics, Hispanic women are significantly more likely to get a break than their white women counterparts (by 4 percentage points). White and Hispanic men remain more likely to get breaks than white women after controlling for demographics. Controlling for human capital attributes including educational attainment and job tenure in Model 3 does not change the pattern of results. The final model in Table 2 controls for economic resources including household income and hourly wages. After doing so, the gap between Black women and white women persists, and the coefficients for Hispanic women, white men, Black men, and Hispanic men are not statistically different from white women.

Next, the nested models in Table 3 examine whether intersectional inequalities in access to work breaks are explained by sorting into occupations (Model 1) and industries (Model 2) or by exposure to worker power through unions, strong labor markets, or regulations mandating breaks (Models 3 through 5). All of the Table 3 models control for the variables included in the final model of Table 2: shift length, year fixed effects, demographics, human capital attributes, and economic resources. Model 1 in Table 3 adds to these covariates the occupation of the surveyed worker as well as their managerial status. If Black women were disadvantaged in an occupational or industry sorting process, that might explain their being relatively less likely to have work breaks. However, the inequality in breaks for Black women relative to White women persists without attenuation in Models 1 and 2.

Next, Model 3 controls for union membership to see whether inequalities in union membership contribute to inequality in access to breaks. The model results show that union membership is significantly associated with access to work breaks. Being represented by a union is associated with an 8 percentage point increase in work breaks. However, union membership does not influence the pattern of differential access to work breaks across groups. Black women remain significantly less likely to have work breaks compared with their counterparts in other race/gender groups after controlling for union membership.

Stronger labor markets may improve workers’ power to advocate for better working conditions, because alternative labor market opportunities are more plentiful. Model 4 considers whether exposure to better labor market conditions contributes to intersectional inequalities in break time. For labor market strength to matter for inequality in access to breaks, labor market conditions would need to be correlated with both access to breaks and with intersectional race/gender identities. However, labor market conditions are not associated with access to work breaks, and the relationships between race/gender identity and work breaks does not change when labor market strength is added to Model 4.

Finally, Model 5 shows that living in a state with a mandated break law is significantly associated with having access to work breaks. Working in a state with a break law is associated with a 6 percentage point increase in work breaks compared with those working in non-break-law states.

Figure 2 displays access in work breaks for each of the 6 race/gender groups of surveyed workers, drawing on the fully-adjusted results from Table 3, Model 5. The left panel displays results for women and the right panel for men. What stands out in this visualization is the noticeably lower level of access to breaks among Black women. As shown, about 85% of white and Hispanic women report receiving a work break compared with just 78% of Black women, a 7 percentage point difference. Between 87% and 90% of men in each of the three racial/ethnic groups reported access to work breaks, and none of the racial/ethnic differences were statistically significant among men. Among men, Black workers are not disadvantaged relative to their white and Hispanic counterparts. However, among women, racial disparities appear and persist after controlling for a wide array of background characteristics, sorting into occupations and industries, and controls for worker power and regulatory environments.

Figure 2:

Access to Work Breaks by Race and Gender

Note: Predicted values of access to work breaks with covariates held at their mean values, shown with 95% confidence intervals. Covariates include shift duration, year, age, parental status, marital status, education, job tenure, household income, hourly wage, occupation, industry, union membership, county unemployment rate, and break law.

In Table 4, we summarize results from two regression models in which union membership or working in state with a break law is interacted with race/gender identities. Here, the goal is to reveal whether conditions that raise the floor on working conditions – increasing access to breaks overall – may also narrow intersectional inequalities across race/gender groups. We consider union membership and working in a state with a break law because these are positively associated with access to breaks, and we omit the strength of the labor market because it is not associated with access to breaks.

