The labour market is highly segmented by gender. As a result, women are often employed in less skilled, less socially valued and lower-paid occupations than men. Do some population groups manage to escape this gendered economic order? One way to explore this issue is to compare the occupations and sectors of activity of people in same-sex and different-sex unions. Using the 2011 Family and Housing Survey (the first French survey to make this type of comparison), Wilfried Rault examines in this article whether gays and lesbians in a union are less affected by the gender divide in the labour market than other people in a couple. He reveals major differences that reflect both the specific strategies applied by gays and lesbians in various occupations and sectors of activity, and the obstacles they may face. This article also highlights the particular situation of women in different-sex unions whose position in the labour market is weaker than that of men or women in same-sex unions.
1Since the 1980s, social science research on homosexuality has expanded rapidly in France, a trend that reflects the rising social visibility of homosexuality, the legal recognition of same-sex unions, and the resulting decrease in stigma. Using interview-based qualitative surveys, numerous studies on sexuality, conjugality, and parenthood have been conducted, partly in response to the politicization of homosexuality brought on by the creation of the civil partnership (PACS) in 1999, the debate surrounding the recognition of same-sex parenting and, more recently, the opening of marriage to gay and lesbian couples in 2013.
2In the field of quantitative research, progress is mixed. Quantitative studies are mainly based on two types of sources: representative surveys on sexual behaviour, and specialized surveys on a sample of volunteers based on the model initiated in the 1980s by Michael Pollak and Marie-Ange Schiltz (1994) at the height of the AIDS epidemic. These surveys made it possible to study the diversity of homo-bisexualities, and more generally, to better understand sexual behaviours and how they evolve over time. It remains difficult, however, to characterize gay and lesbian populations socially, even on the basis of detailed social indicators such as educational qualifications, sector of activity, or occupational category. Two general population surveys – Analyse des comportements sexuels des Français (Analysis of sexual behaviours in France, 1992) and Contexte de la sexualité en France (Context of sexuality in France, 2006) – based on probabilistic samples, revealed that respondents who reported at least one sexual encounter with a person of the same sex in their lifetimes were both younger and more highly educated than the average. It should be noted, however, that this indicator – having had at least one same-sex partner in one’s lifetime – is not an indicator of homosexual orientation (Bajos and Beltzer, 2012; Messiah and Mouret-Fourme, 1993). Surveys based on convenience samples (such as the Gai Pied surveys and the gay press surveys carried out annually from 1985 to 1993, then again in 1995, 1997, 2000, 2004 and 2011 – with the latest edition also addressing women) have always revealed very specific respondent profiles (Pollak, 1988; Rault, 2011; Schiltz, 1998; Velter, 2007): high levels of education, upward social mobility and an over-representation in higher-level and intellectual occupations. The authors ascribe this to the choice of survey administration method (notably paper questionnaires that favour those who are comfortable with writing) and to the effect on the respondents’ life course of having a stigmatized, minority sexual orientation. These observations were made cautiously, as the surveys depended on voluntary participation and did not include detailed indicators of the respondents’ social situations. More recently, routine surveys administered by INSEE, such as the Labour Force survey (Enquête Emploi), have explored certain themes based on the study of individuals in same-sex unions, while taking important methodological precautions, as these configurations are difficult to identify (Laurent and Mihoubi, 2014; Toulemon, 2014; Table 1).
3New and more precise data now allow us to study the social status of gays and lesbians. This article first examines the sources available on the subject, as of the mid-2010s. It shows that the current approach, which aims to characterize gay and lesbian people in terms of education and occupation, brings with it several challenges linked to the representativeness of such surveys, the quality of the indicators, and the low numbers of respondents.
4After this general overview, using data from the Family and Housing survey (Enquête Famille et Logements, INSEE, 2011), a general population survey based on a probabilistic sample, we will propose an alternative approach which explores the hypothesis that social situations may differ by sexual orientation. Unprecedented in its level of detail and its sample size, this approach is based on the analysis of respondents who report being in a union at the time of the survey (whether cohabiting or not). First, we examine the extent to which women and men who report being in same-sex unions are characterized by high levels of education, as also observed in census data from other countries (Canada: Waite and Denier, 2015; United States: Baumle et al., 2009). Drawing upon the work of Pollak and Schiltz, we then put forward the hypothesis of occupational differentiation. Indeed, sectors that require high levels of education are more likely to attract individuals who report being in a same-sex union, if only because of their high level of qualification. We also look for evidence of a differential gender distribution in the various sectors of activity and occupations, which has also been observed in the few North American studies on the subject (Ueno et al., 2013; Waite and Denier, 2015).
5While sectors of activity, occupations and occupational categories are all marked by significant gender divides, the occupations of gays and lesbians may be less polarized. Several mechanisms form the basis of such a hypothesis, though the survey data used here do not allow us to examine their respective importance. From a first perspective, occupational choices reflect the effects of socialization. Thus, for gays and lesbians, the transgression of the heterosexual order could lead to the construction of certain ways of seeing and behaving that make it possible to distance themselves from, and even transcend, society’s social and sexual expectations. In this case, gays and lesbians may be more likely to question “self-evident gendered attributes” (Guichard-Claudic et al., 2008), whether consciously or not. From a second perspective, compatible with the first and mentioned successively by both Pollak and Eribon (Eribon, 2004, Pollak, 1988), distinct occupations, reflecting a strong early investment in education, might reflect the adoption of more strategic behaviours. They may be based on choices that take account of the degree to which various professional environments are perceived to acknowledge the expression of sexuality and other dimensions. Finally, women and men in same-sex unions may have distinct occupations due to their lower projection into gendered familial and marital roles that are more oriented towards the family sphere for women and the breadwinner role for men (Badgett, 2001, Badgett and King, 1997). Thus, regardless of the importance of each of these mechanisms, the sectors of activity and occupations traditionally dominated by one or other sex may be rejected in favour of more mixed or non-gendered sectors.
6These mechanisms are likely to be different for men and women, precisely because of gender norms. For gay men, the desire to distance themselves from spheres associated with the dominant male culture, in which homosexuality may be stigmatized in everyday interactions, [1] could explain their lower investment in these spheres. For lesbian women, these mechanisms may provide a means to avoid heteronormative occupational categories, or a division of labour that follows the asymmetrical organization observed in different-sex couples (Dunne, 1998). One might thus expect to observe lower levels of inactivity, and a lesser presence in less socially valued occupations (such as part-time and low-paid jobs), in which women are particularly numerous.
I – The Family and Housing (2011) survey’s contributions to the study of gay and lesbian social status
7There are many obstacles to acquiring statistical knowledge on gay and lesbian populations, their detailed social status in particular. Available surveys sometimes include variables that can be used to study homosexuality, but also have limitations that compromise the robustness of the reasoning applied. Studying these groups quantitatively, and placing them in a social context with sufficient detail, is dependent on three conditions which have yet to be met simultaneously in a single survey: the existence of satisfactory indicators for addressing both homosexuality and social status, a sample that is sufficiently representative, and a sample of sufficient size. By reviewing the available sources, the various obstacles to this approach can be identified (Table 1).
8The Family and Housing survey (Enquête famille et logements, EFL; INSEE, 2011), a supplementary questionnaire associated with the 2011 census, overcame some of these obstacles by asking the following questions:
- Are you currently in a union?
- Yes, with a person who lives in this dwelling
- Yes, with a person who lives in another dwelling
- No, but I have been in a union before
- No, I have never been in a union
- Concerning the person with whom you are in a union (spouse or partner):
- What is their date of birth?
- Your spouse / partner is:
- □ a woman
- □ a man
9This survey is based on a self-administered questionnaire whose ergonomics and mode of administration are closely modelled on the census questionnaire (though, unlike the census, it is optional). Because of its association with the census, the response rate (83.8%) to the Family and Housing survey is much higher than for general population surveys involving an interviewer, whether they are administered in person or by telephone. Moreover, the survey very explicitly considers same-sex couples, which makes it easier for such couples to report their status, unlike the earlier version of the Family survey which only considered heterosexual unions. [2] The survey’s association with the census also means that the number of respondents is not a problem: administered to 359,770 people aged 18 and older, [3] it recorded a relatively high number of respondents in same-sex unions (730 women and 660 men). Neither is there any problem of representativeness: the persons surveyed were randomly selected from the census sample frame. Finally, the Family and Housing survey provides a large number of indicators concerning social status, which makes it possible to analyse the social background of those surveyed via the occupations of the respondents’ parents. The detailed social characteristics – in particular level of education, detailed occupational category (42 items) and sector of activity (24 items), see Appendix A.1 – are drawn from the individual census bulletin. They are recorded to describe each respondent’s situation at the exact time of the survey; for persons who are no longer economically active, the level of detail of the occupational category is lower (6 items). For this reason, our analysis is restricted to economically active people in the age range of 25-59 years.
10One significant limitation, also observed in other surveys, is that homosexuality is addressed solely in terms of couples. However, these couples are covered much more thoroughly than in previous versions of the survey. The person with whom the respondent is in a relationship is referred to as a conjoint(e)/ami (spouse/partner), the latter term being more in line with the representations of certain population groups who do not identify with labels that stem from traditional matrimonial terminology. Last, non-cohabiting unions (“living apart together”) are also explicitly taken into account. The survey shows that among women and men who report being in a same-sex union, non-cohabiting relationships are more common than among different-sex couples (Rault, 2018). Thus, we propose to examine the social positions of gays and lesbians on the basis of respondents who reported being in a union, a status which not only significantly contributes to the visibility of homosexuality (Courduriès, 2011), but is also one of the main vectors of its societal recognition (Rault, 2009). So the findings of this study do not apply to persons not in a union, who, voluntary surveys suggest, are very numerous.
Contributions and limitations of available sources for studying the social status of gay and lesbian populations


