1Until the early 2000s in France, surveys on couple and family life did not include questions concerning homosexuality. Before that time, only surveys specifically focused on sexuality included questions on the subject. This omission reflects the attitudes of the times, before the legal and social recognition of same-sex couples. For example, in 1999, the year that the PACS (civil solidarity pact)  was implemented, the version of the Family Survey (enquête Famille) associated with the census, the study on family history (Étude de l’histoire familiale) only considered the possibility of heterosexual couples, making it difficult to study same-sex couples (Toulemon et al., 2005). During the same period, the French National Institute for Statistics and Economic Studies (INSEE) recoded census data in ways that contributed to the invisibility of these couples (Digoix et al., 2005). Since then, the headings of the questions in the Family Survey have changed, and the latest version, the Family and Housing Survey (EFL; INSEE, 2011), explicitly mentions the possibility of having a same-sex partner. Couplehood has also been approached more broadly, taking into account non-cohabiting relationships.
2The EPIC survey on couple formation (Étude des parcours individuels et conjugaux), performed in 2013–2014 by INED and INSEE, maintains this broader scope. This survey allowed respondents to report couple relationships and serious intimate relationships with same-sex partners. This contrasts with INED’s two previous surveys on couple formation in France (Le choix du conjoint, in 1959, and La formation des couples, in 1983–1984), which were performed in a context where homosexuality was largely invisible and highly stigmatized, or even punished (Idier, 2013). The EPIC survey, which was designed to capture respondents’ individual trajectories better, approaches homosexuality through several different variables: being in a couple or serious intimate relationship with a person of the same sex, having been in one in the past, as well as having ever had sexual relations with a person of the same sex. In doing so, it follows on from previous surveys about couple formation and connects to major general population surveys on sexuality. 
3Comparisons between previous surveys—the Simon survey (1970), the Analysis of Sexual Behaviour in France (Analyse des comportements sexuels en France, ACSF, 1992), and the Context of Sexuality in France (Contexte de la sexualité en France, CSF, 2005–2006)—revealed a progressive increase in the reporting of at least one same-sex sexual encounter during a lifetime, which was particularly marked for women between 1992 and 2005 (Bajos and Beltzer, 2008). The EPIC survey allows us to update these statistics and to study the reporting of homo-/bisexual practices after the opening of marriage in France to same-sex couples with the law of 17 May 2013.
4The first part of this paper examines the extent of change in the reporting of being in a same-sex couple as well as same-sex sexual encounters, between 2005 and 2014. The second part analyses the individual social characteristics which are associated with reporting at least one same-sex sexual partner over a lifetime. The aim is to observe whether, despite a context that is seemingly less hostile to homosexuality, this type of reporting continues to be associated with a certain level of sociocultural capital, as shown by analyses performed on data from the previous survey on sexual behaviour (Bajos and Beltzer, 2008). A third part explores the diversity of these homo-/bisexual trajectories, how they fit within individual trajectories, and their contrast with strictly heterosexual trajectories. By ‘homo-/bisexual trajectories’, we refer here to respondents who reported at least one same-sex partner over their lifetime, and not to women and men who identify as homosexual or bisexual.  This group is highly heterogeneous, as it includes all those who did not exclusively report different-sex partners, whether their same-sex practices made up all, a majority, a minority, or a single exceptional part of their sexual and intimate relationship history. A final part looks at the normative worlds of this group in the domains of conjugality and sexuality: the EPIC survey highlights the extent to which these can differ from those of the women and men who report exclusively heterosexual trajectories.
I – Increased reporting between 2005 and 2014
5The EPIC survey can be used to study the proportion of individuals in a same-sex couple and compare it to previous surveys. Nine years after the CSF survey, the questions on sex and the number of sexual partners also allow us to update the percentage of people reporting that they have ever had sexual relations with a person of the same sex.
