We investigate whether the subjective well-being of individuals in same-sex unions improved following the legalization of same-sex marriage in England and Wales in March 2014. We employ repeated cross-sectional data from the 2011–2016 Annual Population Surveys on 476,411 persons, including 4,112 individuals in coresidential same-sex relationships. The analysis reveals increases in subjective well-being for individuals in same-sex relationships following legalization. Additional analysis documents higher subjective well-being for individuals in married same-sex couples compared with individuals who are in a civil partnership or an informal cohabiting same-sex union. However, the subjective well-being of individuals from same-sex couples increased after legalization among all subgroups considered, including those who cohabited informally. This result hints at a general reduction in structural stigma as an important mechanism behind the improved well-being of individuals in same-sex unions.
The legalization of same-sex marriage is seen as an important step toward reducing the gap in well-being between sexual minorities and heterosexual individuals (Gonzales 2014; King et al. 2008; Thomeer et al. 2018; Wight et al. 2013). Now that same-sex marriage has been legalized across an increasing number of countries (Trandafir 2015), it is possible to explore the extent to which the subjective well-being of sexual minorities has indeed risen. Access to marriage might have increased sexual minorities’ subjective well-being for at least two reasons.
First, marriage is positively related to health, income, wealth, and family satisfaction (Carr and Springer 2010; Pirani and Vignoli 2016; Vespa and Painter 2011). Some of the underlying mechanisms producing these benefits apply to relationships in general, including emotional support, the pooling of economic resources, and the monitoring of health habits by partners (Donnelly et al. 2018, Herek 2006; Umberson et al. 2018). Other mechanisms are specific to formalized unions, such as less uncertainty regarding the future of the relationship, better social support, and access to institutional benefits (Carr and Springer 2010; Herek 2006). Marriage appears to be more strongly related to such benefits than civil partnerships (LeBlanc et al. 2018). Studies relating to the United States have indicated that individuals in same-sex marriages have less psychological distress than single LGBT individuals (Goldsen et al. 2017) and individuals in both registered and nonregistered same-sex partnerships (LeBlanc et al. 2018; Wight et al. 2013).
Second, the psychological well-being and health of sexual minorities is affected by institutional discrimination (Hatzenbuehler et al. 2009). Sexual minorities experience barriers to achieving relationship goals (Frost and LeBlanc 2014) and less social recognition of their relationships (Frost et al. 2017). Sexual minorities have reported that unequal institutional recognition is a source of stress (Frost et al. 2017; Thomeer et al. 2018). Therefore, access to marriage might not only increase the well-being of those who marry but also reduce any stigma that sexual minorities experience more generally by abolishing a source of institutional discrimination (LeBlanc et al. 2015). In the United States, sexual minorities have reported worse self-rated health, more limited health care coverage, and more psychiatric disorders in states where legislation is less favorable to them (Gonzales and Blewett 2014; Hatzenbuehler et al. 2009; Kail et al. 2015).
Two studies, both based on the United States, have documented changes in the well-being of sexual minorities following changes in legislation. First, Hatzenbuehler et al. (2010) used representative data to show how psychiatric disorders increased among sexual minorities living in states that banned same-sex marriage in 2004 (see also Rostosky et al. 2009). Second, Hatzenbuehler and coauthors (Hatzenbuehler et al. 2012) established that medical care and mental health care visits decreased among a clinical sample of gay and bisexual men from Massachusetts following the state’s legalization of same-sex marriage in 2003.
The current study contributes by using nationally representative data to document changes in subjective well-being following the nationwide legalization of same-sex marriage in England and Wales. In March 2012, the British government announced its plans to legalize marriage for same-sex couples in England and Wales (Eekelaar 2014). The law came into force on March 13, 2014 in both countries, and the first same-sex marriages were registered on March 29, 2014. Britain had a relatively low societal approval of same-sex marriage compared with other Western European countries (Gerhards 2010).1
Data and Method
We employ data from the Annual Population Survey (APS) covering the period April 2011 to September 2016. Every three months, the APS randomly selects private households from a common sampling frame for England, Wales, and parts of Scotland.2 Households are generally interviewed in five subsequent quarters before rotating out of the survey. Information is collected from all household members present through individual computer-assisted face-to-face and telephone interviews, and information on those not present is collected from proxy respondents. Response rates range between 45 % and 50 % for the waves covered.
