Keywords

Until recently, LGBTQ-parent families have been largely invisible in surveys of family life. Yet new understandings of LGBTQ-parent families have emerged in the last decades, and the analysis of several national- or population-based data sources has added new perspectives to the knowledge base on LGBTQ-parent families. It was not until the 1990s that scholars, along with the general public, began to recognize LGBTQ-parent families as a legitimate family form that was not going to go away. The growing research literature on LGBTQ-parent families during the 1990s (see Goldberg, 2010) prompted the designers of large-scale family surveys to begin to consider nonheterosexual family forms. Thus, new possibilities emerged with, for example, the US Census (Simmons & O’Connell, 2003) and the National Longitudinal Study of Adolescent to Adult Health (Add Health; e.g., Wainright, Russell, & Patterson, 2004), which began to include the possibility for respondents to identify same-sex partners in families and households.Footnote 1

With the growing visibility of LGBTQ people, a growing number of large-scale datasets in the United States and around the world have been extended to include attention to LGBTQ-parent families, and for the first time, population samples of LGBTQ people are emerging. These studies offer the potential to greatly advance understandings of contemporary families. In this chapter, we consider the use of large-scale secondary data sources (many of which are population-based and nationally or regionally representative) for the study of LGBTQ-parent families. We include a detailed list of large-scale secondary data sources in an appendix at the end of this chapter. We also discuss the advantages and opportunities that such datasets offer, as well as the challenges that define working with secondary data on such an understudied and marginalized population.

Since the last edition of this volume (Russell & Muraco, 2013), there has been a dramatic shift in the zeitgeist related to reproducible research, transparency in data use and analysis, and data archiving (Winerman, 2017). The impact of this shift for representative data of LGBTQ-parent families is substantial: since the last edition, we have located 47 additional representative datasets which allow for identification of LGBTQ-parent families. The identification of these data sources appears to be due both to the increasing inclusion of LGBTQ measures in population data sources and to greater access to data through public data archives and improvement of the quality of documentation of public data. Two large data enclaves that we utilized to locate these sources were the Inter-university Consortium for Political and Social Research (ICPSR; https://www.icpsr.umich.edu/icpsrweb/) archive and the Integrated Public Use Microdata Series (https://www.ipums.org/). However, there are other new advances beyond these archives. For example, the United Kingdom has a data archive similar to ICPSR (https://www.ukdataservice.ac.uk/). Google also recently released a search engine that searches for publicly available data (Castelvecchi, 2018). Some universities maintain data archives (e.g., Harvard: https://dataverse.harvard.edu/; Princeton: https://opr.princeton.edu/archive/), and there are also individual efforts to accumulate data for a specific population (e.g., http://www.lgbtdata.com/).

We consider several types of datasets that hold potential for the study of LGBTQ-parent families (the appendix includes examples of each of these types of datasets). First are population-based, representative surveys of the general population that may be local, regional, or national in scope and are typically designed to allow for generalizations to the larger populations that they represent and that include measures to identify LGBTQ-parent families. Examples are the US Census, which includes information on same-sex couple householders, or the Add Health study, which includes questions about young adult sexual identity and orientation as well as marital or family status. A subgroup group among representative studies are large-scale cohort studies: The 1970 British Cohort Study (BCS) and the 1958 National Child Development Study (NCDS) are unit in that the design of both studies includes a complete population (rather than a “sample” per se) at a given point in time (all births in one week, followed across childhood and into adulthood). Both studies ask respondents in adulthood about their marital (or marriage-like) relationships and household composition, including information about gender and how study members are related to other householders. Results from these studies are generalizable to similar age cohorts.

A second group of studies are large-scale studies but are not representative of or generalizable to a broader population. Nonrepresentative local, regional, or multi-site samples that provide sufficient numbers of LGBTQ-parent families for study may not be specifically generalizable to a broader population, but may illuminate important associations or processes that characterize LGBTQ-parent family life. An example is the National Longitudinal Lesbian Family Study (NLLFS) .