1 – Being in a same-sex couple
6A look at the results of three successive surveys covering same-sex sexuality and relationships—CSF, EFL,  and EPIC—reveals several phenomena (Table 1). The first is that men more often report being in a same-sex couple than women, regardless of the survey and the terminology used, which varied substantially between these sources (Table 2). Age disparities are also significant; this response is much more common among the youngest respondents. It is tempting to compare these three sources, but doing so is not straightforward for several reasons. First, the three surveys were administered in three different ways: in the first case, by telephone; in the second, by a self-administered questionnaire dropped off and collected by a census enumerator; and in the third, with a face-to-face interview or, less frequently, by telephone. Each of these methods may have had its own effects. Telephone interviews and self-administration, given the associated level of confidentiality, are more favourable for this type of question than face-to-face interviews.
7Secondly, the frequency of reporting may be linked to the type of survey as well as to its topic. It might be easier for respondents to mention a same-sex relationship in the context of a survey on sexual behaviour than in one focused on other topics (EPIC)—all the more so in the context of the census, seen as a population count by the state (EFL, which is associated with the census). In addition, these surveys feature different indicators. The one used in the Family and Housing Survey is more restrictive than in the other two because it is more explicitly based on the notion of the couple. From this perspective, the lesser frequency at which this conjugal type of relationship was reported in that survey is not surprising, as gays and lesbians tend to have a substantially different relationship to conjugality than the majority population. Looking in detail at the configurations reported in the Family and Housing Survey, it is apparent that being in a couple ‘with a person who does not live in the dwelling is much more common among same-sex than different-sex couples (Rault, 2018). The two other surveys, in referring to ‘relationships’ (labelled as ‘stable’ or ‘intimate’), include a wider range of situations. The proportion of noncohabiting relationships seen in the EPIC survey among individuals aged 26–65 years was markedly higher than in the EFL, notably for men: more than a third of intimate or couple relationships reported in the former were non-cohabiting relationships versus 13% in EFL. For women, this proportion was 17% in the EPIC survey versus 12% in the EFL. Finally, these three surveys were performed at intervals of a number of years, in different political and social contexts.
Proportion of women and men who reported being in a same-sex relationship in 2005–2006 (CSF), 2011 (EFL), and 2013–2014 (EPIC)
Proportion of women and men who reported being in a same-sex relationship in 2005–2006 (CSF), 2011 (EFL), and 2013–2014 (EPIC)Interpretation: In the EPIC survey, 1.2% of men aged 26–65 years reported being in a couple relationship or serious intimate relationship with a person of the same sex. CI = confidence interval.
Coverage: Women and men aged 26 to 65 years. The age group is that of the age field in the EPIC survey.
2 – Having had same-sex partners
8Across these different surveys, the reporting of at least one same-sex sexual partner over a lifetime increased. Out of all respondents aged 26–65 years, this proportion went from 3.4% in the CSF survey  to 5%  in EPIC among men and from 3.7% to 6% among women (Figure 1). The observed increase is more marked among women, a trend that extends that observed between the ACSF survey, performed in 1992, and the CSF survey of 2005. But while from the early 1990s to the mid-2000s it seemed that a convergence was occurring between women and men, what is striking here is that the total frequency is higher in women than in men.
9The results of the EPIC survey thus fit with trends observed in major surveys in other countries. Surveys in both the United States (Twenge et al., 2016) and the United Kingdom (Mercer et al., 2013) have shown that the reported frequencies have increased in both sexes over the last decades and that it has become more common for women than men to report having had a same-sex partner (and in slightly higher proportions than in France). In the United Kingdom, this reporting is drawn from the National Surveys of Sexual Attitudes and Lifestyles, or Natsal, performed in 1990, 2000, and 2010, on the basis of several indicators. Over this period, the proportion of men aged 16 to 44 years reporting ‘[a]ny sexual experience with genital contact with a partner of the same sex’ went from 3.6% to 5.4% to 4.8%. Among women, the proportion went from 1.8% to 4.9% to 7.9% across the same period. In the United States, this figure, based on the General Social Survey, administered to people aged 18 to 96 years, is higher than in France: in the early 2010s, 12.7% of women and 9% of men aged 30 to 39 years reported having had a same-sex partner (versus 7.6% and 7.2% a decade earlier), despite this figure having been drawn from answers to a more restrictive question on the number of partners since the age of 18 (Twenge et al., 2016).