Contributions and limitations of available sources for studying the social status of gay and lesbian populations
II – A level of education that is high, but less distinctive among the younger generations
11The four groups studied (women and men in same-sex unions, and women and men in different-sex unions) are structured differently by age group: lesbian and gay couples are less numerous among older generations (Table 2), a result which likely reflects their greater difficulty in living as a homosexual couple in a hostile and repressive context (Idier, 2013). There are also large disparities in levels of education, with same-sex couples more frequently being highly educated than different-sex couples.
12While same-sex couples are most numerous among the two youngest cohorts (born between 1966 and 1985), who generally continued their education for longer, these results do not adequately explain the higher levels of education among same-sex couples. An examination of the relative share of respondents with higher education by age group shows that the contrast between men and women in different-sex and same-sex unions is observable in all categories (Appendix Table A.1). It looks as if having a high level of education is a factor that facilitates same-sex cohabitation (and reporting said cohabitation in a survey). In line with many studies based on voluntary surveys, another possible hypothesis is that homosexual orientation may favour the accumulation of educational capital, a resource that gives more opportunity to escape the constraints of heterosexual norms. This phenomenon may have been significant for the older generations who were particularly exposed to such constraints, and this might explain why level of education is such a distinctive factor for those born between 1951 and 1965. [4]
13The educational divide is less pronounced in the younger generations. This is likely the result of living in a social context more open to homosexuality, in contrast to the oldest generations who were socialized in a particularly hostile social context. [5] Attitudes towards homosexuality in the population as a whole are highly correlated with generation: people born before the 1960s are much more reserved or even hostile towards homosexuality than younger generations (Bajos and Beltzer, 2012; Rault, 2016a). Over the course of their lifetimes, the generations born after the 1960s, and even more so the 1970s, have benefited from greater openness towards equality of sexualities.
14These higher levels of education may also be due to another structural factor: a more privileged social background. The fact that women and men in same-sex unions are more likely to come from the middle and upper classes (Table 2) suggests that this effect of social reproduction probably does exist. The occupations of respondents’ parents illustrate these differences: people in same-sex unions less frequently have fathers who are manual workers or farmers, and more often have fathers in intermediate occupations, or, in the case of women, in higher-level occupations. Women and men in same-sex unions also more frequently have working mothers, who are most often in intermediate occupations. These contrasts are related to the age structure of the groups studied – the youngest being more numerous among persons in a same-sex union – and to the general evolution of the working population, characterized by a rise in higher-level and intermediate occupations, and female employment in general, and a relative decline in manual workers and farmers. However, they are only partially so: the differences persist after controlling for age (Appendix Table A.4).
Structure by age, level of education, and social background of women and men in same-sex and different-sex unions*

