The claim that children raised by lesbian and gay parents thrive, on average, just as well as those raised by heterosexual parents has become a commonplace of opinion journalism, especially since the American Psychological Association reported that conclusion in 2005. However, two studies published online this week in the July issue of the peer-reviewed journal, Social Science Research, put a large question mark over that view of the subject.
In one article Mark Regnerus, an associate professor of Sociology at the University of Texas at Austin, reveals the results of large-scale, robust study showing poorer adult outcomes for children whose parents had same-sex relationships, compared with those raised by their own married mother and father.* The other article, by Loren Marks, an associate professor of Family, Child, and Consumer Sciences at Louisiana State University, reports on a review study that finds no firm basis for the APA's view that gay parenting is equivalent to heterosexual parenting.
"How different are the adult children of parents who have same-sex relationships?" asks the title of Dr Regnerus' article. His answer -- decidedly different -- is based on a "first cut" at data from the New Family Structures Study (NFSS), a new data collection project funded at this point by two conservative organizations, the Bradley Foundation and the Witherspoon Institute.
The study screened over 15,000 current or former members associated with Knowledge Networks, a research firm that obtains samples representative of the US population. Adults between 18 and 39 years of age were asked if their parents had ever had a same-sex romantic relationship, which yielded 163 mothers and 73 fathers in addition to 2,752 parents with other family structures. Regnerus reported results for 40 different outcomes, among them whether the subject ever had suicidal thoughts, identifies entirely as heterosexual, ever had an STI, their closeness to parents, attachment, condition of current relationship, frequency of use of various drugs, and numbers of various sexual partners.
The answers revealed a number of significant differences, even with statistical controls for respondent's age, gender, race/ethnicity, level of mother's education, perceived household income while growing up, experience of bullying as a youth, and state legislative gay-friendliness. Comparisons of the children of intact heterosexual couples to other family forms were generally favorable to those of the former group of parents.
As Regnerus himself writes in Slate, children of women who had had same-sex relationships "were more apt to report being unemployed, less healthy, more depressed, more likely to have cheated on a spouse or partner, smoke more pot, had trouble with the law, report more male and female sex partners, more sexual victimization, and were more likely to reflect negatively on their childhood family life, among other things." Household instability was a major theme among the adult children of same-sex parents, the children of women in lesbian relationships reporting a much higher incidence of time spent in foster care, grandparent care and living on their own before age 18 than those in the rest of the sample.
Box Turtle Bulletin and other internet sites have been ablaze this week with arguments about the strengths and weaknesses of the NFSS, not to mention numerous ad hominem attacks against Regnerus based on the assumption that his work is part of a campaign against gays. In fact, its purpose was to gather objective data that no previous study on this subject had tried or managed to get -- from large samples of gay or lesbian parents from truly random samples of the general population. Previous research, as noted by Dr Marks in his article, has too often relied upon convenience (non-random) samples to study the children of gay or lesbian parents.
Another strength of his study is that Regnerus attempted to look at possible long-term outcomes of different family structures, by surveying adult children. The downside of this important concept is that it forces the reported parenting process back into the 1990s when society was probably less accepting of non-traditional family structures. However, it also might remind us that when we debate controversial topics such as gay marriage, we may not know for decades what the consequences will be of policy changes. For example, no fault divorce was intended to help highly conflicted couples avoid contentious divorce proceedings, but it may have also helped increase the overall heterosexual divorce rate and rates of poverty for women and children -- clearly unintended, often adverse social consequences.
The NFSS study also featured more statistical controls than many previous studies and included a large number of multiple-item scales (many sociological studies rely upon single item measures). One important omission from the tables in the article was the standard deviations, without which effect sizes cannot be calculated. However, Dr. Regnerus indicates that the full data set will be available for re-analysis by other scholars, regardless of their political interests.
Whatever the limitations of Regnerus's research, it appears that most of those reviewing his study commend him for the random sample aspect of the study. More controversy emerges with respect to the nature of the lesbian or gay families aggregated into his same-sex parenting group(s). It appears that very few of the same-sex parental relationships identified by the young adults were stable relationships. For example, only five lesbian couple parents, among the 173 families where a mother had been involved in a same-sex relationship at some point, had been together for 13 or more years of the first 18 years of the participant's life. Given the random nature of the sample, that paucity of long-term lesbian parental relationships suggests that such relationships are relatively rare, at least in terms of percentages. However, it is possible that many such relationships involve higher-income families who may be able to insulate themselves from "opportunities" to participate in such survey research. It also remains possible that some other flaw in the family selection process inadvertently overlooked a disproportionate number of such families.
Several critics have argued that he was actually comparing stable heterosexual parents to MOMs � mixed orientation marriages or to other structures, such as a heterosexual couple in which one partner had a same-sex affair. Because he asked each participant detailed questions about household composition, year by year, from age 1 to age 18, he has a goldmine of data, with a lot of room to deal with the concerns of his critics, if he or others wish. For example, I think his dataset could be used to look into differences among parents who had a same-sex relationship in the past but are now married straight; same-sex parents who raised a child from birth; or parents who came out later in life and now are in a same-sex household.
Furthermore, even if one accepts the MOMs thesis, it is still of interest that many different family structures did appear to predict a variety of outcomes. As noted, the effect sizes of those outcomes were not elucidated in the article but sociology professor Paul Amato, in a comment on the NFSS in the same issue of Social Science Research, suggested they were small to moderate in nature, depending on the group comparisons involved. Sociologists tend to study structures while psychologists tend to study processes. Thus, these two groups of scholars tend to debate the relative importance of structure vs. process. Dr. Regnerus' model academically looks, roughly, like this: structure predicts process, process predicts (in this case, child) outcomes. Probably few would dispute the importance of process (e.g., parents having a loving relationship with each other and their children). Scientifically, the challenge comes when you try to determine which factors mediate the structure-to-process relationship.
Regnerus wasn't trying to do that in this initial report, but I think his data have the potential to permit far more detailed analyses that would help scholars get at some of those issues - that is, far more complex models of how family structure impacts process. The NFSS contains a number of measures that might be useful mediating variables between family structures, family processes, and child outcomes.
One thing is certain: this study represents a serious attempt to obtain objective information that has seldom been available before, and it should not be dismissed simply because of its uncomfortable suggestion that a variety of nontraditional parental structural characteristics may not bode well, on average, for long-term child outcomes.