What can we learn from studies of children raised by gay or lesbian parents?

The papers by Loren Marks and Mark Regnerus, in tandem, provide a challenge to the broad scholarly consensus that children raised by gay or lesbian parents suffer no deficits. Let me be clear, these papers do not prove anything. Instead they offer reasonable arguments for (1) more caution when drawing strong conclusions based on the available science; and (2) that data based on probability samples might lead to different findings than are currently drawn from available data.

The immediate concern that drives the studies on children raised by gay or lesbian parents is obvious. We are currently undergoing a cultural and legal battle over whether same-gender partners should be allowed to marry and this debate is, in part, grounded on scientific conclusions about gay or lesbian couple’s fitness as parents. The 2010 decade review of the literature on lesbian, gay, bisexual and transgender families by Timothy Biblarz and Evren Savci in the Journal of Marriage and Families illustrates this connection clearly:

In the 1990’s, marriages between same-gender partners were not legally recognized anywhere in the world and families formed by gay men, lesbians, and bisexual and transgender people faced considerable opprobrium and intolerance. Researchers were documenting what most social scientists already knew but what much of the public, perhaps inundated by “virtual social science” (Stacey, 1997), did not: that sexual orientations and gender identities per se have almost nothing to do with fitness for family roles and relationships, including parenting

(Biblarz & Savci, 1010, page 480).

While the cultural and legal battles over same-sex marriage certainly are behind the immediate concern, the social science interest in children raised by gay or lesbian parents is also driven by the larger revolution in family life that children are experiencing. The basic contours of the changes are well known. In 1960 88% of children were living with two parents. By 2011 this proportion had fallen to 69% (U.S Bureau of the Census, 2011a). Today, children can be found in variety of alternative family arrangements. For example, nearly 11% of children live with divorced mothers, 1.7% with divorced fathers, and 3.9% with unmarried cohabiting parents (U.S. Bureau of the Census, 2011b). Some children in non-traditional families are parented by lesbian mothers or gay fathers, although, unlike children in other family arrangements, the number is unknown. A commonly cited estimate is between 1 and 9 million children in 1999 (Stacey & Biblarz, 2001).

In tandem with these behavioral changes has been a growing tolerance of less traditional lifestyles. Fifty years ago there would have been broad consensus that these social changes were bad news. This is not the case anymore. A recent opinion poll by the Pew Center for American Life found that a majority of Americans viewed the increases in cohabitation, unmarried couples raising children, women not ever having children, and, significantly, gay/lesbian couples raising children, as either a positive change or that they are benign. Only single women having children was viewed with concern by more than 50% of adults (Pew Research Center, 2010). In short, the current debates about same-sex marriage, embedded in the profound changes in the lives of children and the growing public tolerance of alternative families, together form the backdrop of growing scholarly interest in the well-being of children raised by gay or lesbian parents.

Let me begin with the paper by Loren Marks. On the surface this paper appears to be yet another review of the literature on same-sex parenting and child well-being. The review, however, takes a decidedly different tack than other published reviews: instead of focusing on findings, Marks focuses on the quality of the science behind the findings that formed the basis of the official brief on this subject by the American Psychological Association published in 2005. The Marks paper reads like a legal brief: precise, careful, dry prose festooned with footnotes. The bottom line of this review is understated, but clear: The conclusions in the APA report are unwarranted.

While Marks offers seven criticisms, in my view three seem especially relevant: (1) the limitations of the samples; (2) lack of suitable comparison groups; and (3) the lack of attention to long-term outcomes. None of these complaints are new. Most of the studies I am aware of note one or more of these limitations in their data. The difficulty is addressing these limitations.

Generating the kind of samples that allow some degree of generalization is exceptionally challenging. These sub-groups are not only small, but, based on the demographic work of Gary Gates (2011), widely dispersed across the United States, often living in markedly different contexts. Several recent studies draw on national probability samples in an attempt to deal with this, but each suffers from limitations. A paper by Michael Rosenfeld (2010) utilizes census data to produce very large samples of gay and lesbian headed same-sex couples. However use of census data also presents significant limitations: the focus of same-sex parenting is limited to couples, and the variables that might be used for measuring outcomes are sparse. Papers by Wainright and colleagues (Wainright, Russell, & Patterson, 2004; Wainright & Patterson, 2006; 2008) take advantage of the large probability sample of teenagers in the National Study of Adolescent Heath. However, these analyses are limited to 44 adolescents whose mother reported living in a marriage or marriage-like relationship with a person of the same sex.