Structure by age, level of education, and social background of women and men in same-sex and different-sex unions*
* Missing data were imputed.Interpretation: 21.7 % of women in same-sex unions have upper secondary education.
Coverage: Women and men aged 25-59 in a union.
15While these higher levels of education are not independent of social background, they also reflect greater social mobility. While having parents in higher-level and, to a lesser extent, in intermediate occupations, significantly increases the level of education, being in a same-sex union also correlates with these high levels of education (Table 3). An approach that compares the social mobility of women and men in same-sex unions on the basis of their parents’ occupational categories reveals more frequent social mobility than is observed in different-sex couples. This social mobility is particularly pronounced for women and men from the middle and working classes, while those from the upper classes are more strongly characterized by geographical mobility, most often towards the Paris region (see Rault (2016b), which is also based on the Family and Housing survey, and focuses on the question of social mobility).
Factors associated with the likelihood of having a level of education above upper secondary versus upper secondary or lower, (odds ratios)

Factors associated with the likelihood of having a level of education above upper secondary versus upper secondary or lower, (odds ratios)
Interpretation: A statistically significant odds ratio higher than 1 indicates that for the modalities concerned, when compared to the reference modality of the variable under consideration, the factor increases the chances of belonging to the modelled group. The further the odds ratio is from 1, the greater the influence of the factor in question. For example, for a given age and occupation of the mother, the fact of being in a same-sex union increases the likelihood for a woman that she will report having higher education, compared to a woman in a different-sex union (OR = 1.42).Statistical significance: *** p < 0.001; ** p < 0.01; * p < 0.05.
Coverage: Men and women aged 25-59 in a union.
III – Sectors of activity and occupational categories: distinct and less segregated positions
16These longer educational careers may correlate with different distributions of gay and lesbian women and men in certain sectors of activity, classified here in accordance with the French classifications of sectors of activity (Nomenclature d’Activités Française, NAF) and occupational categories (Professions et Catégories Socioprofessionnelles, PCS). While one would expect a stronger presence in categories requiring a high level of education, [6] they may also be over-represented in other categories. In their analyses of the Gai Pied Hebdo gay press surveys, Pollak (1988) and Adam (1999) noted that certain sectors not characterized by high levels of education, particularly the service industry (e.g. hotels and restaurants), were more likely to be chosen by gay women and men.
17We also put forward the hypothesis that the distribution of same-sex couples across the NAF and PCS categories is less segregated, i.e. that gays and lesbians are more present in categories that are mixed or dominated by the opposite sex, and conversely, less present in categories that are highly segregated and dominated by workers of their own sex.
18Before taking a closer look at these NAF and PCS categories, it is useful to examine the employment statuses of women and men, and their working hours. Here, we find considerable disparities between the groups studied, particularly in the case of women (Table 4). Women in same-sex unions are more frequently employed than women in different-sex unions (83.2% versus 71.8%), and also less frequently work part-time (11% versus 22.1%). These differences are linked to the combined effect of the lesser presence of children in female same-sex unions and their more egalitarian parenting habits (Descoutures, 2010), and to gender norms that often cause women in different-sex unions (more so than men) to leave the labour market or take part-time work.
Different sectors and occupational categories
19Analysis of the NAF and PCS classifications reveals that the social positions of women and men in same-sex unions are very distinctive with respect to their counterparts in different-sex unions. More often than not, these particularities are consistent with our hypotheses (Figures 1, 2, 3 and 4).
20For women, contrasts are observable in several areas. Women in same-sex unions are more likely than women in different-sex unions to work in information and communication (3.7% versus 1.5%), science and technology (7% versus 4.3%), education (13.7% versus 8%), and arts and entertainment (2.3% versus 0.9%), but also in the accommodation and food service industries (4.7% versus 2.5%), characterized a lower level of education. Conversely, women in same-sex unions are much less numerous in the “Other” category, which mainly includes people who do not have a NAF because they do not work. It is on this last point that the difference between the two groups is the most marked. However, a review of NAF categories excluding those who do not work shows that while most of the disparities remain, they are somewhat smaller (data not shown here).
Employment status of women and men by type of union (%)

