Occupational Gender Segregation and Mental Health among Professionals: Women’s Risk Exposure in Five Micro Classes

Abstract

This study explores the intricate interplay between gender, occupation, and mental health using data from the 2020 EU-LFS ad hoc module on 38,066 female professionals in Western Europe. We examine their exposure to work-related risks impacting mental health, focusing on variables such as work overload, violence, and challenging client interactions. Our primary objective is to discern how various occupations contribute to distinct experiences of work-induced strains. Key findings challenge the compensating differential theory, according to which the lower wages in female-dominated occupations are compensated by more friendly working conditions, revealing that interactive service-sector jobs pose higher risks to mental well-being. Health professionals, legal-cultural professionals, and teachers are particularly susceptible, with shift and weekend work exacerbating risk exposure to violence and violent behaviors. This study underscores the significance of a “within-gender” perspective, uncovering nuanced occupation-based inequalities for women. It introduces a novel approach to occupational segregation, highlighting the uneven distribution of work-induced strains among different occupations. It also urges to reassess customer-worker relationships and proposes gender-specific measures to alleviate heightened risks to mental well-being for interactive service occupations. In conclusion, this study analyzes the intersection of gender, occupation, and work-induced strains, emphasizing the role of micro-classes in shaping women’s mental health.

1. Introduction

Over the course of the 20th century, the gradual transformation of the economy from manufacturing-based to service-based altered contemporary gender employment patterns. As a result, services now account for the majority of the employment and economic output. What is more, the female employment rate increased as the service sector expanded. From the mid-20th century onwards, female employment rates have grown by half a percentage point per year on average (Olivetti and Petrongolo 2016), and the constant rise of the service sector has boosted this trend.

Moreover, women’s large-scale entry into the labor market, together with the socio-economic changes that have resulted in a greater percentage of two-wage-earner families, certainly increased women’s independence but also generated “new social risks” (Bonoli 2005; Hemerijck 2012; Taylor-Gooby 2004). An overwhelming amount of research has focused on such risks, which include the difficulties of preserving a balance between work responsibilities and family obligations, an increasing number of single mothers as a result of rising divorce rates, and changing family structure with the decline of the extended family. The difficulties women face in trying to achieve a work–life balance may result in a significant loss of income and of future pension payments should they decide to work part-time or take time off from work because of the inadequate provision of childcare or elderly care services, little or no parental leave, or shorter school days (England 2010).

Another essential phenomenon that has been extensively examined in the literature is that of horizontal and vertical segregation, whereby women are overrepresented in precarious, low-skilled, part-time, and atypical jobs and face a greater risk of economic insecurity (Jacobs 1989; Levanon and Grusky 2016; Torre and Jacobs 2021). Horizontal segregation occurs when women are under (or over) represented in specific occupations or sectors; this tendency is strictly correlated with the expansion of the service economy (Blackburn et al. 2002; Torre 2019). On the other hand, vertical segregation denotes a hierarchical divide (Albrecht et al. 2003; Baxter and Wright 2000; Christofides et al. 2013), with women being penalized at the higher level (in terms of income, prestige, and part-time versus full-time employment). In short, women’s career and earning opportunities are more limited than those of men, and the overrepresentation of women in part-time and temporary jobs has clear gender-related outcomes. This fact has significant consequences on the gender wage gap, which, measured in terms of average earnings, shows that women earn about 80% of what men do, even when they hold the same jobs (Levanon et al. 2009; Levanon and Grusky 2016; Ngai and Petrongolo 2017).

Little research has been conducted beyond the study of inequality in material resources. There has been limited analysis of the possible link between women’s employment growth and segregation and more immaterial aspects such as work strains, gender-based violence, and sexual harassment. Nevertheless, gender-based violence in the workplace represents a severe violation of human rights and an attack on women’s dignity and physical and psychological integrity. Across the world, 35 percent of women fall victim to direct violence in the workplace; of these, between 40 and 50 percent are subjected to unwanted sexual advances, physical contact, or other forms of sexual harassment (European Economic and Social Committee 2015). Gender-based violence comes at a cost to employers as well, whether it takes place in the workplace, in public places, or in the home. According to the United Nations, such violence can impact the workplace in the form of decreased productivity, increased absenteeism, health and safety risks, and increased healthcare costs for the employer. These studies provide precious insights into the social conditions underlying the growth in female employment and reveal the twofold effect of expanding women’s employment opportunities.

Notwithstanding this penalization and the risks affecting the female population, women do not constitute a cohesive social unit of identical individuals experiencing the same circumstances, finding themselves in the same positions or settings, and having a similar status in the labor market. Women’s experiences and social positions vary; as such, these need to be accounted for. Relaxing the homogeneity assumption allows us to investigate whether women’s opportunities in a post-industrial economy mean that all women are better off than before or whether certain categories of women have benefited more than others (Mulè 2023). The scant literature on “within-women” inequality often differentiates women according to educational attainment, income level, or occupation (Korpi et al. 2013; McCall 2000; Hook and Li 2020; Sauer et al. 2022).