A third group of studies have emerged since the first edition of this chapter was written: population-based studies specific to LGBTQ communities. Several studies, some of which at the time of this writing are still in the field, offer the first population-based, representative samples of LGBTQ and transgender US populations: the California Quality of Life Survey (CQLS), the Generations Study, the TransPop Study, and the National Couples’ Health and Time Study.

The potential of these data sources within the context of research on LGBTQ-parent families is important because, historically, research on LGBTQ-parent families developed from and was grounded in a particular set of very different methodological approaches and disciplines. Early questions about child adjustment (with particular attention to sexual orientation, gender identity, and psychological adjustment) in LGBTQ-parent families emerged from the fields of psychology, child development, and family studies, fields that were already attuned to diverse family forms (Patterson, 1992). Further, studies based on small samples of distinct populations that are not population-based were typical in those fields: Early studies were based largely on community or regional samples (Patterson, 2006). These studies focused on child adjustment and the well-being of mothers, both because these constructs were central in these fields and because scholars were responding to fears that lesbians were mentally unwell and would therefore negatively influence their children (Goldberg, 2010). Over time, LGBTQ-parent research extended to include parenting, family processes, and the well-being of LGBTQ parents (Goldberg, 2010). As this body of work grew, it attracted the attention of other fields of study relevant to families and children, including demography, sociology, economics, and health. Thus, new studies from the population sciences provide a vantage point for understanding LGBTQ-parent families that were population-based and generalizable and that allowed comparisons with heterosexual-parent families (see Biblarz & Savci, 2010, for a review).

Today there are a number of large-scale datasets available that afford the possibility of studying LGBTQ-parent families; however, most have rarely or never been used for this purpose (e.g., the Survey of Income and Program Participation [SIPP]). Some nationally representative studies of families and households in the United States have begun to include questions about the LGBTQ identity status of adult householders, many of whom have children (e.g., the Survey of Income and Program Participation (SIPP), the Panel Study of Income Dynamics (PSID), and the US Census). Other large-scale studies began as population-based, longitudinal studies of children: As the study members have grown up and been followed into adulthood, many have become LGBTQ parents themselves. For instance, it is possible with the Add Health study to follow those who reported same-sex attractions or relationships in adolescence into adulthood, affording the opportunity to study their coupling and parenting in adulthood. The prospective birth cohort studies such as the NCDS and the BCS make it possible to identify same-sex couple and parent households when cohort members are adults (Lau, 2012; Strohm, 2010).

Finally, the analysis of representative data of LGBTQ-parent families has been invoked in the promotion of civil rights for LGBTQ people (e.g., Gates, 2013), yet misinterpretation of data has perpetuated misinformation about LGBTQ families. A critical example emerged recently, when findings regarding the well-being of children of LGBTQ parents were inaccurately reported from the New Family Structures Study (NFSS) and used to support legal cases against marriage for same-sex couples (see Manning, Fettro, & Lamidi, 2014, for a discussion). After the original report was published, over 150 social scientists endorsed a letter rejecting the academic integrity and intellectual merit of the study (Gates, 2012b; Perrin, Cohen, & Caren, 2013; Umberson, Cavanagh, Glass, & Raley, 2012), and reanalyses of the data using the NFSS have invalidated the initial findings (Cheng & Powell, 2015). The controversy surrounding the misuse of the NFSS underscores the responsibility of primary investigators, as well as reviewers and publishers, to attend to the political implications of studies of LGBTQ parenting and families.

In this chapter, we review findings based on some of these existing data sources while identifying challenges as well as advantages of using population-based representative datasets to study LGBTQ-parent families. Given the growing number of large-scale representative studies that now allow for the study of LGBTQ-parent families, we identify a number of areas of research that are largely understudied but from which much could be learned in the coming years.