Employment status of women and men by type of union (%)
Interpretation: 11% of women in same-sex unions work part-time.Coverage: Men and women aged 25-59 in a union.
21For men, these contrasts are more pronounced and less structured around the distinction linked to employment status. The most striking difference is in the field of manufacturing and energy-related occupations: nearly one-fifth of the men in different-sex unions reported working in these sectors (18%), compared with just 6.2% of men in same-sex unions. The under-representation of men in same-sex unions is also visible in other categories, such as agriculture (0.6% versus 3.3%) and construction (2.1% versus 10.3%). Conversely, men in same-sex unions are over-represented in several sectors. As with women in same-sex unions, these men are more likely than their heterosexual counterparts to work in information and communication (6.2% versus 3.3%), science and technology (8.5% versus 5,0%), education (7.1% versus 4.1%), arts and entertainment (3.0% versus 1.1%), and the accommodation and food service industries (7.0% versus 2.7%). These contrasts confirm the observations made on various samples of volunteers. However, in certain sectors it is only men in same-sex unions who are over-represented. These include health (8.6% versus 4.5%) and the other service occupations (4.6% versus 1.7%) characterized by a lower level of education on average.
Women’s sectors of activity, by type of union (%)

Women’s sectors of activity, by type of union (%)
Interpretation: 4.7% of women in same-sex unions are employed in the “accommodation and food services” sector. The results have a 95% confidence interval.Note: The green squares correspond to the categories in which women in same-sex unions are relatively under- or over-represented.
Coverage: Women aged 25-59 in a same-sex or different-sex union.
22These contrasts are even more striking when comparing occupational categories. For women, the most considerable difference concerns those who have never had a job: while 15% of women in different-sex unions have never worked, the proportion is just 2.2% of women in same-sex unions. Similarly, part-time work is much less common among women in same-sex unions (12.5% versus 30.2%). As with the NAF categories, a comparison based solely on women who are currently employed shows that most of the particularities observed in the occupational categories remain. Women in same-sex unions are under-represented in the personal services industry, where they are three times less numerous than women in different-sex unions (3.1% versus 10%) and in private-sector clerical occupations (3.9% versus 9.5%). On the other hand, women in same-sex unions are over-represented in a large number of categories: [7] artisans, merchants and business owners (4.7% versus 3.1%), civil service professionals (2.4% versus 1.3%), teachers or scientific professionals (6.3% versus 2.8%), business managers (6.3% versus 3.7%), private-sector technical executives (2.9% versus 1.6%), and in some sectors characterized by lower levels of education: health associate professionals (11% versus 6.8%), civil service associate professionals (3.7% versus 2.1%), police and military (2% versus 0.4%), skilled manual workers (4.1% versus 1.9%) and drivers and freight handlers (2.7% versus 0.7%). Gender segregation may be weaker in these areas – a hypothesis that will be explored later in this article.
Men’s sectors of activity, by type of union (%)

Men’s sectors of activity, by type of union (%)
Interpretation: 7% of men in same-sex unions are employed in the “accommodation and food services” sector. The results have a 95% confidence interval.Note: The green squares correspond to the categories in which men in same-sex unions are relatively under- or over-represented.
Coverage: Men aged 25-59 in a same-sex or different-sex union.
Women’s occupational categories, by type of union (%)