This paper focuses on a specific category of women at work and investigates differences in work-related strains among those women working as professionals. More precisely, it focuses on professionals employed in the following specific areas: STEM (science, technology, engineering, and mathematics), healthcare and medicine, teaching and education, economic and administrative occupations, and legal and cultural occupations. The reason for selecting these categories is that they account for the highest percentage of employees reporting that their mental well-being is at risk. Secondly, these categories reveals a vast spectrum of occupations in terms of tasks and practices.

Specifically, performing interactive tasks is a typical element of the service economy, a sector of economic activity whose expansion is inextricably linked to the increase in female employment, as well as female participation in the labor market in general (England 2010).

Our analysis focuses on the quality of working conditions among those women professionals employed in specific occupational groupings. Our research critically questions the compensating differential theory that assumes that female-dominated occupations have more “women-friendly” working arrangements that offset lower salaries. The data available show, on the contrary, that interactive service jobs are more exposed to risks to the mental well-being of female workers. Moreover, empirical evidence shows the relevance of “within-gender” inequalities, as professional women working in different occupations face a variety of strains deriving from their jobs.

Our analysis contributes to the literature in several ways. It illustrates how female labor market segregation may be linked to work strains, exposure to violent behavior, and the risks to workers’ mental well-being. On the other hand, it develops an approach that is not mainly interested in the customary between-gender inequalities, but that empirically reveals “within-gender” differences between women in positions of workplace authority. Our analysis also offers policy recommendations for improving female workers’ health conditions in certain occupations where control is not only exercised by employers or managers, but also by customers and where emotional labor represents an important aspect of a worker’s tasks.

1.1. The Gender Penalty: Between Compensating Differentials and Devaluation

The structure of the economy is crucial for understanding the determinants of women’s disadvantaged position in the labor market. In a nutshell, the expansion of the service sector stimulates the growth of so-called “pink-collar jobs” since they attract women much more than they do men. These jobs are often located in the secondary segments of the labor market, and they entail more unstable careers and lower wages, to the extent that some scholars call them “occupational ghettos” (Levanon and Grusky 2016).

This phenomenon is influenced by several factors that have been highlighted in the literature on women’s occupational segregation. Firstly, it reveals that some of the activities performed by women are “functionally and symbolically” similar to the traditional unpaid domestic activities performed by women (Charles and Grusky 2004; Levanon and Grusky 2016). As this effect is present to a greater extent in non-manual occupations, it is a source of horizontal segregation and reinforces the concentration of women in certain specific professions (England et al. 2007). Secondly, the growth of the service sector has increased employment demand to the point where it exceeds the supply of non-married women. This makes employers hire previously inactive women (Charles and Grusky 2004; Levanon et al. 2009; Ngai and Petrongolo 2017) and promotes the growth of a new workforce with more pressing domestic responsibilities and a propensity to seek flexible forms of employment, such as part-time working (Reskin and Roos 1990). As a result, the service sector becomes a reservoir of female employment “by default” (Charles 2005; Charles and Grusky 2004; Iversen and Rosenbluth 2013; Ngai and Petrongolo 2017). Thirdly, the advent of a post-industrial society has led to a further differentiation in the division of labor, with the corresponding growth in functional specialization and the routinization of working tasks and HR practices (Charles and Grusky 2004; Ngai and Petrongolo 2017). One effect of this has been the replacement of small, specialized stores with supermarkets and large retail chains. As small-scale businesses decline, sales and clerical occupations in these sectors are routinized and subjected to a deskilling process; moreover, many women have been hired to fill these new positions. At the same time, however, women’s job opportunities in management have increased, as this process requires a greater ability to supervise and coordinate deskilled jobs (Charles and Grusky 2004; Mandel 2012). The clustering of men and women in these occupations may account for up to 40% of the gender wage gap (Barcus 2022; Fuller and Hirsh 2019).

Starting from this premise, an increasing line of research goes beyond the study of inequality in material resources, such as wages, to investigate how segregation may be linked to work strains and gender-based violence, especially sexual harassment (Folke and Rickne 2022; Stojmenovska 2023). Work strains, psychosocial risks, and the risks to workers’ mental well-being have become a highly pressing issue. A recent report by the European Agency for Safety and Health (EU-OSHA) found that “exposure to psychosocial risks is increasing, with mental health prevalence still emerging. Major work-related exposures have grown in the past 15 to 25 years that is, time pressure, difficult clients, longer working hours and poor communication. There is also some evidence that countries with over average employment in sectors like health and care or other human and client-oriented services (education, social work, tourism, entertainment) suffer from longer working hours and more mental burden. The northern countries are at the top of the countries with highest mental burden. The southern countries have a high share of specific psychosocial risks related to work in tourism and entertainment, characterized by atypical working times and issues with difficult clients.” (EU-OSHA 2023, p. 59). Exposure to such strains has become an important aspect of inequality in the labor market.