Challenges in Using Secondary Data to Study LGBTQ-Parent Families

There are a number of challenges in any research based on analyses of existing secondary data sources, some of which are further complicated in studies of LGBTQ-parent families. We consider challenges associated with conceptual breadth as well as measurement inclusion in existing studies. The use of secondary data is relatively new among researchers of LGBTQ-parent families, in part because measures for identifying LGBTQ people and LGBTQ-parent families have only recently been included in secondary data sources and also in part due to the origins of the study of LGBTQ-parent families in disciplines where secondary data analysis was less common. Thus, we also briefly review other basic challenges and suggest strategies to address these challenges.

Conceptual Challenges

At the most basic level, scholars who use secondary datasets must negotiate the discrepancies between their research questions and available data (Hofferth, 2005; Russell & Matthews, 2011). Unless the researcher was directly involved with the data collection process, it is unlikely that full information will be available to address their precise questions. However, they may find that sufficient data exists to partially address their questions or to allow an adjustment of the question based on available data. Most datasets that are focused on broad populations have been developed by economists and sociologists who may not be concerned with many of the constructs that are important to family scholars and psychologists, such as individual or family histories and processes (Russell & Matthews, 2011). Thus, the researcher undoubtedly will be required to be flexible with the conceptual design and creative in posing research questions that can be addressed with available data. At a fundamental level, this is a conceptual problem but one that typically plays out as problems with measurement (what is measured and how).

The most obvious example of this conceptual challenge is that most of what is known from nationally representative studies are based on families in same-sex couple households rather than couples or individuals who specifically identify themselves as lesbian, gay, bisexual, or transgender. For example, since 1990, the US Census has included the option that a primary householder may report an “unmarried partner.” It is difficult to imagine how one could construct a single question to accurately ascertain LGBTQ-parent family status, and we know of no study that does this. Rather, researchers must combine multiple questions to identify households with children in which the parents are same-sex partners and householders or engage in same-sex sexual practices or behaviors. Measures of self-identification as LGBTQ on large-scale surveys continue to be relatively rare; however, participant gender and the gender of their partner/s may be available (Gates & Romero, 2009).

Another conceptual challenge for using secondary data sources to study LGBTQ-parent families is that many of the important constructs in this field are LGBTQ-specific and are unavailable in population-based studies. Thus, important questions specific to LGBTQ-parent families may be missing. For example, how and why do LGBTQ couples decide to have children? How do same-sex couples manage historically gendered parenting roles (Goldberg, 2012; Goldberg, Smith, & Perry-Jenkins, 2012)? What is the impact of LGBTQ-specific minority stress (the experiences of stigma, prejudice, or discrimination due to LGBTQ status; Meyer, 2003) on parenting options, processes, and family life (Chapman et al., 2012)? These questions have been addressed using samples of LGBTQ-parent families, but not population-based samples.

Overall, most of the research literature on LGBTQ-parent families concern constructs that are generalizable to all populations: child adjustment, parent relationship quality, and parenting practices. Yet for questions about LGBTQ-specific dimensions of social or family life (e.g., LGBTQ-specific discrimination; methods for becoming parents and related decision-making), secondary data sources designed for the general population may simply not be suitable.

Measuring LGBTQ-Parent Families

In terms of measurement, there are a number of challenges specific to the availability of measures in secondary data sources. Research based on any one data source must be interpreted in light of other studies, yet there is variability across studies in the specific measures that can be used to identify LGBTQ-parent families. For example, several federally initiated surveys such as the Behavioral Risk Factor Surveillance System (BRFSS) surveys are administered by states, and although some states have begun to include measures that would allow the study of LGBTQ individuals and thus LGBTQ parents and families, the measures are not consistent across states.