The upper left portion of Table 4 presents the results for union membership for women. The first column presents the within-group difference in access to breaks between women who are members of a union and those who are not. For instance, white women who are members of a union are 6.5 percentage points more likely to have access to breaks than white women who are not union members. Similarly, Black women who are union members are 7.4 percentage points more likely to have work breaks than Black women who are not represented by a union. Hispanic women who are union members are 4.7 percentage points more likely to have a work break than their non-union counterparts. As shown, union membership is positively associated with increased access to breaks for women in each race/ethnic group, but the relationship is only statistically significant for white women owing to the relatively larger sample size for this subgroup. The next column of the table displays the “difference in differences” between groups. For instance, the boost in work breaks associated with union membership is around 1 percentage point greater for Black women (7.4 – 6.5 percentage points) and almost 2 percentage points smaller for Hispanic women (4.7 – 6.5 percentage points) relative to white women. These between-group differences in differences are not statistically significant.

The upper right portion of Table 4 contains the union membership results for men. White men who belong to a union are 10.3 percentage points more likely to have work breaks compared with white men who do not belong to a union. For Black and Hispanic men, the union difference in work breaks is smaller at 3.8 and 4.9 percentage points, respectively. Although the difference that union membership makes for work breaks appears substantially larger for white men compared with Black men, the difference in differences between these groups (3.8 – 10.3 percentage points = −6.4 percentage points) is not statistically significant. Therefore, we interpret the between-group differences with caution. The same applies to the difference in differences between white and Hispanic men.

The lower portion of Table 4 presents parallel results for working in a state with a break law. Working in a state with a break law is associated with a 6.5 percentage point increase in work breaks for white women, a 6.4 percentage point increase for Black women, and a 7.5 percentage point increase for Hispanic women. The differences between these groups in the increase in work breaks associated with break laws are not themselves statistically significant.

For men in each of the three race/ethnic groups, working in a state with a break law is associated with improved access to work breaks. For white men, break laws are associated with a 6.9 percentage point increase in work breaks. For Black men, the associated gain is even larger in magnitude at 15.9 percentage points. For Hispanic men, working in a state with a break law is associated with a 4.0 percentage point increase in work breaks. The difference in differences between white men and Hispanic men is not statistically significant. Although the magnitude of the difference between white men and Black men is sizeable at 9 percentage points, this difference is not statistically significant (p=.07). We interpret this result as suggestive and consistent with the possibility that break laws may have been particularly helpful to Black men in accessing work breaks.

Robustness.

In our main analysis, access to work breaks is measured as a dichotomy. As a robustness test, Appendix Table 3 estimates the nested models from Table 3 but substitutes a continuous measure of the number of minutes of break time. Consistent with findings in Table 3, Appendix Table 3 shows that Black women’s disadvantage in break minutes persists after controlling for occupation, industry, union membership, labor market strength, and exposure to work-break laws.

As a separate robustness check, we re-estimated the nested regression models in Table 3 with survey weights applied. These weighted regression results are shown in Appendix Table 4. Applying survey weights does not alter the pattern of disadvantage for Black women in access to work breaks. Survey weights increase the magnitude of the relationship between union membership and access to work breaks, as well as between break laws and access to work breaks.

5. Discussion

Multiple dimensions of job quality, including wages, benefits, and work schedules, are unequally distributed across race and gender groups (Darity and Mason 1998; Greenman and Xie 2008; Kristal, Cohen, and Navot 2018; Storer, Schneider, and Harknett 2020). In this paper, we examine a dimension of job quality for which data and research are scarce: access to work breaks. We focus on the service sector, where workers are often on their feet continually, dealing with a steady stream of customers and working in environments that are often short-staffed (Appelbaum, Bernhardt, and Murnane 2003; Halpin 2015; Vargas 2017). In these environments, work breaks play a crucial role in reducing both physical and psychological strain. Because of the existence of widespread inequalities across multiple dimensions of job quality, we expected that access to work breaks may also be unequal.

In fact, we find robust evidence that Black women are less likely to get a break during a long shift than any of five other race/ethnic groups included in our analysis. The difference across groups is sizeable. For instance, access to breaks is 9 percentage points lower for Black women compared with white women. Even after controlling for a wide range of individual characteristics, occupation and industry, worker power, and labor law protections, Black women remain 7 percentage points less likely to get a break during their shift compared with white women. This residual disadvantage could reflect inequalities in break “giving”—such as managers granting Black women fewer breaks due to gender and racial biases—or inequalities in break “taking”—such as Black women taking fewer breaks due to the unequal workplace structure and culture in which they operate.