Compounding this challenge is the stigma attached to gay and lesbian lifestyles that likely makes volunteering to participate in studies highly selective on level of education, social connectedness to a gay or lesbian community, and perceived “success” as a parent. Confronting these sample deficiencies with new data will take considerable resources. If the past is any indication, the commitment of these resources is most likely to happen when (1) the questions that need to be addressed are of central concern to society and public policy; (2) there is general agreement that current science is inadequate; and (3), that there is a lack of scientific consensus about the conclusions drawn from research. I suspect that as long as the conclusion that gay and lesbian families present no risks for children’s well being is viewed as settled science, the likelihood that funding agencies will commit the significant resources that will be needed to develop large, complex and representative data will be low.

The second criticism offered by Marks is the lack of adequate comparison groups. There are two components to this that need to be addressed: First: to whom should children of gay or lesbian parents be compared? A strong case can be made for using families anchored by continuously married heterosexual couples as the benchmark; these families, although declining as proportion among all families (except among those with at least a college degree), still represent the most common setting of childhood. In addition, as Marks notes, the general consensus is that being raised in these families presents advantages for children. Finally, using intact married couple families as the comparison family structure is typical in studies of the well-being of children in single parent families, remarried families, and cohabiting couple families. On the other hand, with the exception of children born from donor insemination or adopted at birth by lesbian couples, the majority of children of gay or lesbian parents have experienced family transitions not unlike that of children of divorced or remarried parents. The upshot is that whenever possible studies should include comparisons across multiple family structures.

The second component that poses a significant challenge for doing comparative studies is the complexity of these nontraditional families. It is now a given among researchers that entry into a family (e.g. donor insemination, via divorce of parent, by surrogacy, or by adoption) and that parent’s sexual orientation (gay or lesbian) are important dimensions, although the extant literature is unevenly distributed across these types. Even this typology may be simplistic. In their review of the literature on gay and lesbian families, for example, Biblarz and Savci (2010) also argue for more research on bisexual and transgender parents, as well as the nature and effects of gender fluidity.

Compounding the complexity of these diverse family arrangements is the dynamic nature of family life. The presence or absence of effects, as well as their magnitude, are likely to be contingent on the timing of exposure to gay or lesbian parenting in a child’s life (starting at birth, early or mid childhood, or adolescence), the proportion of childhood spent with gay or lesbian parents, and the stability of parent’s involvement with the child. Put another way, a reasonable grasp of the demographic and dynamic contours of children’s family lives is likely to be crucial for understanding the possible link between gay or lesbian parenting and children’s well-being.

Finally, Marks argues that few studies, at least by 2005, have examined outcomes that extend beyond childhood such intergenerational poverty, educational attainment, delinquent or criminal behavior or suicide; outcomes that are central concerns of studies of the consequences for children of other nontraditional families. Clearly longitudinal data, the gold standard for addressing fundamental questions of causality, mechanisms, the relative importance of preexisting conditions, and the possibility of emerging effects would be ideal. The US National Longitudinal Lesbian Family Study of 78 children born via donor insemination to lesbian couples between 1986 and 1992 is one of the few examples of a longitudinal survey directly concerned with the well-being of children (van Gelderen, Bos, Gartell, Hermanns, & Perrin, 2012). Exploiting existing longitudinal surveys is unlikely to be a viable option. Such ongoing nationally representative longitudinal datasets such as the National Longitudinal Surveys or the National Survey of Adolescent Health were not constructed to precisely identify childhood or adolescent experiences with gay or lesbian parents. Neither were these families targeted for oversampling, making it likely that there are too few to cases to get beyond very limited analyses. Given the costly and time-consuming nature of longitudinal data designs, addressing questions of social and behavior outcomes in young adulthood and beyond may rest for the foreseeable future on collecting retrospective data.

