One of the goals of Healthy People 2020 [1], a science-based report issued by the US Office of Disease Prevention and Health Promotion that identifies national objectives for improving the health of Americans, is to specifically improve the health and well-being of lesbian, gay, bisexual, and transgender (LGBT) individuals. Compared with heterosexual women, sexual minority women (SMW; eg, lesbian, bisexual, queer, and pansexual) engage in more heavy drinking and have lower rates of abstention from drinking [2-4]. SMW also experience more negative alcohol-related consequences than heterosexual women, such as driving under the influence of alcohol, having unplanned sex, having suicidal thoughts during or after drinking, fighting or arguing with someone, and engaging in sexual harassment during/after drinking [5,6]. Despite evidence for health disparities between heterosexual women and SMW in alcohol use, much remains to be learned regarding why some SMW in particular are at risk. The goal of this study, Project Relate, was to increase the understanding of hazardous drinking among SMW who are in same-sex relationships using an entirely remote Web-based data collection procedure, including a daily diary component. This approach permits examination of research questions at multiple levels, allowing us to consider between-couple characteristics (ie, couple-level; answering the question: which couples drink more), between-person characteristics (ie, person-level; answering the question: who drinks more), and within-person processes (ie, daily level; answering the question: when does one drink more). The primary objective of this study is to examine the daily associations between hazardous alcohol use and sexual minority stress, affect, and relationship factors among young same-sex female couples. The theoretical framework and existing empirical evidence that contributed to the development of this study are described below.

Members of minority groups experience unique stressors related to their stigmatized and marginalized social status in society. Stigmatization and marginalization based on sexual identity or orientation is known as sexual minority stress [7-9]. Specifically, Meyer’s [9] model of sexual minority stress involves experiences of harassment or discrimination related to sexual identity/sexual orientation (ie, heterosexism), the expectation of and vigilance toward these events, concealment of sexual identity, and the internalization of negative societal attitudes toward one’s sexual identity (eg, internalized heterosexism). Together, these experiences create additional stressors beyond those experienced by majority of individuals. There is ample evidence that sexual minority stress is associated with negative health behaviors, such as alcohol use, through emotion regulation and coping strategies [8,10-14] as well as through less perceived social support [12] and more social constraints and social isolation [13]. Although these studies provide important information regarding how minority stress is associated with negative affect and alcohol use at the between-person level, it is important to note that these studies used primarily cross sectional (ie, single assessment points) and traditional longitudinal designs (ie, using repeated assessments over weeks, months, or years). These types of designs limit our understanding of within-person processes by either studying processes as a between-person variable (ie, using cross-sectional designs) or by studying long-term within-person differences (ie, using traditional longitudinal designs). A handful of studies have used daily or momentary assessment of stress and alcohol use among sexual minorities. For example, in research of young men who have sex with men, alcohol use was greater on days when men indicated they were drinking to cope with stress [15] and on days when general stress was reported [16]. Furthermore, in a 9-day daily diary study, the interaction of structural stigma (eg, state policies that promote discrimination) and rejection sensitivity was associated with alcohol use among young men who have sex with men. Most relevant to this study was the finding that discrimination was associated with contemporaneous and prospective substance use in a group of sexual and gender minority individuals [17]. However, the question of how daily experiences of sexual minority stress and negative affect contribute to SMW’s hazardous drinking remains. In this study, we will test whether greater daily sexual minority stress is directly associated with more alcohol use and associated problems as well as examine indirect associations through negative affect.