Within the literature, evidence has emerged linking gender segregation to gender-based violence. Specifically, women are reported to be more exposed to sexual harassment and gender-based violence in high-paid, male-dominated occupations. At the same time, men experience greater exposure to sexual harassment in female-dominated, low-paid occupations (Folke and Rickne 2022). Stojmenovska (2023) investigated the impact of managerial occupations on women’s well-being; in doing so, she found a greater incidence of burnout and other work strains among men. She argues that exposure to work strains and related health problems should be considered to represent a significant dimension of gender inequality.

These studies provide very valuable insights into the social conditions of women’s employment growth, showing the twofold effect of the expansion of women’s employment opportunities in the high-skilled and low-skilled service sectors. Nevertheless, these analyses propose a trade-off logic based on the idea that segregation constitutes a two-sided process. According to this idea, female-dominated occupations should protect women better from such risks as a result of a non-monetary compensatory mechanism. Basically, the argument put forward is that lower salaries are offset by certain job benefits that facilitate work-family conciliation, making such professions more women- (or family-) friendly (Daw and Hardie 2012; Filer 1990; Fuller and Hirsh 2019). This argument furthers the idea that male-dominated occupations are “culturally hostile” to women and that gender-based violence, harassment, and exposure to work strains should be seen as barriers to de-segregation. Parallel to this argument, there is also a considerable amount of research pointing to specific organizational arrangements that make male-dominated occupations less accessible to women, particularly in terms of working hours and work overload (Battams et al. 2014; Cha 2013). According to this research, the culture of “long working hours” is deeply embedded in male-dominated occupations, not only creating entry barriers for women but also providing reasons for leaving these occupations (Cha 2013; Nemoto 2013).

Compensating differentials and job amenities theories have been contested by advocates of the “devaluation theory” (England 1992; England et al. 2007). According to this line of research, women are paid less for the same level of skills and work experience because they work in professions in which care tasks are performed (England 1992; England et al. 2002; Kilbourne et al. 1994). Various definitions have been used to frame these professions. England et al. adopted the definition of “interactive service work”, distinguishing those professions that involve care and emotional labor from those that do not. Those professions where care is provided include, for example, teaching professions, “cultural” professions, such as those of booksellers and librarians, and medical professions, comprising doctors, nurses, and dental hygienists. According to this view, the devaluation is not offset by job amenities that make these occupations “women-friendly” (Barcus 2022; England et al. 2002). Moreover, service work strongly impacts work strains since it modifies the typical management–employee model of labor relations. Basically, it transforms it into a customer–worker–management triangle (Korczynski 2007; Leidner 1993). In this setting, workers are faced with new challenges and new opportunities. On the one hand, customers have direct control over workers, who now must answer to them. In settings where the relationship is “instrumental or market-driven”, employees seem to experience greater alienation (Gamble 2007; Lopez 2010). However, Lopez (2010) argues that when customers have no more power than workers, as in the case of vulnerable people or specific categories of patients, interactions could also be fulfilling. Secondly, employment in the interactive service care sector entails emotional labor (Hochschild 1983), which exposes workers to burnout and other forms of work strains, especially in the health and medical sector and the teaching sector (Acheson et al. 2016; Kwak et al. 2020; Doyle and Hind 1998; Kwak et al. 2020; Noor and Zainuddin 2011; Rickett and Morris 2021; Yin et al. 2023). Less is known about other occupations that involve care and emotional labor, such as cultural and legal occupations, such as those of librarians, lawyers, and social workers.

1.2. Segregation and Occupations: Within-Gender Differences

The literature discussed up to now has undoubtedly helped advance our knowledge of the determinants of women’s disadvantages in the labor market. However, in studying gender inequality, research has mainly focused on two categories, namely those of women and men, and has implicitly assumed a substantial homogeneity among women. While the gender wage gap has decreased over time, in recent decades, we have observed an increase in wage inequality along other overlapping gender dividing lines. Since de-industrialization has penalized male workers more than women, scholars have focused on income inequality among men. In trying to understand the causes and consequences of the widening wage gap between skilled and unskilled (male) workers, scholars have consistently neglected inequality within the female population (Mulè 2023). Regarding work strains, there are no significant “between-gender” differences (see

Table 1. Percentage of workers who report being exposed to potential risks to their mental well-being according to gender.