Women’s occupational categories, by type of union (%)
Interpretation: 6.3% of women in same-sex unions are private-sector executives. The results have a 95% confidence interval.Note: The green squares correspond to the categories in which women in same-sex unions are relatively under- or over-represented.
Coverage: Women aged 25-59 in a same-sex or different-sex union.
23For men, there are several similar disparities: men in same-sex unions are relatively more numerous among teachers and scientific professionals (5.2% versus 2.3%), information, artistic and cultural professionals (4% versus 1%), and business managers (10.5% versus 5.7%).
24In some fields, it is specifically men in same-sex unions who are over-represented: private-sector clerical workers (4.1% versus 1.8%), sales workers (3.9% versus 1.3%), personal services workers (4.2 % versus 1%). Several of the contrasts observed among men are symmetrical to those found among women: far fewer men in same-sex unions are technicians and supervisors (5.7% versus 10%), farmers (0.2% versus 2.4%), police or military personnel (0.6% versus 2.8%) manual workers, either skilled (5.8% versus 14.2%) or unskilled (3.5% versus 9.4%), or drivers and freight handlers (3.4% versus 7.2%) than men in different-sex unions.
Men’s occupational categories, by type of union (%)

Men’s occupational categories, by type of union (%)
Interpretation: 11.7% of men in same-sex unions are private-sector associate professionals. The results have a 95% confidence interval.Note: The green squares correspond to the categories in which men in same-sex unions are relatively under- or over-represented.
Coverage: Men aged 25-59 in a same-sex or different-sex union.
25After controlling for age, level of education and social background, most of these particularities persist. The test that we implemented (see Appendix Tables for presentation and results) enabled us to better capture the specific characteristics of same-sex couples (more highly educated, slightly more frequently from the middle and upper classes, more often childless, less present in the highest age group, etc.) with respect to people in different-sex unions. This test entailed matching women and men in same-sex unions (the analysed group) to women and men in different-sex unions who had the same age, level of education, and social and family background (control group). This method shows that for a given set of characteristics, the career choices of women and men in same-sex unions are always distinct.
Less gender segregation
26A comparative analysis of the sectors of activity and occupational categories of lesbians and gays in a union with respect to the gender distributions in these categories sheds light on the particularities observed in this study (Figures 5 and 6). We find that women and men in same-sex unions are more likely to transcend the gender norms that lead to the over- and under-representation of women and men in certain sectors. The gender distribution for each sector is represented in the figures, along with the under-representation (green border) or over-representation (black border) of people in same-sex unions in the sector in question. Lesbians and gays are over-represented in some mixed categories, and are present in similar proportions in others (a phenomenon represented by the absence of borders in the figures). In highly segregated sectors and occupational categories, people in same-sex unions are under-represented in categories dominated by members of their own sex, and over-represented in those dominated by members of the opposite sex. In other words, for women and men in same-sex unions, gender segregation is less pronounced than is the case in different-sex unions. While there are exceptions, they most often concern sectors (such as information/communication for men, and education for women), or occupational categories that typically require a high level of education. These are also categories in which people of the opposite sex who are in same-sex unions are also over-represented.
27An analysis in terms of major occupational categories and their composition by sex shows that the distribution of occupations held by women and men in same-sex unions is generally less segregated (Table 5). This approach makes it easier to take level of education into account (Table 6) in order to verify that the observed specificities are still valid after controlling for educational level, the higher-level occupational categories also being the most mixed.
28Based on the occupational categories used in this study, three main groups can be distinguished: those numerically dominated by men (more than 60% men), those that are relatively mixed, with 40-60% of each sex, and those numerically dominated by women (more than 60% women) (Table 5). The occupational categories of gays and lesbians in a union appear to be distributed differently across these three groups. Put together, all of the mixed-sex categories account for 20.6% of women and 23% of the men in different-sex unions; the proportions are much higher for women in same-sex unions (30.7%), and even more so for gay men (38.7%). All female-dominated occupational categories are characterized by a clear under-representation of women in same-sex unions (36.3% versus 52.8%) and an over-representation of men in same-sex unions (23.2% versus 12.8%). These trends mirror those of male-dominated categories, which are relatively more frequent for women in same-sex unions (33% versus 26.6%) and less frequent for men in same-sex unions (38.1% versus 64.2%). For both women and men, the categories dominated by their own sex are less frequently represented than the other two groups, i.e. the mixed-sex categories and, to a lesser extent, the categories dominated by the opposite sex.
Distribution of women and men in a union, by sector of activity and under-/over-representation of women and men in same-sex unions

Distribution of women and men in a union, by sector of activity and under-/over-representation of women and men in same-sex unions
Note: Black borders indicate an over-representation of women (left) or men (right) in same-sex unions in the given sector, and dotted green borders indicate an under-representation. The numbers in parentheses for each sector indicate the percentage of the active population in each sector.Interpretation: The information and communication sector is dominated by men (67%). In this sector, women and men in same-sex unions are over-represented compared to women and men in different-sex unions (black border).
Coverage: Women and men aged 25-59 in a same-sex or different-sex union.
Distribution of women and men in a union by occupational category and under-/over-representation of women and men in same-sex unions

Distribution of women and men in a union by occupational category and under-/over-representation of women and men in same-sex unions
Note: Black borders indicate an over-representation of women (left) or men (right) in same-sex unions in the given sector, and dotted green borders indicate an under-representation.Interpretation: The “teachers, scientific professionals” occupational category is composed of 56% women and 44% men. In the same category, women and men in same-sex unions are over-represented with respect to women and men in different-sex unions (black border).
Coverage: Women and men aged 25-59 in same-sex or different-sex unions.
Distribution of men and women in same-sex and different-sex-unions by gender profile of occupational categories