Within BRFSS, for example, Massachusetts is unusual because it includes measures since 2000 (some that differ across the years) for same-sex sexual behavior as well as sexual identity (whether one identifies as lesbian, gay, or bisexual); beginning in 2007, a measure for transgender identity was included (Behavioral Risk Factor Surveillance System, 2011). As of the 2016 survey, 26 states included the BRFSS sexual and gender identity optional module as part of their survey, leading to a number of studies that account for presence of children in household studies of LGBTQ individuals (Boehmer, Clark, Lord, & Fredman, 2018; Cranney, 2016; Fredriksen-Goldsen, Kim, Barkan, Muraco, & Hoy-Ellis, 2013; Gonzales & Henning-Smith, 2017). Yet, no one, to our knowledge, has used this dataset to directly examine LGBTQ-parent families. One challenge is that not all participating states include the same measures, which hinders cross-state comparison and prevents the study of how state characteristics—such as state laws, policies, or practices—affect LGBTQ-parent families.

There are also a number of measurement challenges particularly relevant for longitudinal studies of LGBTQ-parent families. Sometimes the measures used in prospective studies change over the span of the study (measures for young children will not be identical to those for adolescents and adults; Russell & Matthews, 2011). For repeated cross-sectional studies, there are challenges when measures are changed. For example, the US Census maintains that, as a result of flaws in the way they classified same-sex households in 1990,Footnote 2 the data from 1990 and 2000 cannot be compared (Smith & Gates, 2001). In addition to data errors that result from classifications, some argue that there has been notable change over only a few decades in the diversity of sexual self-identity labels: Some individuals or couples may prefer, for example, the term “queer” to “gay” or “lesbian” (Morandini, Blaszczynski, & Dar-Nimrod, 2016; Russell, Clarke, & Clary, 2009). Further, individuals and couples may change their preferred identity label over time. The existing variability in measures across studies may only be compounded by changes over time in the ways that LGBTQ parents self-label and disclose their identities and family statuses to researchers.

Finally, as the legal basis for LGBTQ family relationships has been in flux, definitions and measures have shifted (and likely will continue to shift). For example, since the first edition of this chapter in 2013, three times as many countries now allow marriage for same-sex couples (10 countries in 2013; 30 countries as of this writing). As legal statuses change, personal meanings change as well. Prior to the legalization of marriage of same-sex couples, couples mostly cohabited, yet in the most recently available data, close to two out of five same-sex couple are married. Same-sex couples are still more likely to cohabit, yet they marry and divorce at rates similar to different-sex couples (Gates, 2015). Beyond marriage, there are other ways that couple and family life is shifting demographically, with implications for the meanings—and measures—of households, parents, and families. For example, “living apart together” (LAT) relationships (nonresidential partnerships) are gaining visibility in Western countries, and LGBTQ people are more likely to be living in these forms of family (Gabb & Fink, 2017; Strohm, Seltzer, Cochran, & Mays, 2009). Such family structural diversity has implications for how individuals and families are captured in population samples (i.e., LAT individuals are often recorded as “single”) and thus who may be included or excluded when we study LGBTQ-parent families.

To address these challenges, it is crucial at a most basic level to carefully sort out the opportunities and limitations of the match between one’s research question and the data available through secondary sources. For example, one could use the National Health Interview Survey (NHIS) to examine same-sex couple household access to health care (the NHIS collects respondents’ gender and the gender of others in the household and their relationship to the respondent). However, if one’s theory of health-care access and utilization relies on arguments about homophobic discrimination in the health-care setting, the absence of data for householders’ sexual identities is crucial. Having a clear understanding of the alignment between one’s research question and the secondary dataset will help formulate a strong case for a study’s rationale and ultimately for persuading reviewers that the opportunity the data affords outweighs any limitations. In the example above, it may be an important first step for the field to simply document differences in health-care access and utilization based on householder couple status. The researcher must be flexible and creative in matching the research question to available data (Russell & Matthews, 2011). In addition to the need for conceptual and analytic flexibility and creativity with regard to measurement, we turn to several other basic challenges and suggestions for addressing them.