10Some respondents did not wish to answer a question on their sexual partners: in the EPIC survey, the refusal rate was 7.7% among men and 7.2% among women (this rate is particularly high among the oldest respondents: 11.5% of men and 11.8% of women aged 56–65 years). The rate in the CSF survey was lower. The very nature of the two surveys may have had an effect here for several reasons: people who accepted to participate in the CSF survey were likely more prepared for questions on sexuality than those who responded to the EPIC survey. Similarly, the CSF questionnaire was much more focused on practices and representations related to sexuality. In this context, questions on the number of partners over a lifetime may have seemed less intimidating, especially since they were questioned over the telephone. As for EPIC, most questions were asked face-to-face, which may have led to embarrassment. However, the protocol for collecting responses to questions on the number of sexual partners of each sex and on the age at sexual debut was different: for these questions, the computer was passed to the respondent, who answered directly and without any discussion on the subject.  Non-responses were imputed based on the sociodemographic profiles of the individuals who did not wish to answer the question.  This made little difference to the results, with the exception of men in the youngest age group (Figure 1).
Reporting of at least one same-sex sexual partner over a lifetime in 2005–2006 (CSF) and 2013–2014 (EPIC) (%)
Reporting of at least one same-sex sexual partner over a lifetime in 2005–2006 (CSF) and 2013–2014 (EPIC) (%)Interpretation: In 2005–2006, 3.1% of women aged 46-55 years reported that they had had at least one same-sex sexual partner over their lifetime. This proportion increased to 6% in 2013–2014 before imputation of non-responses and refusals, and 6.2% after imputation.
Coverage: Women and men aged 26 to 65 years.
11Unlike the EPIC survey, the CSF included other indicators that offer information on sexual relations with a same-sex partner over a lifetime.  Taking these into account significantly increased the reported frequency with respect to the figures given here. It may be assumed that if the EPIC survey had included these indicators, the observed percentages would have been higher.
II – Distinct social characteristics
12Detailed examination of the number of same-sex partners reported in the EPIC survey (Table 3) brings out two important differences between women and men. Men are more likely to mention multiple partners. Among respondents having had sexual relations with a same-sex partner, half of men (versus a little over a quarter of women) reported at least five partners. In contrast, 45% of women (versus 30% of men) reported having had only one. Another difference is that men were more likely to report having exclusively had same-sex partners (1% versus 0.3% of women). These observations parallel findings in the American research cited above.
Individuals reporting same-sex sexual partners over a lifetime, by sex and age group (%)
Individuals reporting same-sex sexual partners over a lifetime, by sex and age group (%)Interpretation: Of women aged 26–35 years, 6.2% reported having had at least one same-sex partner over their lifetime, and 0.6% of women in the same age group had had only same-sex partners over their lifetime. Men in the same age range were slightly less likely to report having had a same-sex partner over their lifetime (4.5%) but more likely to report exclusively same-sex partners (1.0%).
Coverage: Women and men aged 26 to 65 years, weighted data, imputed in case of refusal or non-response.
13Women are more likely than men to report having had both same- and different-sex partners. This result was already discernible in the CSF survey (Chetcuti et al., 2012). In the absence of more detailed data, it is difficult to account for this difference using the EPIC survey. Nonetheless, a number of observations can be made. The greater frequency of exclusively same-sex sexual trajectories in men could be linked to the greater social visibility of male homosexuality, in contrast to female homosexuality, which is less visible and more often associated with bisexuality. The CSF survey (2005) and the Violence and Gender Relations survey (Virage, INED, 2015) show that gay and bisexual men are more likely to have a first same-sex partner than lesbian and bisexual women (Trachman et al., 2018). Other sources show that in terms of identity, men are more likely to describe themselves as gay than women are to define themselves as lesbian (for France, see Bajos and Beltzer, 2008; for the United Kingdom, Mercer et al., 2013; for Italy, La Fauci, 2016). On the other hand, women are more likely than men to describe themselves as bisexual and more likely to report attractions toward people of the same sex. Perhaps reporting one’s desire for or practices with people of both sexes is relatively more acceptable for women (Trachman et al., 2018). These disparities could also be closely linked to available models for identification, which are different for gay men and women. As noted by Natacha Chetcuti-Osorovitz and Gabriel Girard (2015), ‘among gays, social and cultural identifications are found’, whereas ‘the trajectories of most young homosexual women are characterized by a lack of references and a denial of lesbianism’.