Distribution of men and women in same-sex and different-sex-unions by gender profile of occupational categories
Interpretation: 30.7 % of women in same-sex unions work in a mixed occupational category, versus 20.6% of women in different-sex unions.Coverage: Women and men aged 25-59 in same-sex or different-sex unions.
29These distributions can be broken down by level of education (Table 6) to analyse the trends in more detail. The patterns are very different, depending on the gender and level of education of the person in question: for individuals with modest academic capital (below upper secondary), occupational categories are particularly segregated, with mixed-sex categories being quite rare. The differences are large in the most gendered categories and always in the direction previously observed, but even more markedly so: women in same-sex unions are less present in female-dominated categories (27.7% versus 56.9%) and, conversely, much more present in those dominated by men (64.7% versus 34.1%). This differential polarization for women in same-sex unions with an educational level below upper secondary is due in part to the fact that they are much more frequently in employment than women in different-sex unions, and are less likely to be in a part-time job, part-time working being more common among women with lower levels of education. It could also be due to large differences in parental investment between different types of union. Parenthood is much more frequent among different-sex couples, with parenting roles exercised asymmetrically by men and women (Descoutures, 2010) at the expense of the woman’s employment. Moreover, the phenomenon intensifies as the number of children increases (Champagne et al., 2015). It is difficult to study this aspect in detail, and to take the number of children into account, as relatively few women in same-sex unions report living with children. However, disparities are also large in cases where women report having only one or no children in their homes: for women in same-sex unions, full-time employment is much more frequent and part-time work or inactivity much less so. [8]
Distribution of men and women in same-sex and different-sex-unions by gender profile of occupational categories and by level of education (%)