	Women	Men	Total
no risk	47.06	50.93	49.10
possible risk	52.94	49.07	50.90
Total	100.00	100.00	100.00

Source: EU-LFS (European averages). Year: 2020.

). This suggests that the “within-gender” approach is better suited for our purposes.

Quantitative research exploring issues related to gender equality is mostly interested in a multi-group comparison (Bauer et al. 2021; McCall 2005). This also means that disaggregating the “women category” with other variables might be a good solution for gaining new information when gender does not appear to play a crucial role as an explanatory variable.

We believe that the most fruitful way to capture inequality among women is by using the category of occupational class. Occupations are considered essential to stratification accounts for certain important reasons (Grusky and Ku 2008; Williams 2017). First, occupations contain the most significant elements portraying social stratification, and they determine the material and symbolic rewards related to work. The organizational position of an occupation, together with the benefits and privileges it offers, significantly conditions certain key aspects of the daily lives of those who perform the job in question. Hence, occupations provide the basis of socio-economic status and have been found to strongly predict labor market outcomes such as earnings, careers, and job quality (Gallie 2013).

Secondly, occupations are claimed to proxy for sources of labor market stratification that often cannot be measured by means of surveys (Williams 2017). The exact mechanisms as to how and why occupations constitute a significant source of such economic inequality range from the occupational grouping of employment relations as in the Erikson-Goldthorpe model of occupational class (Erikson and Goldthorpe 1992) to occupations representing skill requirements (Estévez-Abe 2011).

In this respect, several works have explored inequalities among women (Hook and Li 2020; Sauer and Van Kerm 2021). Korpi et al. (2013), for example, analyzed the impact of different social policy measures on the opportunities enjoyed by different groups of women. One of their main arguments is that social class operationalized through educational attainment should be taken into account. Other scholars (Mulè 2023) argue that occupational class is a very useful variable for highlighting differences between subgroups of women. This resonates with our initial belief that we cannot ignore the degree to which occupations shape opportunities and distribute material and symbolic rewards. Hence, isolating female workers and disaggregating them according to occupation may prove particularly fruitful.

Indeed, when looking at occupational classes, certain important differences emerge among female workers.

Table 2. Percentage of women exposed to potential risks to their mental well-being according to occupational class (ISCO-08 major groups).

	No Risk	Possible Risk	Total
Armed forces occupations	42.55	57.45	100.00
Managers	40.44	59.56	100.00
Professionals	37.28	62.72	100.00
Technicians and associate professionals	44.01	55.99	100.00
Clerical support workers	55.43	44.57	100.00
Service and sales workers	46.93	53.07	100.00
Skilled agricultural, forestry and fishery workers	55.07	44.93	100.00
Craft and related trades workers	58.68	41.32	100.00
Plant and machine operators and assemblers	55.60	44.40	100.00
Elementary occupations	64.00	36.00	100.00
Total	47.01	52.99	100.00

Source: EU-LFS. Year: 2020.

provides some insights into the matter of risk exposure. This class scheme, however, does not permit us to distinguish between occupations requiring a different amount of emotional labor. For this reason, it is preferable to work at a more disaggregated level, more specifically at the third-digit level of the ISCO-08 code. Three-digit ISCO-08 codes largely correspond to what we can define as “micro classes” (Grusky and Sørensen 1998; Jonsson et al. 2009; van Leeuwen 2017). Workers who belong to the same micro class “share not only the same life chances and labor market circumstances as those in a large class, but they have something more in common, for instance, the techniques and equipment they use, the products they make, the capital they need, or the customer service skills required to serve a local community. Micro classes can be seen as occupational networks through which resources (human, cultural, social, and economic capital) are exchanged, creating layers of stratification both within and between major classes” (van Leeuwen 2017, p. 9). At this level of disaggregation, we can therefore identify those “interactive service” occupations that require workers to perform emotional labor.

Following the “micro-class” approach, we believe that it would be more useful to study how different occupations can expose workers to work strains within the same major class. We are interested in studying work strains among professionals. This class constitutes an interesting case for two reasons. Firstly, this class contains the highest percentage of workers reporting exposure to risks to their mental well-being. Secondly, this class features a very broad spectrum of occupations in terms of the tasks and practices concerned. It essentially includes five categories of occupation: STEM (science, technology, engineering, and mathematics) occupations, health and medical professionals, teachers, economic and administrative workers, and the legal and cultural professions. These workers carry out different kinds of tasks regarding interactive service work. Service work and interactions with other people are of key importance for health professionals, teachers, and legal-cultural professionals. Moreover, these are occupations where emotional labor is a crucial part of the job.

Table 3. Occupational groupings.