14While women are more likely than men to report having had at least one same-sex partner, in both cases this likelihood varies with age. As in the case of couple relationships, those in the oldest age group are markedly less likely to report having had at least one same-sex partner. The sexual socialization of the oldest respondents is more likely to have happened in contexts involving explicit social hostility towards homosexuality: either having or reporting a same-sex partner may be less likely in these cohorts.
15Finally, respondents with higher education were more likely to report having had same-sex partners, and regardless of number for men (Table 4, levels III and IV).  This result fits with the findings of previous studies on homo-/bisexual practices based on general population surveys (Messiah and Mouret-Fourme, 1995; Bajos and Beltzer, 2008), on surveys of volunteer samples (Schiltz, 1998; Adam, 1999; Velter, 2004; Velter et al., 2013 for the most recent versions of the gay press survey), or surveys targeting the social characterization of people who report being in a same-sex couple (Buisson and Lapinte, 2013).
16This observation could be the effect of a number of mechanisms, whose respective importance is difficult to distinguish. There may primarily be a reporting effect: having educational capital may favour both a practice that continues to be socially discouraged in some cases and the reporting of it in a major survey. The relatively high proportion of men who indicate a high level of formal education and who report having had only same-sex partners (Table 4) fits with this hypothesis. Another effect might be linked, for a small portion of the analysed group with a homosexual orientation, to this higher level of education among gays and lesbians (Rault, 2016a).
Reporting of same-sex sexual partners over a lifetime, by sex and level of education
Reporting of same-sex sexual partners over a lifetime, by sex and level of educationInterpretation: Of women without educational qualifications, 3.9% reported having had at least one same-sex partner over their lifetime; 0.2% of women without educational qualifications had had only same-sex partners over their lifetime. Men with the same levels of education were slightly more likely to report having had a samesex partner over their lifetime (4.2%) and more likely to report exclusively same-sex partners (0.6%).
* For definitions of levels of education, see Appendix Table A.1.
Coverage: Women and men aged 26 to 65 years, weighted data, imputed in case of refusal or non-response.
III – A place within diverse sexual trajectories
17Homo-/bisexual practices have a place within a heterogeneous range of sexual trajectories, which, on average, are markedly distinct from those involving exclusive heterosexuality (Table 5). The situations of women and men who report having had at least one same-sex partner over their lifetime are diverse: a minority were in a same-sex couple at the time of the survey (13% of women and 23% of men), whereas most were in a different-sex couple, or, in a lesser number of cases, did not have a partner. A large contrast between the sexes was found: 65% of women who had had a same-sex sexual partner were in a heterosexual couple, versus 48% of men. This observation reproduces the results of previous studies showing that men are much more likely than women to have had only same-sex partners (cf. Section II). Another result is particularly noteworthy: respondents without a partner at the time of the survey reported a significantly higher number of partners than those who were in a couple. For example, among men who had had one or more same-sex partners, those who were not in a couple at the time of the survey had had an average of 26 partners, versus 15 for those in a different-sex couple, and 10 in a same-sex couple. Such differences can also be seen in the results for women. This contrast may reflect a distinction between forms of sexual sociability which are partly structured through conjugality and others that are expressed predominantly outside this context, through a large network of partners.
18Comparing two groups, one of people who have had a same-sex sexual partner and one of people who have not, highlights some very marked differences, particularly for women (Table 5). The median age at sexual debut of women who have had homosexual practices was earlier than for other women: 16 years among respondents aged 36–45 years, versus 18 among those who had only engaged in heterosexual practices. No such differences were found in men. The reported lifetime number of sexual partners also differed between the two groups: it was significantly higher among women and men who had already had a same-sex partner, although the disparities were slightly more marked among the two groups of women. The median number of lifetime partners was 4 times higher in gay/bisexual women than among strictly heterosexual women (12 versus 3). Among men, this difference was equally consequential but slightly less pronounced (14 versus 5).