Distribution of men and women in same-sex and different-sex-unions by gender profile of occupational categories and by level of education (%)
Coverage: Women and men aged 25-59 in a same-sex or different-sex union.Interpretation: 64.7 % of women in same-sex unions with lower secondary education work in an occupational category dominated by men (> 60%). These women represent 17.4% of all women in same-sex unions.
30The cumulative effect (Maruani, 2006) for women in different-sex unions, characterized by both vertical segregation (with more women working part-time or inactive) and horizontal segregation (presence in specific sectors that are highly gendered and socially less valued), is less pronounced for women in same-sex unions. Their greater presence in occupations dominated by men gives them access to less precarious and more socially valued jobs. From this point of view, being in a different-sex union is more likely to produce gender inequalities for women. This contrast echoes the findings of demographic studies conducted in the United States on income and same-sex couples. Using the 1995 and 2000 United States censuses, these studies found that women in different-sex unions had lower incomes than women in same-sex unions. However, the latter still reported lower incomes than men. Among men, those in different-sex unions tended to have higher levels of income than men in same-sex unions (Baumle et al., 2009, Black et al., 2000). [9] The composition of the occupational categories of gay men with low levels of education followed the same logic, but to a lesser degree than women: gay men are relatively more numerous in occupations numerically dominated by women (and are under-represented in those dominated by men), but the differences are less pronounced than those observed for women.
31For men in same-sex unions, distancing themselves from male-dominated occupations may reflect a dual avoidance strategy: first, an avoidance of sectors that are potentially more conducive to the expression of hostility towards gay men, as men are generally less accepting of homosexuality than women (Bajos and Beltzer, 2012); and second, an avoidance of sectors where professional identities are often based on representations of masculinity that denigrate male homosexuality, and see it as incompatible with the sectors in question. [10] For gay men, opting for female-dominated occupations may therefore have a lower relative cost than for men in different-sex unions. Moreover, the glass escalator mechanism (Williams 1992), whereby male career advancement is faster in sectors where men are in a minority, may also work in their favour (Buscatto and Fusulier, 2013).
32For each group analysed as a whole (women and men in same-sex unions, and women and men in different-sex unions), the extent to which gays and lesbians opt for occupations dominated by the opposite sex should nonetheless be put into perspective. While 65% of women in same-sex unions with a lower secondary educational level are employed in a sector numerically dominated by men, they represent only 17.4% of all women in same-sex unions. Women in different-sex unions with these same characteristics (34%) account for 15.2% of all women in different-sex unions. In this regard, the differences are small. Likewise, 6.6% of men in same-sex unions, and 5.5% of men in different-sex unions have a lower secondary educational level and work in a female-dominated occupational category.
33Among the most educated respondents, the distribution follows a different pattern: people in same-sex unions have a strong presence in mixed categories. Nearly 60% of highly educated women and 64% of highly educated men in same-sex unions were employed in these categories, versus 49% and 52%, respectively, in different-sex unions. The former are relatively less numerous in the occupations dominated by their own sex, and their presence more marked in those dominated by the opposite sex. Each occupational category dominated by one or other sex is represented in fairly similar proportions for each type of couple: all of the male-dominated categories account for 12.7% of all women in same-sex unions and 15.4% in different-sex unions. All of the female-dominated categories account for 12.4% of men in same-sex unions, and 11.2% in different-sex unions.
34The contrasts in the distribution of occupational categories of women and men in same-sex unions are smaller than for women and men in different-sex unions. The former are less present in the categories dominated by their own sex than in mixed categories and, to a lesser extent, categories dominated by the opposite sex. This distribution is indicative not so much of a “gender reversal” (Guichard-Claudic et al., 2008) but rather of a lesser – though still measurable – degree of segregation.
Conclusion
35The Family and Housing Survey provides an opportunity to analyse both the diversity and particularities of the employment situations of gays and lesbians in same-sex unions. Their over-representation in sectors of activity and occupational categories with a mixed gender profile and their less asymmetrical presence in highly male- or female-dominated categories than men and women in different-sex unions reflects the distancing of same-sex couples from gender norms. Whether or not these specificities are seen as the expression of personal strategies, the effects of socialization, or the result of gender norms that weigh directly upon conjugal and family configurations, this approach emphasizes the value of not limiting the analysis of minority sexualities to the angle of sexual behaviour alone, and of “desexualizing” homosexuality, as proposed by Simon and Gagnon (1967) in response to approaches that emphasized deviance in the late 1960s. This study must be supplemented by further qualitative and quantitative analyses to identify the mechanisms that lead to these differences. It is also important to better understand the possible reporting effect of such a survey: certain socially privileged individuals may be more willing to report being in a same-sex union than others. In this respect, the observations described in this article do not suggest that all gays and lesbians occupy more privileged positions; this is the case for only a share of their total population. And indeed, this does not imply the absence of discrimination (see studies cited above).
36Moreover, this approach, limited here by a necessary restriction of the analysis to people in a union, serves not only as a contribution to gay and lesbian studies, but also encourages reflection on the effects of sexual orientation on the gendered structure of society. The examination of men and women in same-sex and different-sex unions thus highlights how heterosexuality – not just gender – is associated with more differentiated social and professional roles. In this regard, it is the comparison of the two groups of women that reveals the most marked differences. Although these two groups are heterogeneous, women in different-sex unions often have a number of specific characteristics, such as employment in strongly female-dominated and socially undervalued occupations, part-time working, and inactivity, that are not observed among women in same-sex unions. According to Nicky Le Feuvre (2008), certain life experiences such as lone parenthood, being single, and migration are likely to direct women towards an employment profile similar to that of “breadwinners”, and to adopt career strategies that are typically associated with men. The Family and Housing survey shows that a private life structured around a same-sex partnership is a form of life experience that gives rise to “transgressive acts” (Le Feuvre, 2008) with regard to the gendered division of labour. It is thus the institution of heterosexuality (Rubin, 1988) that could be accused of favouring the production of asymmetric, differentiated and ultimately less egalitarian social roles.
Acknowledgements
The author is grateful to the Population reviewers for their helpful remarks and suggestions. He would also like to thank Marc Thévenin for his methodological support and Yaëlle Amsellem-Mainguy and Arnaud Régnier-Loilier for their constructive comments on an initial version of this article.Appendix A.1. French classification of sectors of activity (NAF) and occupational categories (PCS)
NAF classification
37The French classification of sectors of activity (Nomenclature d’Activités Française, NAF) is the national statistical classification system for economic activities. It is based on the information provided in the individual census form. This classification system is made up of 5 levels, with 732 items in the most detailed level. The first level, which includes 21 sections, is used in this study. We have grouped some activities together because of their similarity (sections B, C, D and E were grouped together, as were sections K and L).
38http://recherche-naf.insee.fr/SIRENET_Template/Accueil/template_page_accueil.html
PCS classification
39The system of occupational categories (Professions et Catégories Socioprofessionnelles, PCS) is a statistical system for classifying occupations and trades. This system was created by the National Institute of Statistics and Economic Studies (INSEE) in 1982, and replaced the occupational categories (Catégories Socioprofessionnelles, CSP) system first established in 1954.
40The occupational category is determined by one’s occupation, i.e. what an individual does at his or her place of work, and the social status linked to that occupation: employment status (self-employed or salaried), position in the occupational hierarchy, employer (private or public), and sector of activity. The third level of the PCS system (which includes 42 occupational categories) is used in this analysis. Some activities are grouped because of their similarity, especially when the number of respondents was particularly low and/or when certain sectors were related. The categories of farmers, skilled manual workers and unskilled manual workers were grouped together, as well those of artisans, merchants and business owners.
Appendix A.2. Analysis of sectors of activity and occupational categories using an exact matching method [11]
42This approach [12] consists in matching the individuals in the groups analysed (women and men in same-sex unions) with those in the control groups (women and men in different-sex unions) who have the same characteristics in terms of age, social background (proxied by the occupation of the respondents’ parents), and whether or not they have children. The aim is to examine whether groups with similar characteristics nonetheless show disparities with regard to sectors of activity, occupations, and occupational categories when the type of union (same-sex or different-sex) is taken into account. The effects of several factors can thus be considered jointly rather than in an isolated manner, something that a logistic regression renders impossible by its very nature.
43All of the individuals in the control group who have similar characteristics are matched (one-to-many matching) using a method that excludes fewer observations than strict one-to-one matching. Similarly, matching was conducted on five-year age groups, and not on specific ages. Weighting based on the number of individuals in the control group was applied to take account of the fact that each individual in a same-sex union is matched with a different number (ki) of individuals in different-sex unions (a weight of 1/ki). After grouping together similar sectors of activity and occupational categories, the technique was applied to produce Tables A.1 and A.2. In all the configurations studied, the specificities of individuals in same-sex unions – men especially – can be observed, although they are less marked than those found in the results mentioned in the article.
44In total, 97.1% of the women in same-sex unions (the analysed group) were matched to 37.3% of the women in different-sex unions (the control group), and 99.2% of the men in same-sex unions (the analysed group) were matched to 56.7% of the men in different-sex unions (the control group).
NAF code by sex and type of union, after matching