The Regnerus paper attempts to address some of the criticisms of this literature leveled by Marks. The heart and soul of this paper is the introduction of a new data set, the New Family Structures Study (NFSS). These data depart from the types of data common to the vast majority of previous studies by its focus on young adults, the potential for comparisons with wide array of family groups, the detailed calendar data of living arrangements from birth to the interview, and the inclusion of a generous compliment of outcome measures. The most significant difference from previous data is that the sample of young adults who report a gay or lesbian parent is, with the exception of the sample drawn from census data used by Rosenfeld, larger and closer to being nationally representative than any other data set used in previous studies.

Nevertheless these data are far from ideal. A larger sample does not translate into a sufficiently large sample. Given some of the complexity of the family lives detailed above, a sample of 163 young adults who report a lesbian mother and 73 who report a gay father are frustratingly inadequate for doing anything but broad comparisons across family characteristics.

In addition there is little indication in this paper what other variables are contained in the dataset. The models estimated in Tables 2–4 use mostly standard demographic controls, with the exception of the bullying measure and the gay friendliness of the state of residence; a defensible strategy for descriptive paper. However the potential of these data for doing the essential work that advances beyond description and enables researchers to examine proximal factors such as parenting style, division of labor, or social support, is not described. Of course, this concern may certainly be allayed once the complete variable list is made public.

Finding a significant number of negative correlates of well-being for children with gay or lesbian parents, even if they are derived from simple models, invites thinking about some possible mechanisms. It is hard to imagine explanations that point to the quality of parenting per se. Parents, regardless of sexual orientation, are equally motivated to provide the best care possible for their children. It is reasonable, however, to posit that gay and lesbian parents and their children face challenges that may make parenting more difficult.

For example, anti-gay sentiments, expressed or subtle, experienced by children of gay or lesbian parents from peers, teachers, or neighbors may be difficult for parents to shield from their children. Parents themselves can also be confronted with the anti-gay sentiments of neighbors, parents of their child’s peers, or other adults such as teachers or pastors. Some studies find that these encounters are not a significant predictor of children’s well-being (Patterson, 2006; van Gelderen, et. al, 2012). However it is possible that these anti-gay sentiments, and their impact, are quite variable depending on the particular setting. Some schools intentionally foster a supportive and tolerant culture, while others may be less successful at this. Some neighbors are more accepting than others. Some communities and their child-oriented organizations make a strong effort to be inclusive; others may not see this as necessary or may even be hostile to non-traditional families. More studies that take into account these variable contexts are needed to gauge the impact they may have on gay and lesbian parents and their children as well as how parents cope with these issues.

Building and maintaining social support networks may be more difficult for gay and lesbian parents. There is considerable research which shows that the quality of intergenerational family relationships, friendships with other parents, and ties to community groups that encourage good parenting such as parent-teacher organizations, boy and girl scouts, or churches, matter for both quality parenting and for children’s well-being. Support networks may be easier for gay and lesbian parents to come by in urban areas or university dominated towns that tend to be tolerant of diverse families. The same may not be true for small towns characterized by conservative or religious values and traditional family structures. The lack of social connectedness may also retard the development of community social capital. Community social capital refers to the linkages to the larger world that parents provide for their children such as being advocates for their children in school or other settings, plugging children in the kinds of social or work-related networks that benefit children, or teaching children how to negotiate the adult world (Pleck, 2007). Of course these ideas are only suggestive and it may not be possible to carefully examine these explanations or other proposed mechanisms with these data.

In summary, the Marks paper, by turning a bright light on the shortcomings of previous work, challenges researchers to develop better data and conduct the kinds of analyses that allow more confidence in generalizations. The Regnerus paper introduces a data set based on a national probability sample that has the potential to address some of Mark’s criticisms. The analyses in the Regnerus paper are provocative but far from conclusive. These very preliminary findings should not detract from the real importance of this paper, the description of a new data set that offers significant advantages. Whether the New Family Structures Study has the possibility of unsettling previously settled questions depends in equal parts on richness of the information collected, as well as the willingness of scholars to make use these data.