STEM	Health and Medicine	Teaching	Economic-Administrative	Legal-Cultural
Physical and Earth Science Professionals	Medical Doctors	University and Higher Education Teachers	Finance Professionals	Librarians, Archivists, and Curators
Mathematicians, Actuaries, and Statisticians	Nursing and Midwifery Professionals	Vocational Education Teachers	Administration Professionals	Social and Religious Professionals
Life Science Professionals	Traditional and Complementary Medicine Professionals	Secondary Education Teachers	Sales, Marketing and Public Relations Professionals	Authors, Journalists, and Linguists
Engineering Professionals (excluding Electrotechnology)	Paramedical Practitioners	Primary School and Early-Childhood Teachers		Creative and Performing Artists
Electrotechnology Engineers	Other Health Professionals	Other Teaching Professionals		Legal Professionals
Software and Applications Developers and Analysts
Database and Network Professionals

below shows how we grouped occupations (at the third digit level of the ISCO-08 code) from the above-mentioned categories.

We excluded architects, who are in the same ISCO sub-group as the “engineering occupations” but do not qualify as “STEM”, and we have also excluded vets from health and medical professionals since our focus is exclusively on occupations where the services provided are entirely aimed at humans.

A “within-gender”-oriented research approach is very important due to the fact that these occupations are heavily embedded in that “system of social control” constituting segregation (Jacobs 1989). When we observe the gender composition of these occupations, we find that STEM professions are male-dominated, with women representing less than 33.3% of the workforce (Torre 2019), while health and medicine and teaching are female-dominated (women constitute more than 66.6% of the workforce); economic and administrative occupations and the legal-cultural professions, on the other hand, are “gender-neutral”, although the second group is much closer to the female-dominated threshold (see

Table 4. Gender composition of grouped occupations.

	Gender
	Men	Women	Total
STEM	78.17	21.83	100.00
Health-medicine	29.05	70.95	100.00
Teaching	29.46	70.54	100.00
Economic-administrative	50.68	49.32	100.00
Legal-cultural	39.58	60.42	100.00
Total	44.96	55.04	100.00

Source: EU-LFS (European averages).

).

We advance two arguments. Occupations heavily shape people’s life chances. These life chances encompass both the material dimension, such as wages and compensation, and the immaterial dimension, such as how working conditions expose workers to risks to their mental health and well-being. For this reason, we further this first hypothesis.

H1. 

We expect to observe relevant variations within women in the exposure to work strains in different occupations.

Moreover, segregation is a system of social control (Jacobs 1989). For this reason, compensation, pay, and working conditions should be better in occupations where women are largely excluded. Specifically, health and medicine, teaching, and legal-cultural occupations involve job tasks that require a high amount of emotional labor, which previous research hinted could be a major factor in exposing workers to strains and burnout. For this reason, we also further the following hypothesis.

H2. 

Work strain exposure is higher in women in health and medicine, teaching, and legal-cultural occupations.

2. Methods

2.1. Participants

We used the 2020 EU-LFS ad hoc module, which includes survey items relating to exposure to both physical and mental risks1. We were particularly interested in the second kind of risk. We restricted our sample to women who work in occupations listed in Table 3. In this regard, we have gathered data on the following Western European countries: Austria, Belgium, Germany, Denmark, Spain, Finland, France, Greece, Ireland, Italy, Netherlands, Norway, Portugal, and Sweden. The final sample comprised 38,066 women.

2.2. Instruments

Our dependent variable was derived from the survey item regarding “Exposure to mental well-being risk factors” (variable “ahm2020_mentrisk”). This item is a categorical variable structured in the following way (see

Table 5. Mental risk factors reported in the LFS 2020 ad hoc module.

Exposure to Mental Well-Being Risk Factors	Freq.	Percentage	Cum.
No significant risk factor for mental well-being	14,148	37.28	37.28
Yes, mainly severe time pressure or overload of work	11,117	29.30	66.58
Yes, mainly violence or threat of violence	681	1.79	68.37
Yes, mainly harassment or bullying	392	1.03	69.41
Yes, mainly poor communication or cooperation within the organization	2504	6.60	76.00
Yes, mainly dealing with difficult customers, patients, pupils, etc.	5262	13.87	89.87
Yes, mainly job insecurity	1806	4.76	94.63
Yes, mainly lack of autonomy or lack of influence over the work pace or work processes	841	2.22	96.85
Yes, mainly another significant risk factor for mental well-being	1197	3.15	100.00
Total	37,948	100.00

Source: EU-LFS (European averages). Year: 2020.

).

As a first step, we recoded the variable as a dummy that simply tells us if a woman is exposed to any risk to her mental well-being.