In addition to the link between sexual minority stressors, negative affect, and alcohol use, it is important to consider the associations between relationship factors and alcohol use. A review of studies on heterosexual couples concluded that unpartnered individuals drank more compared with those in serious romantic relationships and that drinking is associated with poorer relationship quality as well as conflict and violence [18]. Among SMW, several studies have demonstrated associations between partner violence and alcohol use and related problems in samples of SMW. For example, in a study of lesbian couples who experienced intimate partner violence, approximately 64% of both batterers and victims report using alcohol or drugs before the violent incidents [19]. Alcohol use was also associated with SMW’s nonphysical domestic violence (eg, verbal threats and damage to property) [20]. Research also suggests that intimate partner violence among SMW is associated with hazardous alcohol use [21], which may stem from emotional distress [22] or sexual minority stressors [23].

Another important characteristic when studying relationships and hazardous alcohol use is the degree to which partners drink together versus separately. Among heterosexuals, drinking together has been associated with more intimacy and fewer negative partner behaviors than drinking apart [24]. Conversely, drinking apart is associated with more negative relationship events [24] as well as less satisfaction, more conflict, and marital dissolution [18,25]. Among lesbian women, when controlling for physical and psychological aggression, discrepant drinking (ie, differences in partners’ drinking quantity) is associated with poorer relationship adjustment [26]. Furthermore, in a longitudinal examination of discrepant drinking and intimate partner violence among lesbian partners, a cyclical pattern of alcohol use and violence occurred such that discrepant drinking predicted psychological aggression 6 months later, which in turn predicted later discrepant drinking. Physical violence also predicted subsequent discrepant drinking 6 months later [27]. Considering the small body of work connecting alcohol use to relationship variables in SMW, in this study, we will examine how instances of partners drinking together versus apart and discrepant drinking are associated with relationship functioning on the same day and subsequent days and in turn, how relationship functioning may then be associated with drinking on subsequent days.

The impact of minority stress and negative relationship factors may be modifiable at the couple- or individual-level [9,28], but more information is needed to understand what factors may protect SMW from problematic alcohol use [29]. Therefore, the final aim of this research is to investigate societal-, couple-, and person-level characteristics that could be associated with daily experiences of sexual minority stress, relationship quality, and drinking or that may strengthen or weaken the associations among them. For example, connection to the sexual minority community, a stronger sexual minority identity, and social support may all protect against the deleterious effects of sexual minority stress and related negative mental and behavioral outcomes [30-32]. Similarly, being in a legally recognized relationship may also be a protective factor against sexual minority stress and negative health outcomes [33-35]. Although marriage between same-sex individuals is now legal in all states in the United States, other protections (eg, housing, employment, and health care benefits) are not federally mandated and, thus, vary by state. The lack of these state protections (ie, structural stigma) has been associated with health outcomes. For example, in a study published in 2014, sexual minority individuals who lived in states without protections for sexual minorities (eg, in housing and employment benefits) and that banned same sex marriage had a higher prevalence of psychiatric disorders than those who lived in states with such protections [36]. In this study, we will assess a variety of potential factors that may protect against hazardous drinking, including both factors specific to the person or couple (eg, connection to sexual minority community and being in a committed relationship), as well as structural factors (eg, living in a state with protections for people in same-sex relationships). Identification of potentially modifiable individual risk and protective factors can facilitate development of culturally tailored inventions to reduce hazardous drinking. Furthermore, identification of structural (societal) risk and protective factors may suggest important directions for advocacy efforts.

The primary objective of Project Relate is to increase our understanding of hazardous drinking among SMW. Using a daily diary methodology, this project investigates the associations among sexual minority stress, affect, relationship factors, and alcohol use. The development of this study was guided by minority stress theories [7-9] that offer a way to understand the unique stressors that SMW experience related to their marginalized societal status. In addition, because relationship factors are important correlates of alcohol use among heterosexual couples [18], we consider how relationship characteristics are associated with alcohol use among young adult same-sex female couples. In addition, we also look at potential protective factors. Collecting information on sexual minority stressors, emotional functioning, relationship experiences, and alcohol use from both partners will allow us to assess the interactions among these variables as they enhance or reduce the quality of life of SMW. Extensive background information, including personality characteristics, social support, history of discrimination, and community context, will provide information on personal and contextual factors that moderate these associations. A secondary objective of Project Relate is to examine other physical and mental health concerns, including other substance use, eating, physical activity, and general mental health and well-being among SMW. The opportunity to examine daily-, individual-, and couple-level effects is a particular strength of the study methodology.

