Despite this progress of empirical investigations, two methodological issues remain. First, there is limited analysis of data collected originally for assessing changes associated with SSM legalization [10]. Despite the ability to identify changes in outcomes in parallel with policy changes [16], a caveat in these studies involves the historical threat as researchers were not able to surely determine whether the observed outcomes were directly due to the SSM legislation or other confounding factors or parallel policy changes [16, 22]. A viable response to this methodological problem is to measure the outcomes specific to sexual minority individuals, such as exposure to distal sexual minority stress or disclosure status since they are intrinsic to the impact of legalization of SSM. A prospective design to assess LGB-specific outcomes in the same individuals is another solution by treating subjects as their own controls so that within-subject changes could be largely attributed to the policy changes [18].

Another gap pertains to the exploration of the effects of SSM for un-partnered sexual minority individuals [23] given that many studies [19, 24–26] only sampled partnered individuals. While certain benefits afforded by legalization of SSM are particularly relevant for LGB individuals in relationship, such as allowing them to form a family, increasing the sense of love, closeness, and commitment, and improving their relationships with family of origin [27], the reduction in discrimination and bias following the legalization [6] is supposed to benefit the entire LGB communities. However, studies have yielded inconsistent findings. A study with sexual minority women found that single participants perceived lower impact of SSM legalization on their personal lives and social climates than their partnered counterparts did [14]. In contrast, Hatzenbuehler, O'Cleirigh [18] found that changes in healthcare utilization following the legalization of SSM were not significantly different between partnered and un-partnered sexual minority men. Similarly, Everett et al. [16] found comparable health effects of legalizing same-sex civil unions for both partnered and un-partnered sexual minority women. Given the inconsistency of the study findings, we also accounted for the relationship status to determine whether SSM legalizations produced effects among wider sexual minority populations.

To address these research gaps, this study employed a prospective, panel design to evaluate the short-term outcomes of SSM legalization in Taiwan. The study makes two empirical contributions. First, instead of capitalizing on pre-existing data, primary data were collected to delineate changes in mental health status and LGB-specific variables as functions of the legalization of SSM. The data gathered by this study are closely relevant to sexual minority individuals’ personal experiences and perceptions and therefore are more able to capture the direct implications of the marriage equality policy. Second, the sample comprised both partnered and un-partnered individuals, thus allowing us to ascertain whether the effects of SSM legalization varied by one’s relationship status. Notably, acknowledging the heterogeneity within sexual minority communities and the gender-based nature of experiences, we collected data from cisgender gay and bisexual men only.

The study was prospective in nature, with baseline data collected 1 months prior to the enactment of SSM in Taiwan on May 24, 2019 and follow-up data gathered 1.5 years following the legislation. Similar with several previous studies [16, 18], we adopted this timeframe to assess the short-term changes associated with SSM legalization.

Data were derived from a community panel of self-identified gay and bisexual men living in Taiwan. We utilized Facebook advertising to recruit a baseline sample between May 11 and 23, 2019. A set of parameters were specified in the metrics to direct advertisements about the study to Facebook users who indicated in their profiles that they were: 1) Men, 2) aged ≥18 years, 3) residents of Taiwan, and 4) interested in lesbian, gay, bisexual, and transgender community–related issues and information. Interested users clicked the link embedded in the advertisement and were transferred to the online survey platform, SurveyMonkey, to read about the purposes of the study and eligibility criteria, which were: 1) Being assigned male at birth, 2) being aged 18 years or older, 3) self-identifying as gay or bisexual, and 4) residing in Taiwan. They pressed the button “I agree to participate” to give consent and spent approximately 12 min completing the survey. At the end of the survey, they were given the option of providing their contact details (e.g., email, mobile phone number, or communication mobile applications) to receive an e-coupon valued NT$100 as an honorarium.

Between September 14 and October 16, 2020, we sent invitations and information about the follow-up survey to respondents who had provided contact details. Respondents were informed of the purpose of the follow-up survey and ethical issues involved when considering whether to participate. After providing consent, respondents spent approximately 13 min completing the survey and were prompted to leave their contact information again to receive an honorarium. The data from the two waves were linked based on the respondents’ contact information.