2.3. Procedure

After this, we isolated specific risks that are relevant to the theoretical discussion about compensating differentials perspective: “severe time pressure or overload of work”, “violence or threat of violence”, and “dealing with difficult customers, patients, pupils”. We recoded these variables as dummies. Work overload and dealing with difficult people are the most common risks reported by women. We also included violence because of the considerable relevance of this issue, even though, in this case, we are not dealing with gender-based violence but with a more general form thereof.

2.4. Data Analysis

Our explanatory variable was a categorical variable that tells us about the occupational field in which a person works, with STEM occupations as a reference category.

Our controls included age, marital status, educational attainment, weekly working hours, part-time work, firm size, presence of shift work, presence of “weekend working”, and the country in which a person works.

We carried out a series of logistic regressions, firstly to compute the probability of being exposed to work strain (which represents a possible risk to a worker’s mental health). We then estimated separate models for the other specific risks mentioned above.

3. Results

In this section, we report the results of the logistic regressions performed. We show the Average Marginal Effects2 (AMEs) for the most important results. We computed the AME for full-time and part-time workers since spending less time in the workplace might impact a person’s exposure to work strains.

We found that women working as health professionals, on average, display a 16% greater chance of being exposed to mental health risks than women working in STEM occupations (see Figure 1 and

Table 6. Average Marginal Effects: full-time workers. Number of obs = 31,345.

	Difference in Probabilities	Std. Err.	z	P > z	(95% Conf. Interval)
Health-medicine	0.161	0.012	13.750	<0.001	0.138	0.184
Teaching	0.117	0.010	11.270	<0.001	0.097	0.137
Economic-administrative	0.036	0.011	3.300	0.001	0.015	0.058
Legal-cultural	0.142	0.012	12.130	<0.001	0.119	0.165

Source: authors’ own elaboration from EU-LFS micro-data. Note: dy/dx for factor levels is the discrete change from the base level.

). Teachers and legal-cultural professionals are 12% and 14% more likely to be exposed to risks to their mental well-being than women working in STEM occupations, respectively. On the other hand, women working in economic and administrative occupations are much less likely to be exposed to mental well-being risks, even though they are 4% more likely to be exposed to mental well-being risks than women working in STEM occupations. Generally, we do not observe important differences between full-time and part-time workers.

Then, we looked at the three specific dimensions of work strains we are interested in here, namely (1) work overload and time pressure, (2) violence or the risk of violence, and (3) difficult interactions with customers/patients/pupils.

3.1. Work Overload

With regard to time pressure and work overload, we observed a similar pattern (see

Table 7. Average Marginal Effects: full-time workers. Number of obs = 20,779.

	Difference in Probabilities	Std. Err.	z	P > z	(95% Conf. Interval)
Health-medicine	0.125	0.013	9.350	<0.001	0.099	0.152
Teaching	0.044	0.011	3.830	<0.001	0.021	0.066
Economic-administrative	0.024	0.012	2.010	0.045	0.001	0.047
Legal-cultural	0.074	0.014	5.470	<0.001	0.048	0.101

Note: dy/dx for factor levels is the discrete change from the base level. Source: authors’ own elaboration from EU-LFS micro-data.

and Figure 2). Health professionals are about 13% more likely to be exposed to risks related to work overload than STEM workers, while teachers and legal-cultural professionals are 5% and 8% more likely to be exposed to such risks than STEM workers (see

Table 8. Average Marginal Effects: part-time workers. Number of obs = 20,779.

	dy/dx Difference in Probabilities	Std. Err.	z	P > z	(95% Conf. Interval)
Health-medicine	0.126	0.014	9.330	<0.001	0.100	0.153
Teaching	0.045	0.012	3.810	<0.001	0.022	0.068
Economic-administrative	0.024	0.012	2.010	0.045	0.001	0.048
Legal-cultural	0.075	0.014	5.470	<0.001	0.048	0.102

Note: dy/dx for factor levels is the discrete change from the base level. Source: authors’ own elaboration from EU-LFS micro-data.

and Figure 2). Economic-administrative occupations are very similar to STEM occupations in this regard. Relative to overall exposure, teaching professionals are more similar to economic-administrative professionals in terms of exposure to time pressure and work overload. Moreover, in this case, we did not observe important differences between full-time and part-time workers.

3.2. Violence or the Threat of Violence

In terms of violence or the threat of violence, we saw no important differences between occupations. However, such differences increase dramatically when we compute AMEs for workers who perform shift work and Sunday work (see

Table 9. AMEs: No shift or Sunday work (full-time workers). Number of obs = 11,858.

	Difference in Probabilities	Std. Err.	z	P > z	[95% Conf. Interval]
Health-medicine	0.052	0.006	8.090	<0.001	0.040	0.065
Teaching	0.042	0.004	10.370	<0.001	0.034	0.050
Economic-administrative	0.002	0.003	0.940	0.349	−0.003	0.007
Legal-cultural	0.066	0.007	9.050	<0.001	0.051	0.080

Note: dy/dx for factor levels is the discrete change from the base level. Source: authors’ own elaboration from EU-LFS micro-data.