19Another indicator also revealed differences between individuals who had engaged in some homosexual practices and those who had only heterosexual practices. In addition to asking respondents about ‘couple relationships or serious intimate relationships’, the EPIC survey also asked them whether in their lifetime they had had ‘other less serious intimate relationships’ and how many. For both men and women, there were substantial differences on this point between the two groups: 48% of women in the ‘gay/bi’ group recounted having had at least three such relationships, versus 30% of women in the ‘exclusively hetero’ group. For men, these differences were slightly less marked (53% versus 40% overall). For both sexes, the differences were particularly large among respondents aged 26–35 years. This observation is associated with a larger number of sexual partners, but it also reflects differences in the relationship between the couple or serious intimate relationships and sexual relationships.
20While the survey does not allow us to characterize the sexual trajectories of the respondents in detail, analysis of the homo-/bisexual trajectories distinguished here, and their comparison with strictly heterosexual trajectories, does bring out some differences. While the homo-/bisexual group is heterogeneous, on average they are more likely to have a relatively large ‘sexual network’, with affective and sexual relations extending into social contexts beyond the couple. They show distinct practices and perceptions around couplehood, sexuality, and how they are connected.
Characteristics of the groups studied
Characteristics of the groups studiedCoverage: Women and men aged 26 to 65 years who have already had sexual relations, data imputed in case of refusal or non-response.
IV – Distinct representations of sexuality and conjugality
21The EPIC survey can be used to study this dimension in more detail, using indicators of representations that offer information on how the respondents understand the connection between sexuality and love. Three questions at the end of the questionnaire were explicitly focused on this issue. They addressed sexual exclusivity, the possibility of disconnecting sexuality and love, and the exclusivity of love:
- Do you agree or disagree with the following statements:
- It is possible to love someone and have affairs on the side.
- It is possible to have sex with someone you do not love.
- It is possible to be in love with more than one person at the same time.
22The answer choices were completely agree; mostly agree; mostly disagree; completely disagree; don’t know.
23The analysis showed that the homo-/bisexual practices studied here were associated with a greater representational dissociation of sexuality and couplehood (Table 6). In addition, men were more likely to agree with all three statements, regardless of whether or not their sexual experiences were exclusively heterosexual, a result that fits with findings from the CSF (Bajos et al., 2008).
24It is important to ensure that these effects remain after controlling for other variables. The analysis performed by the CSF team showed that agreement with these views is associated with age and level of education. For example, the youngest age groups were less likely to accept the possibility of having affairs while staying with a partner. Conversely, the CSF survey found that the acceptance of sexual relations without love was less widespread among members of the oldest age groups (Bajos and Beltzer, 2008).
25Analysis by sex shows that the disparities are especially pronounced between women in the ‘homosexual/bisexual’ group and those in the exclusively heterosexual group, in particular for the indicator concerning the dissociation of love and sexuality (‘It is possible to have sex with someone you don’t love.’). Among men, being in the ‘homosexual/bi’ group is weakly associated with this opinion (OR = 1.4, p < 0.05). This observation is not entirely surprising given sexual norms that encourage men to accept this dissociation, independently of sexual orientation. Among women, the association is very strong (OR = 2.6, p < 0.001), extending the pattern seen in data on numbers of sexual partners. Women who report homo- or bisexual practices are more distant from a model that closely associates sexuality, love, and conjugality than are women who report strictly heterosexual practices. Analyses of the two other variables (‘It is possible to love someone and have affairs on the side’ and ‘It is possible to be in love with more than one person at the same time’) yield similar results. On average, men express less agreement with the association between love and sexuality. Reporting a sexual trajectory involving at least one same-sex partner weakens it, which is particularly marked in women.
Representations of sexuality: completely or mostly agreeing with the following opinions (logistic regression)
Representations of sexuality: completely or mostly agreeing with the following opinions (logistic regression)Interpretation: A statistically significant odds ratio which is greater than 1 indicates that, for this category, the factor increases the chances of belonging to the modelled group with respect to the reference category. The further the odds ratio is from 1, the greater the influence of the factor with which it is associated. For example, reporting at least one same-sex partner over the lifetime (‘homosexual/bisexual’ group) significantly increases the likelihood that a respondent will consider that it is possible ‘to be in love with more than one person at the same time’.
Coverage: Women and men aged 26 to 65 years who have already had sexual relations, data imputed in case of refusal or non-response.
Significance levels: *** p < 0.001, ** p < 0.01, * p < 0.05.