In the baseline survey, 2,525 people visited the survey website, and 1,422 completed the questionnaire. Data cleaning was undertaken by inspecting respondents’ sociodemographic data, IP addresses, and contact information. Forty-one respondents were removed for: 1) being younger than 18 years (n = 3), 2) not identifying as gay or bisexual (n = 24), or 3) being repeat respondents (n = 14), thereby reducing the final sample of the baseline survey to 1,381. In the follow-up survey, only 952 responses were returned (attrition rate = 31.1%) of which 89 were discarded because: they were from an identical IP address as another respondent (n = 9); or had missing values in the full survey (n = 25), questions on the variables of interest (n = 30), or the age question (n = 5). Responses were also discarded if they had error messages (n = 1), were unmatched with the corresponding data in the baseline survey (n = 1), or not self-identified as bisexual or gay (n = 18). The final sample size for analysis was 863. The sociodemographic data of respondents from the follow-up survey are presented in Table 1.

We first described respondents’ sociodemographic characteristics stratified by their relationship status. Group differences were examined by chi-square tests and independent sample t-tests. Means and standard deviations of the continuous variables in the baseline and follow-up surveys were calculated separately and changes over time were examined by paired sample t-tests. The effect sizes of the t-tests were also computed. Designed to test paired nominal data, McNemar’s chi-square was used to assess the significance of the change in disclosure to mothers and fathers.

To examine the within-individual changes of the variables before and after the SSM legalization, we conducted repeated measures analysis of variance (ANOVA) to compare the mean differences across two time points. Because this analysis required individuals to have data at both time points, list-wise deletion was used to handle missing values. In the analysis, the independent variables were time as a 2-level within-subject (baseline and follow-up) measure and the relationship status in the follow-up survey as a 2-level between-subject factor (un-partnered and partnered). We tested the interaction effect of time and relationship status to examine whether changes over time were comparable between un-partnered and partnered respondents. These analyses were conducted using SPSS 24.0.

Provided with the panel data, we also used fixed-effect regression analysis to determine whether and how the changes in minority-specific factors, including distal sexual minority stress, internalized homophobia, and outness, contributed to the change in participants’ depressive symptoms. Fixed-effect analysis relies on within-person variation to estimate coefficients while automatically controlling for unobserved time-invariant factors when each individual serves as their own control [34]. Therefore, we only controlled for sociodemographic confounders that may change over time, including sexual orientation, relationship status, religious status, monthly income, and education attainment. To adjust for the time and contextual factor that may simultaneously affect people in Taiwan (e.g., COVID-19), we also created a dummy variable of time and incorporated it into each model. Each minority-specific factor was tested in separate models due to their inter-correlations. The Fixed-effect regressions were conducted using STATA 14.0.

Several limitations are noted. First, despite the longitudinal design, this study is limited in establishing causality of the effects of SSM legislation, which requires comparative samples of heterosexual individuals or LGB people not exposed to SSM legislation. Only through these comparisons are we able to ascertain whether the changes observed among LGB individuals are also attributable to other societal changes. Second, the representativeness of survey respondents is unknown because they were recruited through non-probability sampling and we could not determine how Facebook advertisement distributed the recruitment posts. The non-probability sampling procedure thus requires extreme caution in interpreting the result of inferential statistics. Third, as a common issue for Internet-based surveys [37], attrition bias could not be fully eliminated although no systematic patterns in drop out were found for the key variable (Appendix Table A1). Fourth, relative to objective measures, the self-reported measures on which the study relied could introduce self-report biases and measurement errors.

It should also be noted that the effects of legislation could be time-sensitive in that passage and implementation of the bill could produce differential impacts [16]. This consideration is also important for this study because the Supreme Court’s ruling on May 24, 2017 mandated the Legislative Yuan to either pass a new bill or amend the existing one to enact marriage equality within 2 years. Because the pre-legalization data were collected during this transitional period, participants might have been exposed to the anticipation effect of the policy shift. Meanwhile, the period preceding the enactment of SSM legislation was also marked by heated debate and a proliferation of opposing opinions and anti-gay comments. These social dynamics could have provoked emotions among members of LGB communities [38, 39]. The post-legalization data were collected during the COVID-19 pandemic that has imposed a collective influence on citizens’ mental health. All these events need to be considered in the interpretation of the findings while additional follow-up investigations will be necessary to examine the longer-term effects of SSM legalization.

Advocacy of marriage equality is driven by values and evidence. While the ethos of equal rights aptly justifies the legalization of SSM, this study has generated explicit evidence for a wide range of psychosocial benefits afforded by SSM legalization to sexual minority men in Taiwan. The finding indicates that after the legalization of SSM, sexual minority men in Taiwan reported fewer depressive symptoms, perceived the environment as less hostile and exclusionary, and felt more willing to disclose their identities. These findings not only enrich knowledge about the effects of SSM policy but lend strong support for progression towards marriage equality.