,

Table 10. AMEs: No shift or Sunday work (part-time workers). Number of obs = 11,858.

	Difference in Probabilities	Std. Err.	z	P > z	[95% Conf. Interval]
Health-medicine	0.059	0.009	6.780	<0.001	0.042	0.076
Teaching	0.048	0.006	7.410	<0.001	0.035	0.060
Economic-administrative	0.003	0.003	0.930	0.351	−0.003	0.008
Legal-cultural	0.074	0.010	7.350	<0.001	0.054	0.093

Note: dy/dx for factor levels is the discrete change from the base level. Source: authors’ own elaboration from EU-LFS micro-data.

,

Table 11. Average Marginal Effects: shift work and Sunday work (full-time workers). Number of obs = 11,858.

	Difference in Probabilities	Std. Err.	z	P > z	[95% Conf. Interval]
Health-medicine	0.158	0.048	3.280	0.001	0.064	0.253
Teaching	0.121	0.043	2.780	0.005	0.035	0.206
Economic-administrative	−0.016	0.030	−0.520	0.600	−0.075	0.043
Legal-cultural	0.210	0.059	3.590	<0.001	0.095	0.326

Note: dy/dx for factor levels is the discrete change from the base level. Source: authors’ own elaboration from EU-LFS micro-data.

and

Table 12. Average Marginal Effects: shift work and Sunday work (part-time workers). Number of obs = 11,858.

	Difference in Probabilities	Std. Err.	z	P > z	[95% Conf. Interval]
Health-medicine	0.165	0.025	6.690	<0.001	0.117	0.213
Teaching	0.138	0.023	6.080	<0.001	0.093	0.182
Economic-administrative	0.009	0.010	0.920	0.356	−0.010	0.028
Legal-cultural	0.199	0.031	6.490	<0.001	0.139	0.259

Note: dy/dx for factor levels is the discrete change from the base level. Source: authors’ own elaboration from EU-LFS micro-data.

, Figure 3 and Figure 4).

For those who work shifts or on Sundays, the chance of being exposed to such risk to their mental well-being is 16% higher than that of STEM workers in the case of health professionals, 12.1% in the case of teachers, and 21% in the case of legal-cultural professionals.

3.3. Dealing with Difficult People

With regard to the likelihood of dealing with difficult people, we observed the same pattern as in the previous two cases. However, the likelihood of such exposure is generally greater for all interactive service occupations. This probably reflects the very nature of these jobs, which revolve around interactions with customers, patients, clients, or students (see

Table 13. AMEs: working with difficult people (full-time workers). Number of obs = 15,725.

	Difference in Probabilities	Std. Err.	z	P > z	[95% Conf. Interval]
Health-medicine	0.264	0.013	19.870	<0.001	0.238	0.290
Teaching	0.260	0.010	26.340	<0.001	0.240	0.279
Economic-administrative	0.044	0.010	4.480	<0.001	0.025	0.063
Legal-cultural	0.256	0.013	19.250	<0.001	0.230	0.282

Note: dy/dx for factor levels is the discrete change from the base level. Source: authors’ own elaboration from EU-LFS micro-data.

and

Table 14. AMEs: working with difficult people (part-time workers). Number of obs = 15,725.

	Difference in Probabilities	Std. Err.	z	P > z	[95% Conf. Interval]
Health-medicine	0.269	0.014	19.040	<0.001	0.241	0.297
Teaching	0.265	0.012	22.440	<0.001	0.242	0.288
Economic-administrative	0.045	0.010	4.440	<0.001	0.025	0.065
Legal-cultural	0.262	0.014	18.360	<0.001	0.234	0.290

Note: dy/dx for factor levels is the discrete change from the base level. Source: authors’ own elaboration from EU-LFS micro-data.

, Figure 5).

4. Discussion

We posited significant differences in exposure to work strains among women using micro classes as a strategic variable. Our data support this hypothesis since we observed (1) significant differences compared with the reference category, i.e., STEM occupations, and (2) differences among the other four micro classes, particularly in general risk exposure between the legal-cultural, teaching, health and medical, and economic-administrative professions. In addition, we also observed differences in exposure to specific forms of work strains among the above-mentioned occupations.