Project Relate is an ongoing daily diary study of same-sex female couples. Young adult SMW (aged 18-35 years) are the target sample for this study because in community samples they frequently endorse indicators of hazardous drinking [3]. Both partners of the romantic relationship are enrolled to capture information about the relationship from both members’ perspectives. Participants are recruited from across the United States, and data collection is occurring remotely via Web-based surveys. Both women in each couple complete a comprehensive baseline survey and then a brief daily survey each morning for the following 14 days. Each member of the couple begins the daily survey on the same day to ensure we have corresponding daily data for each couple. The Old Dominion University Institutional Review Board approved all study procedures (Project #839097).

A power analysis to determine the sample size needed for planned analyses was conducted in 2 steps. First, a power analysis was conducted for traditional structural equation modeling using Monte Carlo simulation methods [37], focusing on powering the hardest effects to detect (the small indirect effects, beta=.10). Similar models were conducted incorporating a small-to-medium effect–moderating variable (beta=.20). Monte Carlo simulation methods indicated that for the effect size expected, a sample size of 200 cases yields sufficient power (0.808) to detect the relevant effects. This, however, assumes no correlation among residuals, which we know is not true in nested designs. To determine how to appropriately account for the correlated observations and identify the minimum number of participants needed for power of .8 in the multilevel design, a formula taking into account the expected intraclass correlation coefficients (ICCs), or degree of relatedness within couples, for key variables was used [38]. Assuming an average of 11 days of assessments (80% response rate to account for attrition) and 150 couples, an ICC up to approximately .70 would yield the necessary number of cases to maintain a power of .80. Previous daily diary alcohol research suggests we can expect to see lower ICCs, likely ranging from .4 to .7 [39], yielding more power. Therefore, recruiting 150 couples (N=300 people) is more than adequate to provide sufficient power.

Recruiting large samples of sexual minorities is difficult to do in many localities, and therefore, we are recruiting participants from across the United States. We partnered with Community Marketing and Insights (CMI), a leading market research firm specializing in Web-based research with the LGBT community. CMI maintains a proprietary research-only panel of over 90,000 individuals who identify as sexual minorities and regularly take part in market research and health research (although psychological research studies are less common). CMI recruits participants to take part in this study from their existing panel and potential new panel members who may be interested in participating in research studies. CMI is consistently recruiting new research panel members through LGBT print, digital, and event outreach activities as well as LGBT-specific outreach on social media sites (eg, Facebook, Craigslist, and sexual minority-specific social media).

To be eligible for the study, both partners have to meet the following eligibility criteria: (1) aged 18 to 35 years, (2) self-identify as a cisgender woman (meaning self-identify as a woman and was assigned female at birth, ie, she is not transgender), (3) currently in a romantic relationship with a woman for at least 3 months, (4) see partner in person at least once per week, and (5) able to respond to daily surveys between 6 am and 12 pm for 2 weeks. In addition to these 5 criteria that must be met by all participants, at least one person in the couple must also meet the following 3 criteria: (6) only or mostly attracted to women, (7) drank alcohol at least 3 days in the previous 2 weeks, (8) drank 4 or more standard alcoholic drinks in 1 sitting at least once in the previous 2 weeks (ie, met criteria for a binge drinking episode).

