… and found to have much in common with less controversial gay parenting studies.
In the July issue of the scholarly journal Social Science Research, Professor Mark Regnerus (pictured) published an article detailing initial results from his New Family Structures Study. His results suggested that adult children who had been raised, for at least a brief time, in families with a gay, lesbian, or bisexual parent were more likely to report dysfunctional adult outcomes than those who had been raised in other family structures, especially families with continuously married heterosexual parents.
In the same issue of the journal, three other scholars rendered comments on the NFSS results and Regnerus addressed their comments as well. His study raised a huge outcry of protest, which led, among other things, to the University of Texas conducting a preliminary investigation into the ethics of his study (he was cleared of any malfeasance). Subsequently, in the November issue of SSR, several scholars, including the editor of SSR and an auditor of the review process, rendered their verdicts on the study. Professor Regnerus also provided a revised analysis of the data, attempting to address some of the criticisms of his study.
How different was Regnerus’s method to that of other studies?
I also weighed in on the discussion with a commentary pointing out, as SSR editor Dr. James D. Wright noted, “that many of the most controversial methodological and measurement decisions made in the Regnerus paper have well-established precedents in the larger social science literature”.
My approach was different than those of the other commentators, who were generally “for” or “against”. My main question was, “How different was what Regnerus did methodologically compared to what other scholars have done in the past ten years when investigating similar issues?” I considered sample selection, sample size, definition of family forms, measurement of sexual orientation, statistical analysis, funding, and consistency of results with other research, citing over 110 examples of other previous research.
Space here does not permit me to detail all of my findings but I will discuss a few of my comments. While some have tried to vindicate the Regnerus study by citing my comments, my intent was not to laud the study but to place it in context relative to other social science research. The conclusion to be drawn may be “similar methodological limitations” as much as anything.
Criticisms would apply to others
Use of Knowledge Networks data. Regnerus was criticized, for example, for using Knowledge Networks (KN) for obtaining his sample. KN does not appear to conduct random national samples per se but provides a hired panel of respondents thought to compare favorably to the characteristics of a national random USA sample. For myself, I am concerned about turning over one’s data collection to another organization because you may not understand all of the small decisions (or errors) made in that complex process.
However, my search of the academic literature led to over 20 published studies that had used KN, including research by Professor Gregory Herek, an internationally renowned gay psychologist, as well as research sponsored by the Human Rights Campaign, a pro-gay advocacy group.
Thus, while I might be “cool” towards the use of KN, it is clear that the research company has been used by a wide variety of scholars, who have had success getting published in leading journals with data drawn from its panels of respondents. Regnerus reported that his response rate was approximately 65%; in contrast, Herek’s research with KN featured a response rate of only 30%, apparently far lower.
Use of mixed orientation households (MOMs). Much has been made of the possibility that Regnerus succeeded in gathering data from children from mixed orientation marriages (MOMs). However, many other attempts to study GLB families have involved such marriages. One study, for example, featured 72% of children who had been born into a previous heterosexual marriage before joining a lesbian couple family at an average age of over 4 years. However, the results of such studies are heralded as showing us how well lesbian families are doing, even though they involve many of the same limitations vis-à-vis MOMs as Regnerus’ NFSS study.
Funding issues. With respect to funding, many published studies have been funded by pro-gay advocacy groups and yet few report doubts about the influence of such funding on research outcomes; but since the NFSS was funded by conservative groups, such doubts are brought to the forefront.
Outcomes for children. There is considerable research – detailed in my commentary in SSR – that notes the instability of lesbian and gay parental relationships, the tendency of their children to be involved in substance abuse, and the tendency of such children to experiment with or adopt same-sex sexual behaviors or identities -- results similar to those that Regnerus reported. In other words, at least some of Regnerus’s findings were very similar to results from many other studies from around the world.
Surely some will attempt to portray my comments as a whitewash, as if I agree with all of Regnerus’s methodological decisions. That is hardly the case.
First, I think that it would have been wise, especially once it became apparent that there were few very stable GLB families in the Knowledge Networks panel, to contract with a pro-gay research organization to collect data from at least 30, preferably 100 or more, stable same-sex families, to permit a more valid (although still limited because of possible selection effects) comparison with similarly stable heterosexual families.
Second, I think Regnerus should have reported a type of bracketing of his results – worst case, best case, and modal case – in his first article. However, in his first article, he presented a worst case analysis. In his second paper, he presented more of a modal case analysis. As far as I can tell, we have yet to see a detailed best case analysis.
Third, it is not clear to me that using average comparisons of outcomes for different family structural forms is the best approach for assessing the intertwined role of family structure and family process in the development of children. I hope to try some of my different ideas over the next few months to see what some of those better ways might be, if permitted by the characteristics of the data in the NFSS.
As my own review of other studies shows, however, the wholesale ad hominem attacks on Professor Regnerus and the complete dismissal of all of his methods and professionalism continue to be unjustified, even though his research – as all research - deserves careful scrutiny. His decisions about research design and analysis were within the ball park of what other credible and distinguished researchers have been doing within the past decade.
Walter Schumm is a Professor of Family Studies at Kansas State University, with over 300 publications, including approximately 250 journal articles. He is also a retired colonel, U.S. Army. Dr. Schumm served as a paid consultant (6-7 days) in the early stages of the development of the NFSS. Dr. Schumm presented each of the articles or comments related to the NFSS from Social Science Research in his assessment of the NFSS at a round table discussion at the National Council on Family Relation’s annual conference, November 2, Phoenix, Arizona.
* “Methodological decisions and the evaluation of possible effects of different family structures on children: the New Family Structures Survey (NFSS)”, Social Science Research, Vol. 41(6) pages 1357-1366. (My apologies for the confusion between “Survey” and “Study” in the NFSS title).
* Links to other articles on the NFSS study in the current issue of Social Science Research can be found here. They are for purchase only.