Following political interest in noneconomic measures of well-being, the U.K. Office for National Statistics incorporated subjective well-being measures in various surveys, including the APS (Hicks et al. 2013). Respondents aged 16 and older who gave a personal interview completed these questions.3 We restrict this sample further to adults in a coresidential union (dropping 21 % of the sample) who lived in England or Wales (excluding 4.7 % of respondents).
Respondents were asked to use a scale ranging from 0 to 10 to rate the following: (1) how anxious they were yesterday; (2) how happy they felt yesterday; (3) how worthwhile they find things that they do in their life; and (4) how satisfied they are with their life. These measures of subjective well-being, even though empirically related to objective measures of well-being (Dolan et al. 2008), capture distinct components of well-being, including overall evaluations of life, the experience of positive and negative emotions, and the perception of meaning in life (Hicks et al. 2013). These, and similar measures, have been used to document the impact of individual events and policy changes on subjective well-being (Deeming 2013; Helliwell and Huang 2014; Sage 2015). Our main analysis focuses on an index averaging answers to each question into an overall scale (Cronbach’s alpha = .73; anxiety values are reversed for this procedure). This scale ranges from a minimum value of 0 to a maximum value of 10 (with an average of 7.6 and a standard deviation of 1.5; see online appendix A for the complete distribution). In additional analysis, we present results for each measure separately (see online appendix B).
The independent variable of interest is whether the respondent is in a coresidential relationship (regardless of marital status) with someone of the same sex or with someone of a different sex. This information is derived from a household roster indicating the sex of individuals present and the relationships connecting them. Because of the lack of information on sexual orientation in the data, we use a measure based on coresidence with a partner of the same or different sex. This is a limitation common to many studies on sexual minorities (Gates and Brown 2015; Rosenfeld 2010). Checks of data from the U. S. Census Bureau revealed that if individuals in same-sex couples are identified based on coresidence, relatively uncommon coding mistakes (e.g., on the sex or relationship variable) can lead to a nonnegligible share of individuals in different-sex couples being marked as living in a same-sex couple (Gates and Brown 2015). It is unclear to what extent this issue applies to our data, but we ran robustness checks using face-to-face interviews only, which are arguably less prone to recording mistakes; the results did not change (available upon request). One advantage of our data over commonly used sources, such as the U.S. Census, is that the option of “cohabiting partner” directly follows the option of “spouse” when respondents are asked about their relationship to other household members. This was signaled as a factor easing the answering of such questions and one of the improvements recently introduced by the U.S. Census Bureau to better identify same-sex couples (Kreider et al. 2017).
We include a range of confounders: ethnicity (black, Bangladeshi, Chinese, Indian, mixed, other Asian, Pakistani, white, other); region (11 government office regions); sex; age; age squared; education (no qualifications, GCSE+, or university); year; interview mode (face-to-face or telephone); net weekly pay (five categories: no pay and four pay quartiles); the presence of children (none, at least one child under age 5, or only children between ages 5 and 18); being foreign-born; and the presence of a disability.
The final sample includes 446,141 individuals, of whom 4,112 are in a same-sex relationship.4,5 We employ ordinary least squares (OLS) regression models. Even though respondents provide information on well-being in their first and last interview (fifth quarter), the publicly available APS data sets do not allow for specific individuals to be followed over time. We therefore look at cross-sectional changes among groups as a whole. In addition, even though several individuals per household can be interviewed, the absence of household identification numbers in the publicly available data prevents us from clustering observations according to household.6 We therefore estimate robust standard errors through all our analysis.
Table 1 compares descriptive statistics for individuals in same-sex and different-sex relationships. Across the period studied, average well-being was lower among individuals in same-sex couples. Socioeconomic differences include higher education, higher employment, and fewer children among individuals in same-sex relationships. In the sample, 83 % of individuals in different-sex relationships were married, compared with 5 % of individuals in same-sex relationships. Nearly one-half (45 %) of individuals in same-sex relationships were in a civil partnership, a legal option available only to same-sex couples in the United Kingdom (Ross et al. 2011).
Figure 1 shows descriptively how average overall well-being has steadily increased for individuals in different-sex relationships (thick dashed line). Individuals in same-sex relationships, in contrast, experienced decreasing levels of well-being in the period 2011–2013 (thick solid line). Since 2014, the year same-sex marriage was legalized, their well-being has increased.
