*2.2.2 Legal marriage and partnership*

CHIS asked all participants the standard marital question, "Are you now married, living with a partner in a marriage-like relationship, widowed, divorced, separated, or never married?" The response options do not include same-sex marriage or legal partnership. Previous research has shown that lesbian/gay women and gay men under-report marriage and legal partnership when responding to standard marital status questions [19]. To address under-report of marriage and to reflect the 2008 CSCD, CHIS asked all participants who reported having sex with someone of the same sex within the preceding 12 months whether that sexually-active respondent had a legal same-sex spouse or domestic partner. In addition to legallypartnered and married, we consolidated remaining status into *other*, a category including unmarried people who may be divorced, widowed, never-married, or living with a partner without legal recognition. We also constructed a binary variable for married/legally-partnered to increase the power of the data to find statistically significant results. There is no way to separate married and legally-partnered for data prior to 2009 because of questionnaire wording. In 2009, revised question wording distinguished married from legally-partnered. We compared percent of moderate psychological distress between married and legally-partnered lesbian women and gay men and found no statistical difference. Therefore, we collapsed married/legally-partnered as one group for our analyses in order to compare pre to post CSCD.

#### *2.2.3 Statistical analyses*

We performed all statistical analyses using CHIS data pooled from survey years 2005–2015 and weighted to the California population. Lesbian/gay women and gay men were compared to their heterosexual counterparts on sociodemographic variables. Data collected before 2008 were considered prior to the CSCD and data collected in 2008 and later were considered after the legal decision. The proportion of the sample experiencing moderate mental distress was plotted over time by gender, sexual identity, and couple status. Joinpoint analysis tested if trends in moderate psychological distress changed at specific years. Joinpoint uses weighted least squares to fit the trend model, using the inverse of the standard error as the weight variable.

Bi-variate and logistic regression analyses compared psychological distress using the K6 scale between the periods before and after the CSCD. Bi-variate analyses were replicated only for lesbian/gay women and gay men and compared stress levels between those legally married or partnered as compared with those not. Independent variables of the logistic regression included the main effects of sexual identity, marriage/legal partnership and the timing of CSCD (before or after) and all the two-way interaction effects (i.e., sexual identity x marriage/legal partnership, sexual identity x timing of CSCD, marriage/legal partnership x timing of CSCD) and the three-way interaction effect of the three variables (i.e., sexual identity x marriage/legal partnership x timing of CSCD) while adjusting for the following sociodemographics: Race, marital/partnered status, children in home, education, work status, income, geography, age. The conditional adjusted odds ratios compare the odds of reporting moderate mental health distress for that specific group pre- versus after-CSCD while holding all other variables constant.