Methodological Challenges

Becoming familiar with a large and complex existing dataset is time-consuming, and researchers often overlook the “costs” of learning. One must understand a study’s design, data structure, and distinct methodological characteristics that may influence analyses (Hofferth, 2005). Studies often employ complex sampling designs which require specialized statistical analytic techniques: Researchers may need to learn methods for adjusting for complex sample designs (e.g., nested samples or cluster designs) or methods for the use of weighted data responses (Russell & Matthews, 2011). There is a common perception that using existing data simply circumvents a data collection phase of research; however, recoding existing variables into useful constructs is time-consuming (after 20 years of experience, the first author has found it necessary to estimate the time it will take and multiply by four!). At the same time, there are often opportunities for learning: Many large-scale studies have user groups or conferences designed to allow researchers to network with one another.Footnote 3 These networks offer possibilities for collaboration or the sharing of strategies for analysis, as well as for learning about others’ questions and research efforts. Although when working with publicly available data there is a possibility of having one’s idea “scooped” (i.e., taken, tested, and published before one is able to do so oneself), participating in scholarly networks of study users can keep one abreast of developments by other scholars in the field.

Professional Challenges

Finally, a unique challenge in using secondary data is potential professional costs. In many fields and at many institutions, for various reasons, original data collection may be more highly valued. In some fields, original data has value in itself. At the same time, in research-intensive institutions where grant funding is an important marker of career success, the higher costs and thus larger extramural grants required to collect data may be valued above analyses of secondary data. As research-focused institutions place greater demand on researchers to receive external funding, it is important to acknowledge that grants for secondary data analyses tend to require less overall time and staff. The challenge of acquiring grant funding for LGBTQ-parent research using secondary data analysis may therefore be a disincentive for junior scholars concerned with meeting academic tenure requirements. Yet despite these challenges, the availability and access to a growing number of secondary data sources offers a new array of research possibilities for studying LGBTQ-parent families.

Advantages of Secondary Data for Studies of LGBTQ-Parent Families

Having discussed some of the challenges, we now describe the potential advantages of using large-scale or population-based secondary datasets for the study of LGBTQ-parent families. Important advantages include generalizability to broad populations, large sample sizes (including sufficient numbers of underrepresented populations and power for statistical analyses), and the ability to conduct comparative analyses with populations of heterosexual-parent families. Some data sources allow for additional advantages: They may be longitudinal, include data from multiple reporters, allow insights about multiple contexts and processes of development, or allow cross-historical or cross-national comparisons (Russell & Matthews, 2011). An obvious practical advantage is low cost and ease of access (Hofferth, 2005) compared to the labor-intensive work of sample selection and data collection to begin a new study of LGBTQ-parent families.

First, the possibility for making generalizations to broader populations of LGBTQ-parent families is a crucial advantage that can advance this field of study. For example, the 2000 US Census counted 594,391 same-sex couples (Simmons & O’Connell, 2003); of those same-sex couples, about a quarter reported a child under the age of 18 living in their household (Gates & Ost, 2004). Never before had there been a true census of LGBTQ-parent families (or more accurately, households headed by parenting same-sex couples): For the first time, researchers asserted that they had “identified same-sex couples in every state and virtually every county in the United States” (Sears, Gates, & Rubenstein, 2005, p. 1) and provided population estimates of the proportion of households headed by same-sex couples who are parenting in every state (the proportion of same-sex couples out of all households ranged from .27% to .80%). Notably, the same statistics have also been challenged because, with data only available for relationships among adult householders and thus on couples, they dramatically undercount the total number of single LGBTQ people and single LGBTQ-parent families in the United States. Yet, these results were groundbreaking for establishing the presence of these families for policy makers and planners. The results have also been instrumental in challenging stereotypes about LGBTQ-parent families, for example, that they are typically White, affluent, coastal, and urban. Indeed, these data have established that, although same-sex couples without children are more likely to reside in California and Vermont, same-sex couples with children are more likely to reside in rural states (Mississippi, Wyoming, Alaska, Idaho, and Montana; Gates, 2013, Gates & Ost, 2004). Yet California is where gay and lesbian adoptive and foster families are most likely to live (Gates, Badgett, Macomber, & Chambers, 2007). Further, same-sex couples of color are more likely to have children compared to their White counterparts (Bennett & Gates, 2004; Black, Sanders, & Taylor, 2007; Carpenter & Gates, 2008; Gates, 2012a; 2013).