Table 2 evaluates to what extent changes in the gap in well-being between groups have been statistically significant. The analysis divides the observation period into 12-month intervals based on the time between the interview and the month of legalization (March 2014). Model 1 confirms that individuals in same-sex relationships had lower subjective well-being across the observation period. Model 2 shows how the gap between individuals in same-sex and different-sex relationships decreased significantly in the years following legalization compared with the 12 months before legalization. The reductions in this gap of 0.28 and 0.24 points correspond to roughly one-fifth and one-sixth of a standard deviation, respectively.7 Interestingly, the gap had been increasing in a significant way in the years leading up to legalization—an issue we discuss later.
The changing gap in well-being over time holds once changes in background characteristics over time are controlled for (Model 3). Additional analysis on subcomponents of well-being (online appendix B) showed that this pattern was reproduced most clearly for the experience of negative and positive emotions (happiness and anxiety). Patterns were different for life satisfaction, which was higher for individuals in same-sex relationships two years before legalization, but converged with the satisfaction reported by individuals in different-sex relationships in subsequent years.
Table 3 restricts the sample to individuals in same-sex couples to scrutinize two possible and complementary explanations as to why the well-being of individuals in same-sex couples increased over time. First, well-being might have increased because of the direct benefits related to marriage. Model 1 shows that during the period after legalization, married individuals in same-sex relationships had higher subjective well-being than individuals who informally cohabited or were in a civil partnership.8 Model 1 hence provides initial support for the argument that access to the direct benefits of marriage increased average well-being among same-sex couples. However, it could also be that this cross-sectional result is produced by selection into marriage based on unobserved characteristics.
A second possible explanation for the increased well-being of individuals in same-sex couples is the lifting of institutional discrimination against same-sex couples. Reductions in the experience of structural stigma might have increased the well-being of individuals in same-sex couples regardless of whether they got married. Models 2 and 3 split the analysis according to union type, and Fig. 2 reproduces the results graphically. For individuals in same-sex civil partnerships or marriages, initial declines in well-being are observed until the year of legalization, after which sustained increases in well-being took place. Subjective well-being among informally cohabiting individuals in same-sex relationships was relatively stable until legalization and subsequently increased significantly, providing support for a general positive effect of legalizing same-sex marriage on the well-being of individuals in all types of same-sex unions. These increases in well-being among the informally cohabiting are less pronounced two years after legalization. This result might be due to sampling error (the differences between the last two years are not statistically significant); alternatively, it might indicate that other changes in society reduced well-being to some extent again in later years among same-sex cohabitors or that part of the benefits of legalization for cohabitors are temporary.
However, overall, the benefits of legalizing same-sex marriage do not seem to be limited to individuals in formal relationships. This claim is further strengthened by the observation that only a small minority of individuals in same-sex relationships actually married after legalization. In 2016, 10 % of individuals in a same-sex relationship were in a marital relationship.
For this study, we used data on subjective well-being from England and Wales to document that the size of the gap in subjective well-being between individuals in different-sex and same-sex relationships decreased after the introduction of same-sex marriage in March 2014. We examined two alternative but complementary explanations for these patterns: the direct benefits of marriage and general reductions in stigma. After legalization, individuals in same-sex marriages had higher subjective well-being compared with individuals in same-sex civil partnerships or cohabiting unions. However, subjective well-being increased over time among all groups of individuals in same-sex unions, including those in an informal cohabitation.
This result was to be expected if the legalization of same-sex marriage affects well-being by reducing any structural stigma experienced by individuals in all types of same-sex couples. Previous research has shown that structural stigma, including institutional discrimination, affects the well-being of entire groups and not only the well-being of those directly involved (Hatzenbuehler et al. 2009). The abolishment of an institutional source of discrimination is therefore likely to have increased the well-being of individuals in same-sex couples regardless of union status. Given that our analysis identified same-sex couples based on coresidence and lacked information on sexual orientation, future research can investigate whether the well-being of sexual minorities who are not in a relationship has also been affected by changes in legislation.
We noted that subjective well-being decreased in the years before legalization among individuals in same-sex civil partnerships but not among those cohabiting informally. One possible interpretation of this pattern is that increased exposure to negative messages and interactions during the public debate preceding legislation reduced well-being among individuals in same-sex relationships. This had been observed when same-sex marriage was banned in some parts of the United States (Rostosky et al. 2009). Given these legislative and public opinion tussles, individuals in registered civil partnerships might have perceived that the legitimacy of their union was being questioned by parts of society. Previous research suggested that individuals in same-sex registered partnerships appear to display lower psychological well-being compared with individuals in same-sex marriages (Wight et al. 2013). On some occasions, informally cohabiting couples showed better mental health than same-sex couples who were in a domestic registered partnership or civil union (LeBlanc et al. 2018). The results of our study suggest that comparisons of the benefits of marriage with those of registered partnerships might be sensitive to the period studied. Similarly, the direct benefits of marriage and its impact on average well-being among same-sex couples might unfold over a longer period as more couples get married. Future research could investigate how well-being develops among the different subgroups studied here across longer periods.