NAF code by sex and type of union, after matching
Coverage: Women and men aged 25-59 in same-sex unions, matched with individuals in different-sex unions with similar characteristics.Occupational category by type of union, after matching

Occupational category by type of union, after matching
Coverage: Women and men aged 25-59 in same-sex unions, matched with individuals in different-sex unions with similar characteristics.Post-secondary qualification by type of union, age group (%), and hazard ratio

Post-secondary qualification by type of union, age group (%), and hazard ratio
Interpretation: 43.8% of women aged 45-59 in a same-sex union have a post-secondary qualification (bachelor’s degree or higher). These women more often have a post-secondary qualification than women of the same age group in different-sex unions (HR = 2.44).Coverage: Women and men aged 25-59 in a union.
Factors associated with being in a same-sex (versus different-sex) union, (odds ratios)

Factors associated with being in a same-sex (versus different-sex) union, (odds ratios)
Interpretation: A statistically significant odds ratio (higher than 1) indicates that for the modality studied, when compared to the reference modality of the variable under consideration, the factor increases the chances of belonging to the modelled group. The further the odds ratio is from 1, the greater the influence of the factor in question.Significance: *** p < 0.001; ** p < 0.01; * p < 0.05.
Coverage: Women and men aged 25-59 in a same-sex or different-sex union.
Notes
-
[*]
French Institute for Demographic Studies. Correspaondence: Wilfried Rault, Institut national d’études démographiques, 133 Boulevard Davout, 75980 Paris Cedex 20, email: wilfried.rault@ined.fr
-
[1]
For examples, see Bernstein and Swartwout (2012), Connell (1995), Eribon (2004), and Pruvost (2008).
-
[2]
The 1999 Family History study (Etude de l’histoire familiale, EHF) addressed women by referring to their conjoint (the male noun for spouse or partner) and addressed men by referring to their conjointe (the female noun). For more information on the EHF as a source for studying same-sex couples, see Toulemon et al., 2005.
-
[3]
The Family and Housing survey is based on a sample comprising two-thirds women and one-third men. This numerical imbalance between male and female respondents was deliberate, as survey’s intention was to question twice as many women as men. This characteristic is linked to the history of Family surveys (Desplanques, 2005), with EFL being the latest version. Originally designed to measure women’s fertility, the survey did not start to include men until 1999.
-
[4]
It would also be useful to distinguish those who identified as homosexual during or before their education from those who began to identify as homosexual later in life. For the former, a higher level of education would support the idea of life choices structured by homosexuality. A lack of difference between the two groups would rather indicate that a high educational capital is a social condition for living in a same-sex union (and reporting said union). In the absence of an adequate indicator in the survey, we cannot take this idea any further.
-
[5]
For changes in the social acceptance of homosexuality, see Mossuz-Lavau (2002).
-
[6]
NAF categories: information and communication; health; education; finance, insurance, and real estate; scientific and technical; arts and entertainment.
PCS categories: independent professionals; scientific professionals and teachers; managers; primary school teachers; health associate professionals; information and cultural professionals. -
[7]
This asymmetry between the numbers of categories dominated by women and by men is largely due to the structure of the classification system: there are fewer female-dominated categories than male-dominated categories (Amossé, 2004).
-
[8]
This tendency was confirmed after controlling for educational level and age. The matching methodology used (presented in the Appendix), which takes into account the presence or absence of children, also supports these observations (Appendix A.2).
-
[9]
In France, few studies have been carried out on the subject. Using data from the Labour Force survey (enquête Emploi) and an indirect identification of same-sex couples, Laurent and Mihoubi (2013) show that elevated social status is associated with forms of discrimination, in line with North American studies on the subject (Badgett and King, 1997, Badgett et al., 2007). The INSEE tax and social revenues survey (enquête Revenus fiscaux et sociaux) could contribute new elements from a study on couples (rather than individuals) in the future.
-
[10]
See, for example, Geneviève Pruvost ‘s work on the police (2007, 2008). Pruvost also shows how the presence of women in a given sector can sometimes help to attenuate certain forms of hypermasculine homosociability based on sexist attitudes and praise of (hetero)sexual performance. However, some women choose to enter the police force precisely in order to do a “man’s job” (Pruvost, 2008).
-
[11]
Method implemented by Marc Thévenin and Wilfried Rault.
-
[12]
For more details on the application of this method to the analysis of small groups versus large control groups in the social sciences, see Ichou (2013).