The inequality in access to breaks during service sector shift work likely contributes to entrenched racial health inequalities. Extensive literatures have documented the higher rates of stress, morbidity, and mortality among Black populations (Hayward and Heron 1999; Hummer and Chinn 2011; Williams 2018; Wrigley-Field 2020). Research finds that Black women exhibit the highest toll of accumulated stress in terms of their allostatic load (Geronimus et al. 2006). Against this backdrop, our results shed light on one pathway through which the routine daily stress of service sector work may disproportionately take a toll on the health of Black women. Breaks may also help workers manage the demands of their personal lives, relationships, and families throughout the workday. Given our findings, Black women would be disproportionately deprived of these possible benefits of breaks on work-life balance.

Importantly, unions and labor laws each play an important role in expanding access to work breaks. Those employed in a state with a break law were significantly more likely to report getting a break during their shift, but only 19 states have such laws on the books, and therefore only one-third of service sector workers are covered by this mandate. Workers represented by a union were more likely to report getting a break during their shifts. However, only 9% of retail and food service workers are represented by a union, mostly those working in grocery stores, meaning that the vast majority of workers do not enjoy the enhanced access to breaks associated with union membership. Expansions in union representation in the service sector, such as in recent successful union organizing efforts at Amazon, Apple, and Starbucks (Mickle and Scheiber 2022; Rosenbaum 2022), could lead to improved break access for newly unionized workers.

Among workers who were represented by a union or covered by a break law, overall access to breaks increased, but the gap in break access between Black women and other workers persisted. Although union membership and break laws did not narrow within-sector inequalities in access to breaks, there is reason to expect that they reduced inequality in the broader labor market. Because the retail and food service sector is disproportionately staffed by workers of color and women, raising the floor in this sector disproportionately benefits these groups.

In interpreting these results and their implications, some limitations should be kept in mind. First, our study relies on a non-probability sample of service sector workers that the Shift Project recruited using targeted advertisements on Facebook and Instagram, raising concerns about sample selectivity and representativeness. Earlier published research using the Shift Project survey data found that the data replicated findings on wages and the wage-tenure relationship compared with probability samples from the Current Population Survey and the National Longitudinal Survey of Youth, allaying concerns about sample selectivity (Schneider and Harknett 2022). Further, even if our sample were selective in terms of having or lacking access to work breaks, we would expect that bias to affect all race and gender groups. Therefore, when it comes to comparing across intersectional race and gender groups, we can think of no reason why sample selectivity would bias these comparisons. Second, we interpret the pattern of intersectional inequalities in break time as likely stemming from processes of discrimination and bias that disadvantage black women in service sector workplaces. We acknowledge that we do not directly measure discrimination or bias but rather assume these processes are at work given the residual inequality in working conditions after controlling for a stringent set of potential confounders, including employer and occupation. This approach of inferring discrimination from residual inequalities has a strong precedent in the research literature but is subject to omitted variable bias (Pager and Shepherd 2008). Another limitation is that our measure of work breaks does not distinguish between paid and unpaid breaks. Unpaid breaks create a tradeoff for workers between getting a respite during a long shift and maximizing their earnings for time spent at work. Ideally, future work would distinguish access to paid and unpaid break time.

The pandemic drew attention to the essential work that service sector workers conduct to meet the basic needs of the population and led to some calls for improving working conditions. In most U.S. states, employers are not legally required to offer break time, and one out of six workers report that they do not get any break time during a shift of 6 hours or longer. In a historically tight labor market in which employers are having difficulty attracting and retaining workers, offering breaks during work should be considered alongside other proposals for improving work conditions such as raising wages and stabilizing work schedules. However, and importantly, our research shows that inequalities in access to breaks were impervious to policies that required breaks during work time. Even when breaks are mandated by law, Black women were still less likely to get a break. Policies can and should raise the floor on working conditions for service sector workers, but doing so will not eliminate subtle and entrenched workplace inequalities in informal workplace relationships that benefit some social groups and disadvantage others.