26This article reveals an increase in the number of individuals reporting a same-sex spouse or sexual partner between the mid-2000s and 2014. It is difficult to establish the exact reasons underlying this change, which involve multiple mechanisms. It is likely linked to transformations in social context, which has become more favourable to homosexuality. The increased visibility of homosexuality, its growing albeit incomplete social acceptance, and the legal and social recognition of same-sex couples and LGBT parenting may favour the reporting of practices that have been highly stigmatized and viewed as deviant in the recent past. The very possibility of reporting these conjugal and sexual situations, in surveys on sexuality and in other contexts, reveals that speaking about them has become more acceptable. But this new political and social context could also be behind an increase in these practices, reflected in turn by increases in reported levels. The more favourable context may constitute a form of social authorization, permitting the existence of sexual and conjugal behaviours that would otherwise have been subject to repression or self-censorship. It is likely that both mechanisms (increase in both practices and reporting) are at work here.
27The observations made possible by the questions on sexuality in the EPIC survey fit more broadly into the pattern of recent transformations in sexuality previously highlighted by the CSF team in France (Bajos and Bozon, 2008), particularly with respect to women. As in the United States and the United Kingdom, where surveys on sexuality have provided evidence of this increasing trend between the late 1990s and the early 2010s (Gartrell et al., 2012; Twenge et al., 2016, for the US, and Mercer et al., 2013, for the UK), women in France are now more likely than men to report having had sexual relations with a person of the same sex over their lifetime.
28The sexual trajectories of women, now less systematically linked to conjugality and heterosexuality, are marked by some diversification of practices and a context that is more favourable to their expression. Nonetheless, heterosexuality remains the dominant framework, and virtually all women begin with heterosexual experiences. Homo- and bisexual practices continue to be more common among people with certain social resources (notably educational capital). In this sense, the recognition and increased acceptance of same-sex couples and homosexuality are not yet complete. Among men, the trend toward greater reporting of same-sex partners is less prominent. This could be a differential effect of heteronormativity, which stigmatizes homosexuality as a violation of norms of masculinity and which can also be seen in the greater acceptance of homosexuality among women than among men (Mercer et al., 2013; Rault, 2016b; Tissot, 2018).
Construction of the education variable
Construction of the education variableNote: Level of education is captured here by a variable that aims to take into account change in levels of education over time, across cohorts. Four levels are distinguished.
The PACS, a civil union contract aimed at both same-sex and different-sex couples, has seen continuous growth since its creation. In 2016, four PACS were concluded for every five marriages.
However, it presents a weakness with respect to sexuality: the absence of questions on self-identification and attraction.
According to the CSF survey (2005), which contained both an indicator of self-definition and questions on the sex of the respondent’s sexual partners, the number of individuals included by this definition is larger than that of those who define themselves as homosexual or bisexual (Bajos and Beltzer, 2008, p. 250).
For a detailed analysis of this survey, see Rault (2017).
Question asked in the CSF (by telephone): ‘Overall, in your lifetime, how many men/women have you had sexual relations with?’ (Preamble: ‘We will now talk about the persons you have had sexual relations with over your lifetime. These questions concern the person you live with as well as other regular or occasional partners, including prostitutes.’)
In EPIC (self-administered by computer): ‘Overall, in your lifetime, and including your current situation, how many men/women have you had sexual relations with? Enter a number. If you do not wish to answer this question, hit [a specified] key.’
On the choice of this protocol, see the presentation of the EPIC survey in this issue of Population (Rault and Régnier-Loilier, 2019).
This procedure is based on the strong hypothesis that the behaviours of respondents who abstained from answering the question on the number of sexual partners are similar to those of persons with the same sociodemographic characteristics who did respond.
These consisted of the following questions: 1) Sex of the first sexual partner, 2) Sex of the person with whom the respondent had their most recent sexual encounter, and 3) ‘Before the age of 18, did you have any sexual experiences with someone of the female/male sex?’ Taking the responses to these questions into account, 4% of women and 4.1% of men aged 18 to 69 reported having had sexual relations with a person of the same sex over their lifetime.
See Appendix for the principles of the construction of this variable into four levels designed by the CSF team to take into account increasing levels of formal education across cohorts.