Second, large sample sizes are beneficial because they allow for both the study of small and often marginalized subpopulations and statistical power for complex analyses (Russell & Matthews, 2011). Obviously, LGBTQ people and LGBTQ-parent families are present in all large-scale studies: The question is whether data are obtained to acknowledge them or whether they are invisible. Given their very small proportion within the total population, only huge studies will yield sufficient numbers of LGBTQ-parent families to allow for statistical analyses. For example, over 20,000 adolescents were included in the in-home portion of the Add Health study collected in 1994–1995; over 17,000 of their parents completed surveys. Wainright et al. (2004) were among the first investigators to use these data to investigate the well-being of adolescents growing up in same-sex parent households. They investigated psychosocial adjustment, school outcomes, and romantic relationships for 44 adolescents determined to be parented by same-sex couples based on parent reports of their gender and the gender of their partner (all were mothers; there were too few two-father families for inclusion in the study). Compared to a matched group of adolescents from heterosexual-parent families, no differences were found in adolescent adjustment (Wainright et al., 2004).

The Add Health study was the first of its kind based on a nationally representative sample to allow comparisons across family types, yet even with over 17,000 responding parents in that study, only 44 adolescents parented by female same-sex couples were identified. It is important to note that these low numbers may also be explained by heteronormative assumptions in the design of the household measures in the original waves of the Add Health study that (a) did not ask the sexual orientation/identity of responding parents, (b) gave preference to female parents on the parent survey, and (c) precluded the possibility for adolescents to indicate on the adolescent-reported household roster that an adult living in the household could be the same-sex partner of a parent.

Add Health data have since been utilized for a number of studies examining children of mothers in same-sex couples. Wainright and Patterson (2006) found that regardless of family type, adolescents whose mothers described closer relationships with their children reported less delinquent behavior and substance use. Further, Wainright and Patterson (2008) found that regardless of family type, adolescents whose mothers described closer relationships with their children reported higher-quality peer relations and more friends in school. These findings support the assertion that the quality of the parent-adolescent relationship better predicts adolescent outcomes than family type (Wainright & Patterson, 2006, 2008). Future studies should examine whether such findings remain true for children of male same-sex couples.

An additional benefit of very large samples is the possibility to study differences among LGBTQ-parent families based on demographics such as race/ethnicity, class, age, and gender. Gates (2013) reports that among same-sex couples in the United States, people of color are twice as likely as their White counterparts to have children under 18 living at home: 41% of non-White women in same-sex couples have children under 18 living at home, compared to 23% of their White counterparts. Among non-White men in same-sex couples, 20% have children living at home relative to 8% of their White counterparts (see Bennett & Gates, 2004; Black et al., 2007; Carpenter & Gates, 2008; Gates & Romero, 2009). Some studies have also begun to measure socioeconomic diversity among LGBTQ-parent families, displacing stereotypes of affluence and reporting higher rates of poverty relative to their heterosexual counterparts (Cenegy, Denney, & Kimbro, 2018; Gates, 2013; Schneebaum & Badgett, 2018; Sears & Badgett, 2012). These findings are groundbreaking in identifying more diversity in LGBTQ-parent families than has been represented in the existing literature, which has been largely derived from community-based samples of LGBTQ-identified parents who, until recently, consisted of primarily White lesbian mothers.

Another advantage to the use of population-based data sources is that some utilize longitudinal designs (Russell & Matthews 2011). Some, like the General Social Survey (GSS) and the National Health Interview Survey (NHIS), collect data longitudinally by collecting representative data across time (but do not follow the same participants prospectively from year to year); few if any published studies based on these data have examined LGBTQ-parent families. Other datasets, such as Add Health, the National Child Development Study (NCDS), and the British Cohort Study (BCS), allow for the study of individuals across time so that hypotheses concerning human development and change can be explored. The members of the Add Health and both the NCDS and BCS cohorts are now adults or young adults, many of whom are becoming parents. These datasets offer unique opportunities to study characteristics from the early life course (childhood and adolescence) that may be associated with the well-being of LGBTQ adults and their children or the adult lives of children who were parented in same-sex households; again, we are aware of no studies that have taken this approach.