One limitation of this study is that we were not able to follow individuals’ well-being over time to test whether our results are driven by changes in composition or by changes in well-being among a given set of individuals. This opens the door to several alternative explanations for the patterns observed. First, the legalization of same-sex marriage might have increased the number of individuals in a same-sex union, and high levels of well-being among recently formed couples might have driven average levels of well-being upward. Second, certain individuals in same-sex unions were perhaps more prepared to disclose their relationship status to interviewers after legalization. Both of these possibilities would mean a jump in the number of individuals in same-sex couples recorded in the APS following the legalization of same-sex marriage. The share of individuals in same-sex couples reported in the data, however, has risen quite steadily over time (see online appendix E). In addition, the characteristics of individuals in same-sex couples did not change much across time in the sample used (see online appendix F). Finally, there might be other macro-level changes that affected the well-being of individuals in same-sex relationships but not the well-being of those in different-sex relationships. It is certain, though, that the subjective well-being of individuals in same-sex relationships increased following the legalization of same-sex marriage even if it had been decreasing to that point (but increasing among different-sex couples). This sudden break in the time trend suggests that an event that took place in 2014 is responsible for the results observed rather than a process that developed over an extended period.
Future studies using longitudinal data might shed a better light on which mechanisms produced the changes in well-being related to specific union-types. Dyadic data might help disentangle to what extent legalizing same-sex marriage has reduced minority stress on the couple-level (e.g., processes experienced as a couple) or whether increases in well-being are the result of processes that occur at the individual level, such as a reduction in discrimination experienced by all individuals belonging to sexual minority groups (Kroeger and Powers 2019; LeBlanc et al. 2015).
To conclude, we began the article by restating the expectation of scholars that legalizing same-sex marriage would reduce the subjective well-being gap between sexual minorities and heterosexual individuals (Gonzales 2014; King et al. 2008; Thomeer et al. 2018; Wight et al. 2013). Our results show that the well-being of individuals in same-sex couples indeed increased following the legalization of same-sex marriage in England and Wales. That increases in well-being were observed regardless of union status provides empirical support for the argument that abolishing institutional discrimination affects the well-being of individuals involved, both directly and indirectly, by reducing structural stigma.
We would like to thank attendants at the 2018 Population Association of America annual meeting (Denver) and the 2018 European Population Conference (Brussels) for their insights and comments. Diederik Boertien acknowledges research funding from the Beatriu de Pinos program of the Generalitat de Catalunya (2016-BP-00121), the EQUALIZE project (ERC-2014-STG-grant agreement No 637768), and the GLOBFAM project (RTI2018-096730-B-I00).
Given the presence of universal public health care in Britain, access to health care through a married partner’s employment is likely to be a less relevant mechanism to explain well-being disparities compared with other countries, such as the United States (Gonzales and Blewett 2014).
These data are complemented by annual boost samples. For more information, see https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/methodologies/annualpopulationsurveyapsqmi.
Dropping cases without information on subjective well-being leads to the exclusion of 48 % of the sample, mainly children and respondents not present at the time of interview for whom proxy respondents provided information. Well-being questions were not put to proxy respondents. Sample weights designed for nonresponse on well-being measures are included. We dropped 0.3 % of cases because of missing household-level information. Less than 0.1 % of individuals were excluded because merging household to individual data based on observable characteristics (given the absence of personal identification numbers in the publicly available data) led to unequivocal matches in 99.9 % of cases only (see also footnote 6).
The process leading up to legalization was very similar for England and Wales (Eekelaar 2014), and the well-being of sexual minorities is often studied for both countries jointly (e.g., King et al. 2003). There are therefore no a priori expectations regarding differences across countries in results. Nonetheless, we excluded the 10 % of respondents living in Wales from the analysis in robustness checks (online appendix C); the results remained unchanged.
Listwise case deletion excludes 1.7 % of the original sample. Using multiple imputation renders practically identical results (online appendix D).
Household and personal identification numbers in the APS are available only under a special license agreement to U.K.-based researchers.
Robustness checks based on face-to-face interviews alone displayed very similar substantial results, albeit with lower levels of precision.
Additional tests show that the difference between individuals in a marriage and those in a civil partnership is statistically significant (p = .03).
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.