These inclusion criteria were developed for several reasons. First, although we recognize that people with other gender identities (eg, transgender and nonbinary gender) and sexual identities (eg, bisexual women not in a relationship with a woman) are also at risk for experiencing both sexual minority stress and hazardous alcohol use, individuals with different gender and sexual identities likely experience stressors that differ from one another [29]. In other words, a transgender and cisgender woman’s or a lesbian and bisexual woman’s daily experiences might differ from each other. Thus, this study focuses solely on individuals who identify as cisgender female and are in a relationship with a woman. Second, we require the couples to see each other in-person at least once a week to avoid potential confounds related to being in a long-distance relationship. Third, although 1 participant has to report only or mostly attracted to women, her partner can describe her attraction in other ways (eg, attracted to men and women equally); this criterion was used both to obtain a sample size large enough to test study aims and to have some variability in attraction (eg, attracted to men and women) and identity (eg, lesbian, bisexual, and queer) that can be explored statistically. Finally, the alcohol inclusion criteria for 1 participant is used to identify participants belonging to the desired at-risk alcohol use population of interest and to increase the likelihood of enrolling couples in which at least 1 partner drinks with some frequency, resulting in occurrence of drinking during the 14-day daily diary reporting period.

All measures included in the baseline survey are described in Table 1. The measures are organized by those designed to test the primary study aims described above and secondary measures that will supplement testing of the primary aims. For each measure, we provide a brief description of the construct assessed. In addition, several modifications to measures were made and are worth noting. First, most measures in the baseline survey referencing a specific time frame were adapted to 3 months to be consistent throughout the survey and ensure we could examine these constructs during similar time periods. The time frames of the following measures were not adapted: the Conflict Tactics Scale [45], the Psychological Maltreatment of Women Inventory [46], the physical health questions Short Form-20 [47], the Daily Heterosexist Experience Questionnaire [48], and the Modified Eating Disorder Examination Questionnaire [49,50]. These measures were not changed to a 3-month time frame because their original time frames (eg, last year and last month) were more appropriate for addressing the study aims. Second, changes were made to several of the measures to be appropriate for same-sex female couples. These adaptions included changing male or gender-neutral pronouns to female pronouns and modifying sexual behavior items that require male partners. Finally, because of the length of the baseline survey, attention check items were added to ensure that participants are paying attention and selecting their answers carefully. Attention check items are included as items in some of the longer measures as well as stand-alone questions. Rationale for inclusion of these items is discussed in more detail in the Discussion section below.

The daily survey measures are described in Table 2 and are similarly organized by primary and secondary constructs of interest. Most of the primary and secondary measures included in the daily survey were adapted from existing daily diary or ecological momentary assessment (EMA) studies or were developed for this study to assess daily behaviors, particularly drinking behaviors and sexual minority stressors. Of note, when adapting items from existing scales, for example, to assess daily sexual minority stressors, we selected and edited items to be appropriate for a daily time frame (ie, would vary day-to-day). To balance the survey length on days when a participant did not drink or interact with her partner, filler items regarding media use, time management, and general social interactions were developed and are administered on nondrinking and noninteraction days. Finally, the format of the survey was optimized for mobile delivery. For example, visual analog slider scales are used instead of matrix tables in some cases so that all response options are visible to the participant when using a smaller screen (eg, on a smartphone). Additional details regarding the daily survey measures can be found in Table 2, and all daily survey items can be found in Multimedia Appendix 2.

The end-of-study survey contains questions concerning whether the participant believed her experiences were captured by the questions asked (eg, “While answering questions related to your gender and/or sexual identity, did you think that the questions you answered were inclusive of the way you describe yourself?”) as well as what was her preferred method of accessing the survey materials (ie, computer, tablet, or smartphone). Participants are also asked about their motivation to improve their health and whether they would be comfortable using mobile technology to monitor their physical and mental health. Survey items are included in Multimedia Appendix 3. The feedback gathered from the end-of-study survey will be used to gauge the participant experience in this study as well as provide an opportunity to improve future studies and survey materials.