Other benefits of large-scale survey studies (e.g., Fragile Families, https://fragilefamilies.princeton.edu/) include perspectives from multiple reporters such as children and parents, which allow for more than one perspective on family life. Finally, another potential advantage is the ability to conduct cross-historical or cross-national comparisons (Russell & Matthews, 2011). For example, a component of the GSS, the International Social Survey Program, was specifically developed to allow for cross-cultural comparisons between the United States, Australia, Great Britain, and West Germany. Such surveys may allow for future comparisons of LGBTQ-parent families across multiple countries.

New LGBTQ-Focused Population Studies

Several methodological innovations have allowed for in-depth study of LGBTQ individuals and families drawing from general population samples. As marketing and research samples have grown in size and online methods of data collection have been developed, new possibilities have emerged for reaching LGBTQ populations (see chapter “Methods, Recruitment, and Sampling in Research with LGBTQ-Parent Families”). In one of the first examples to use two-phase sampling, LGBTQ participants in the California Health Interview Survey (CHIS) were recontacted for participation in the California Quality of Life Survey (CQLS) which included all participants of CHIS who reported their sexual identity as gay, lesbian, or bisexual or as having had same-sex sexual activity and who agreed to participate in future surveys on the CHIS (Strohm et al., 2009). The CQLS was designed to include questions specific to LGBTQ individuals and families.

More recently, the first nationally representative probability study of LGBTQ adults, the Generations Study (http://www.generations-study.com/), and the first nationally representative probability study of transgender health, the Transpop Study (http://www.transpop.org/), were begun. These interdisciplinary study teams are composed of scientists across fields including psychology, sociology, demography, human development and family sciences, and public health—a testament to a growing recognition of the importance of diverse perspectives in the study of LGBTQ lives. These projects will allow for some of the first nationally representative evidence from the United States about the lives and health of LGBTQ and transgender adults and provide more accurate estimates related to stigma and health.

Lastly, researchers are in the process of collecting data for the National Couples’ Health and Time Study (http://u.osu.edu/kamp-dush.1/about-me/), which will provide the first representative sample of same-sex couples’ family functioning, experiences of stigma, and coping. This dataset will address several critical gaps in prior LGBTQ-couple data, including limitations to analysis of dyadic data, a lack of detailed information about family functioning and stress mechanisms, and limited ethnic/racial diversity.

Implications for Future Research

There is a rich tradition of population-based survey research in the social and behavioral sciences that has provided a baseline for scientific and public understanding of the social and economic health and development of families, yet for generations, LGBTQ people and families were invisible. Developments in recent decades have changed that. More large-scale surveys now include possibilities to identify, study, and understand LGBTQ-parent families. Such large-scale representative studies are one path for building scientific understanding of LGBTQ-parent families. The appendix includes descriptions of relevant data sources, some of which to our knowledge have never been used for the study of LGBTQ-parent families. To compile our appendix, we used five sources: (a) our knowledge of available datasets, (b) Inter-university Consortium for Political and Social Research (ICPSR), (c) UK data archive; (d) a search of EBSCO host for articles using representative data since the publication of the first edition of this book, and (e) a search of Integrated Public Use Microdata Series (IPUMS) data for national census data. The search had two main criteria: The data had to be representative of a population (i.e., national, state, regional) and have the potential to identify same-sex parents or lesbian, gay, bisexual, trans, or queer parents.

