According to conventional wisdom, research regarding outcomes for children of parents in same-sex relationships shows “no difference.” For example, the American Psychological Association (APA) stated in 2005 that “not a single study has found children of lesbian or gay parents to be disadvantaged in any significant respect relative to children of heterosexual parents.” This conclusion has been cited in the judicial proceedings on the nature of marriage. For example, Judge Vaughn Walker’s decision regarding Proposition 8, California’s constitutional amendment defining marriage, stated: “The research supporting this conclusion is accepted beyond serious debate in the field of developmental psychology.”
However, a number of researchers have pointed out significant methodological problems with the research that would cast doubt on the conclusiveness of the “no difference” findings. Even some researchers supportive of same-sex parenting have acknowledged the significant methodological limitations in the research to date. Examples of the methodological problems are discussed below, as well as a profile of the recently released New Family Structures Study, which has raised the bar on quality of research in this area. Given the magnitude of the policy issues under debate, much more study is needed.
In order to generalize a set of social scientific evidence to a broader group, the samples upon which the evidence is based should accurately represent the broader group. In a probability sample, every member of the broader group has an equal chance of being selected into the sample, and the selection is random. Non-probability samples, on the other hand, cannot make valid generalizations about the broader group because they do not represent it. These samples “may give us interesting leads, and suggest possible insights, but nothing reliable can be inferred from them outside the individuals studied.”
A 2010 article reviewed 44 same-sex parenting studies and noted that only five had probability samples; moreover, only three used a nationally representative sample. However, this same sample used by all three studies contained only 44 children of same-sex parents.
The remaining studies were all based on convenience samples, which are comprised of volunteers recruited through targeted advertisement (e.g., in specific publications), the “snowball” method (when respondents identify additional subjects), and/or an existing group.
Consequently, most of the no-differences findings “describe samples of lesbian families that are disproportionately middle class, White, and highly educated.” For example, a 2012 review of the 59 studies reported in the 2005 APA brief on same-sex parenting “reveals a tendency towards not only non-representative but racially homogeneous samples.” Even studies of donor-inseminated (DI) parenthood “do not know the extent to which the comparatively high socioeconomic status of DI parents studied accurately reflects the demographics of lesbian and gay parenthood generally.”
Moreover, in many studies, even the opposite-sex comparison groups are not representative of their respective populations. Researchers also acknowledge the nearly non-existent research on gay fathers and parents who are bisexual or transgender.
Failure to Reflect Diversity
Emerging demographic data suggest significant diversity among same-sex parents. For example, using the Census Bureau’s 2009 American Community Survey, a 2012 report found that, contrary to popular perception, at the national level, same-sex couples raising children are the most likely to live in southern states, not on the west coast or in the northeast. Moreover, minority same-sex parents and less-educated same-sex parents are more likely to be raising children than are white and more-educated parents.
A 2010 review reported that the “mean number of children of gay or lesbian parents in these studies is 39, and the median is 37.” The two studies that used longitudinal, probability samples identified only 44 children of same-sex parents in a sample of more than 12,000 adolescents and 18 lesbian mothers out of 14,000 mothers in the study, respectively. One was supplemented with 21 mothers through the “snowball,” or referral, strategy.
Small sample sizes increase the likelihood of finding false negatives—that is, concluding that there are no differences when they do in fact exist. Of the 22 studies (out of the 49 reviewed) that had a different-sex comparison group, a 2001 review found only one study that had a sufficiently large sample size, which had 25 percent probability of finding a false negative.
For the remaining 21 studies, the false negative rate ranged from 77 percent to 92 percent. Researchers have also acknowledged that “some of the findings of no differences may miss real differences…because some studies use levels of significance that may be too restrictive for their very small samples.”
Inconsistent or Non-Existent Comparison Groups
To conclude that two groups are different, research needs to compare a study group (e.g., children of same-sex parents) with a comparison group (e.g., children of opposite-sex parents). Ideally, these two groups should be identical except for the characteristic of interest that is being tested—in the case of same-sex parenting, the parents’ sexual orientation or relationship status. However, a surprising number of same-sex parenting studies have no opposite-sex comparison groups.
For example, a 2012 review that evaluated the 59 studies in the 2005 APA report noted that only 33 had comparison groups, of which 13 clearly used single-mothers. In the remaining 20 studies, however, the comparison groups were often vaguely defined, with general references to “mothers” or “couples,” and only in rare cases was the comparison group explicitly defined as opposite-sex intact married families.
Moreover, some studies fail to account for all the significant differences between the study and the comparison groups—e.g., mothers’ age or education level—that might bias the no-differences findings.
Other Methodological Problems
Researchers have noted several other methodological limitations in the research on children of parents in same-sex relationships. They include a variety of measurement issues, such as questionable reliability and validity and potentially biased participant responses (e.g., given by parents) due to social desirability considerations.
In addition, in many studies, participants, and researchers were not blind to the nature of the study, which may have introduced biases during the data collection and processing stages. Furthermore, few studies examined longer-term outcomes, as some effects may not be observable until late adolescence or early adulthood.
New Family Structures Study
In contrast to most studies on children of same-sex parents, a new national study called the New Family Structures Survey (NFSS) offers the most representative picture to date of outcomes for children whose parents had a same-sex relationship. It compares a large, national, random sample of such children with their peers from intact families on 40 outcomes and finds significant disadvantages across many of the outcome areas for children whose parents had a same-sex relationship. The NFSS is a rich data source for quality research on children raised by parents who had same-sex relationships. While the NFSS does not answer every research question, it represents an important contribution to the literature on the issue.
More Study Needed
Better data and theories, greater detail, and more rigorous methods over an extended period of time are needed before a broader understanding can emerge. Today, the issue of children’s welfare in association with same-sex parenting calls for more rigorous research based on large, nationally representative samples. The NFSS is an important contribution in this regard. The need is all the more serious given the magnitude of the policy questions at hand related to the institution of marriage.
Christine C. Kim is Policy Analyst in the Domestic Policy Studies Department at The Heritage Foundation.