The data being collected in this study are inherently multilevel in nature, with days (level 1) nested within individuals (level 2) nested within couples (level 3). In addition, our research questions incorporate aspects of mediation and moderated mediation. As such, we will utilize an approach that combines multilevel modeling (MLM) [86,87] with mediation pathways available through structural equation modeling. This combined approach, multilevel structural equation modeling (MSEM) [88], allows for bootstrapping of the models, which can accommodate non-normal underlying distributions, particularly relevant for alcohol use outcomes and indirect effects [89,90] such as those incorporated in our proposed models. Regarding missing data, like MLM, MSEM is robust to missing data (eg, skipped observations). In addition, analyses will be conducted in Mplus [91] using maximum likelihood estimation, which allows for the estimation of parameters using all available cases (ie, no listwise deletion for missing variables within a time point). Participants with complete data will be compared with participants with missing data to identify potential attrition biases, and significant predictors will be included as covariates in all subsequent analyses. The MSEM approach allows for the simultaneous examination of level 1 (daily variation), level 2 (person-specific levels), and level 3 (couple factor) associations among variables, including moderating effects that strengthen or weaken those associations.

Recruitment of couples into research endeavors is always a challenge, and recruiting same-sex couples creates additional challenges. Consequently, we devoted substantial effort to develop and refine processes for participant recruitment and retention. Below, we discuss the challenges faced when designing and implementing Project Relate.

EMA studies of health behaviors [99], as well as EMA studies conducted in the context of couples or families (eg, in which more than one person in a couple/family reports EMA data) [100], are rapidly increasing in popularity because they can examine more nuanced associations that consider individual, dyadic, and contextual variables that may influence behavior within people and couples over time. Despite growing interest in daily health factors and related stressors such as minority stress and alcohol use, research on daily experiences of sexual minority individuals lags far behind [101]. There are 2 important challenges that researchers are facing when conducting this work. First, as has been discussed at length elsewhere with respect to health behavior theories [102], existing theories may need to be refined or redeveloped to be able to guide our understanding of within-person processes that occur over relatively short time frames (eg, minutes, days, and weeks). Minority stress theories, including those that guided the development of Project Relate [8,9] at times, suggest predictions that involve within-person processes that may occur over short time frames but have largely been tested using cross-sectional designs (testing between-person processes) or traditional longitudinal designs (testing longer-term within-person processes). Project Relate is designed such that it will be able to test aspects of minority stress theories and presents an opportunity to extend and refine these theories to further the understanding of within-person processes in daily life.

A second challenge we encountered when designing Project Relate is the relative lack of well-established, validated measures that are appropriate for daily assessments. In Table 2, we provide a description of where we adapted measures from, based on either previous daily diary research with heterosexual couples or from existing retrospective measures. Importantly for Project Relate, we were unable to identify a measure of sexual minority stress that was appropriate for daily administration and, thus, created a daily sexual minority stressor measure [41] to use in this study. The lack of validated measures that are appropriate for daily or EMA studies highlights the importance for future work in this area to create new measures.

At present, the ability to provide education, mental health counseling, and interventions that are optimally effective for SMW is limited by our lack of understanding of key variables to incorporate in these efforts. Furthermore, the limited data available about relationship factors and alcohol use among SMW are typically derived from 1 respondent who provides information about herself and her partner. To address this limitation, our approach requiring both partners to participate is generating higher quality data on both partners’ affective, behavioral, and relationship experiences. In addition, we are including other variables both specific to SMW and heterosexual couples that may impact hazardous alcohol use. We believe it is critical to consider how couple-level dynamics/variables may impact hazardous alcohol use and related issues. In addition, we believe that understanding both daily and between-subjects’ mechanisms that may increase risk for hazardous alcohol is critical as is identifying protective factors that may attenuate this risk. By testing an empirically informed model that articulates probable within- and between-person factors and relationship variables that increase hazardous alcohol use, we believe this work will provide a necessary intermediary step toward informing culturally tailored prevention and intervention studies for SMW and contribute to a goal of the Healthy People 2020 initiative to improve the health and well-being of SMW.