In addition to the challenges and opportunities we have discussed, we note some areas in the study of LGBTQ-parent families that have been particularly underexamined and for which the use of secondary data sources may provide important new possibilities. Gay fathers are fewer in number than their female counterparts, which may help to explain why they have been underrepresented in existing studies of LGBTQ-parent families. In 1990, one in five female same-sex couples was raising children compared to one in twenty male same-sex couples (Gates & Ost, 2004). By 2000, one in three female same-sex couples and one in five male same-sex couples were raising children (Gates & Ost, 2004). Data from the American Community Survey from 2014 to 2016 found that 8% of male same-sex couples were raising children while 24% of female same-sex couples were raising children (Goldberg & Conron, 2018). Although datasets such as the National Longitudinal Lesbian Family Study (NLLFS) exist to expand research on same-sex female couples, no existing data source is comparable for the study of male same-sex couples raising children (Gartrell et al., 1996). The NLLFS is not population-based and thus is not representative of all lesbian-parent families; however, it is a large sample that includes a birth mother and a co-mother with at least one child from whom data have been collected five times (before the child was born and then when the child was 2, 5, 10, and 17). Data from the NLLFS have allowed researchers to explore the lives of lesbian mothers to debunk common myths. Results find, for example, that the development of psychological well-being in children of lesbian mothers over a 7-year period from childhood to adolescence is the same for those with known and unknown donors (Bos & Gartrell, 2010); no similar information exists about the children of gay fathers using known and unknown donors. Although some studies are beginning to address the importance of examining gay male parenting (e.g., Golombok et al., 2014; Green, Rubio, Rothblum, Bergman, & Katuzny, 2019; Carneiro, Tasker, Salinas-Quiroz, Leal, & Costa, 2017; see chapters “Gay Men and Surrogacy” and “LGBTQ Adoptive Parents and Their Children”), more attention to gay male parenting is warranted, especially using longitudinal data.

Further, there are few, if any, studies based on population-representative data sources that examine bisexual- or transgender-parent families (there are few existing studies of bisexual or transgender persons and family life in general; see chapter “What Do We Now Know About Bisexual Parenting? A Continuing Call for Research”, for a review of bisexual-parent family research, and see chapter “Transgender-Parent Families”, for a review of trans-parent family research). Of the sources included in the appendix, the Behavioral Risk Factor Surveillance System (BRFSS, select states) , British Cohort Study (BCS), and the National Child Development Studies (NCDS) include measures that allow identification of transgender people. Even though these sources are largely untapped, they afford unprecedented opportunities for scholarship. Lastly, little is known about LGBTQ-parent families and socioeconomic status; much of the existing research focuses on middle-class LGBTQ-parent families (see chapter “LGBTQ-Parent Families in the United States and Economic Well-Being”). Yet studies using new sources of population data have shown, for example, that it is socioeconomic status rather than same-sex family structure that is associated with children’s economic well-being (Brown, Manning, & Payne, 2016) and that LGBTQ parents in middle and upper socioeconomic classes are more protected from discrimination (Cenegy et al., 2018).

In conclusion, we have identified challenges as well as opportunities for scholars who may pursue the study of LGBTQ-parent families through analysis of secondary data sources or large-scale surveys. There are many new possibilities for the study of LGBTQ-parent families (and even more possibilities to study LGBTQ individuals). To date, findings from such studies have been groundbreaking. Not only have they demonstrated, for example, that child and family well-being do not differ in LGBTQ-parent and heterosexual-parent families (Bos, Knox, Rijn-van Gelderen, & Gartrell, 2016; Gartrell, Bos, & Koh, 2018; Rosenfeld, 2010; Wainright et al., 2004; Wainright & Patterson, 2006, 2008), but also they have dispelled myths about who LGBTQ parents are and where they live (Gates, 2013; Gates & Ost, 2004; Gates & Romero, 2009) and have shown simply—yet radically—that LGBTQ-parent families are everywhere (Simmons & O’Connell, 2003; see chapter “LGBTQ-Parent Families in Community Context”). There are remarkable possibilities waiting in these data sources. They are opportunities to propel the field of LGBTQ-parent families, and thus our understanding of all contemporary families, forward.