Specifically, we argued that, among those occupations in which women are called upon to perform tasks involving interaction with customers in addition to with their employers and peers, women are subject to greater exposure to risks to their mental well-being. The rationale behind this hypothesis is that interacting with customers, patients, and audiences adds another layer to the organizational structure within which working tasks are performed. The employees, in such cases, are answerable to their superiors (employer, managers, and supervisors) and are also accountable to a third party, the customer. Customer control over workers represents a potential source of alienation, and it directly impacts workers’ exposure to strains and risks to their mental well-being. Our data corroborate this hypothesis since we found a greater degree of exposure to work strains among those occupations in which worker–customer relationships are crucial, namely the health and medical, teaching, and legal-cultural professions.

These are also occupations that incorporate emotional labor. As argued, customer pressure and the obligation to conduct themselves in a manner characterized by friendliness and fairness towards the customer increases workers’ exposure to work strains (Kwak et al. 2020). Our data corroborate this hypothesis since emotional labor is central to the tasks performed in the above-mentioned occupations. Indeed, from this point of view, it is hardly surprising that dealing with difficult people is indeed a risk to which such workers are more greatly exposed, meaning that customer relations in contexts where the customers enjoy greater power are a source of distress for the workers involved. Results regarding time pressure-related risks and violent behavior, especially among those employees working shifts or at weekends, further support the idea that interactive service work can be a source of risk for workers’ well-being.

These risks coalesce into a constellation of “work-induced forms of strains”, which can be categorized as “class-based”. Such strains are intricately linked to micro-level class distinctions and can be interpreted as integral components of a worker’s “life chances” (Weber 1918). How does this nexus interrelate with gender segregation? The theory of compensating differentials posits that women experience a trade-off between material resources and adversarial work environments. Occupations predominantly performed by men tend to harbor more hostile working environments for women; however, they also offer greater material rewards. Conversely, female-dominated occupations tend to provide a less antagonistic working environment but comparatively lower material gains. Based on this conjecture, one would anticipate an elevated exposure to work-induced strains among women engaged in STEM occupations. Nevertheless, our empirical data show that women employed in the health and medical fields, teaching, and the legal-cultural professions experience heightened greater exposure to the risks concerned. The first two occupations are dominated by women, while the latter is close to the threshold for female-dominated categorization.

We do not deny the existence of gender-based risks for women working in male-dominated occupations. However, our data indicate that segregation is not merely a trade-off between material resources and hostility within the working environment, as there are occupation-related risks whose prevalence is more pronounced in those areas of employment where women constitute the vast majority of the workforce.

5. Conclusions

There are three important conclusions to be taken from this paper. The first concerns the fresh evidence regarding the interaction between gender and micro class. We found that the risks to workers’ mental well-being are greater among those occupations in which customer-worker interaction is central to the tasks performed, and emotional labor represents a crucial dimension of said interaction. Moreover, we found no difference between part-time and full-time workers, implying that the occupation is a much more relevant source of work strain than the amount of time spent in the workplace, regardless of the occupation.

The second conclusion regards the adoption of a “within-gender” perspective. Descriptive data show no significant differences between men and women regarding exposure to risks to their respective mental well-being. We adopted an approach that studied differences among women, using occupations as the primary variable of interest for observing such differences. Following this approach, it was possible to identify a source of occupation-based inequality, which is a more pressing issue for women than men.

The third finding we made regards segregation. With few exceptions (Stojmenovska 2023; Folke and Rickne 2022), segregation has often been analyzed from the point of view of material inequalities, such as pay, working hours, and overtime. Previous research focused on studying the different opportunities men and women have to access certain occupations deemed more desirable in terms of salaries (Torre 2014, 2019; Jacobs 1989; Baxter and Wright 2000), or it studied how specific working arrangements provide a more hostile environment for women working in male-dominated occupations (Nemoto 2013; Cha 2013). We chose to focus on a different dimension of segregation, which relates to work strains rather than material rewards. This offers us a more complete view of segregation as a system of social control in which both material resources and exposure to work strains in the workplace differ between men and women in diverse occupations.

Finally, one issue of relevance for future research and another of relevance to policy also emerged. Our findings show the need to categorize gender-neutral occupations better. Economic and administrative professions behave in almost the same way as STEM occupations, while the legal-cultural professions are more similar to female-dominated occupations as far as risk exposure is concerned. This suggests the need for a more nuanced categorization of occupations in which researchers do not simply rely on numeric quotas to classify an occupation as segregated but also consider the specific arrangements, tasks, and skills involved in performing the job.

The second aspect is of significance for policymakers. Our findings would seem to indicate a need to reconsider the configuration of interactions between customers and workers. As outlined by Lopez (2010), the customer–worker dynamic might be satisfactory and fulfilling when the former wields comparatively less influence over the latter, which primarily applies to those roles where the “customers” are in vulnerable positions. However, it is imperative to reassess these relationships within different contexts. More specifically, there is currently a call for gender-specific interventions, given the observed greater exposure to risks to workers’ psychological well-being in those occupations where women represent the majority of the workforce.