In the July 2012 issue of Social Science Research (SSR), Associate Professor of Sociology Mark Regnerus of the University of Texas-Austin (UT-A) published a research article with the leaden title: “How Different Are the Adult Children of Parents Who Have Same-Sex Relationships? Findings from the New Family Structures Study.” His conclusion? That among a large, random sample of young adults ages 18 through 39, there were “numerous, consistent differences especially between the children of women who have had a lesbian relationship and those with still-married (heterosexual) biological parents.” Predictably, all hell broke loose.
“The most reasonable conclusion is that there is some negative effect of gay/lesbian parenting on children’s outcomes in these data.”
This study should have been welcomed by social scientists in 2012. Why would they not want legal decisions that could profoundly affect the institution of marriage and the fate of children to be informed by large-scale controlled studies? After all, the volatile issue of constitutionality of state prohibitions of same-sex marriage, integrally tied to assertions about gay parenting outcomes, was swiftly working its way through federal appeals. SCOTUS would soon be settling the issue. Most published social science research does not matter much in the real world. But this research clearly mattered.
Moreover, Regnerus’ study offered what little same sex parenting research had. For one thing, using controls, it examined an impressive 40 highly relevant outcomes. It also addressed them in young adults, not children or teenagers. As Richard Redding pointed out, this was not typical for such research.
Regnerus also utilized a large, random sample over against the common practice in such research of using small, non-random, even “convenience” and “snowball” samples unsuitable for making statistical claims about significant differences in general populations.
For example, one source of research showing “no difference,” was the National Longitudinal Lesbian Family Study (NLLFS). As Glenn Stanton documented, the NLLFS (which is ongoing) only had a tiny number (“initially 84 lesbian families”) invited through “informal networking and word of mouth referrals” and “announcements at lesbian events, in women’s bookstores, and at lesbian newspapers,” in three large cities. It was not representative of gays, or even of lesbians. High status, highly educated, overwhelmingly white participants knew what they were signing up for and could connect favorably with achieving gay rights goals.
As Redding noted, many other studies had samples of just “15 to 55 gay or lesbian participants.” And the Regnerus’ study? Redding pointed out that it had “a sample of 236 adult children of parents (175 mothers, 73 fathers) who had a same-sex relationship and a total sample across all comparison groups of 2,988.”
Of course, Regnerus did not compare children being raised in stable same-sex headed (much less, married) homes with those in married opposite-sex households. But he did not claim to, stating right up front that he compared “how the young-adult children of a parent who has had a same-sex romantic relationship fare … when compared with six other family-of-origin types” (emphasis added).
But then, how could one perform such comparisons, using large, random samples, at that time? According to research cited by Regnerus, in 2010 there were an estimated 98,600 same-sex (out of more than 35 million total) households with children. In a general sample of Americans, how could you pick up enough people raised by same-sex couples to analyze and compare them with otherwise comparable opposite-sex parents? How many of these same-sex households had children old enough to test many important, measurable outcomes? How many young adults were raised in such households?
The numbers reality that Regnerus confronted is still faced by any researcher trying to do controlled gay parenting research. As of the 2023 Census’ American Community Survey (ACS), among married couples raising their own children, only one-half of one percent — 221,076 — were same-sex. In Massachusetts, which has had same-sex marriage since 2004, that percentage was just eight-tenths of one percent. UCLA’s Williams Institute used the 2019-21 ACS to estimate that 293,986 children were being raised by same-sex couples (married and cohabiting). In those years, the estimated total number of children fluctuated between 74.3 to 73.1 million.
So, even with its limitations (which he mostly stated up front), the Regnerus study was ground-breaking. Yet rather than reasonably absorbing and discussing it, many social scientists and numerous media pundits unleashed the furies on Regnerus’ hapless head. If he could not be silenced, he had to be punished, his reputation reduced to a smoking ruin as a warning to any social scientists with the temerity to offend the LGBT gods.
The vilification was almost instant.
As the National Association of Scholars (NAS) reported, on June 24, a blogger and gay activist named Scott Rose (really, Rosensweig) demanded that UT-A investigate Regnerus’ “scientific misconduct.” About 200 “PhDs and MDs” wrote to the SSR editor, demanding an inquiry into “the process by which this paper was submitted, reviewed, and accepted.” Rose also published that on his blog.
Ultimately, UT-A found no grounds to formally investigate Regnerus. But they still put him into a gauntlet through the end of August. The smears intensified. This included not just criticism of methodology but, as the NAS said, “character assassination.”
Many of Rose’s charges would have probably been constitutional violations if the school pursued them (they did not). As the Foundation for Individual Rights and Expression (FIRE) reported, amidst a few allegations that could legitimately be investigated, the charges included such things as: Regnerus is an ex-evangelical and current Catholic; he received support from the conservative Witherspoon Institute “where Robert George of the anti-gay National Organization for Marriage is a Senior Fellow”; “Sociologists from Brigham Young University were involved in the study design … akin to asking the Ku Klux Klan to design a study about Jews.”
UT-A “sequestered” (i.e., took away) Regnerus’ six computers and 42,000 emails. As FIRE complained, this was “a major disruption of his scholarly activity.” Not to mention unbelievably “intrusive.”
Things did not stop there. An audit of the journal’s peer-review process of Regnerus’ submission by Darren Sherkat was to be published in the November issue of SSR. In July, the SSR senior editor shared a draft version of this audit with the Chronicle of Higher Education, which covered it, interviewing Sherkat and the editor. In the interview, Sherkat called Regnerus’ article “bullshit” that should have been “disqualified … immediately.”
Sherkat accused reviewers of “both ideology and inattention.” He complained that three of the six were opposed to gay marriage (though this percentage reflected the American public at the time and could be viewed as ensuring fairness). As for the editor, he “suffered sleepless nights since the publication of Regnerus’s paper, and has received a steady stream of angry e-mails, from both colleagues and irate strangers.” If this is what he was being put through, we can imagine Regnerus’ email.
Meanwhile, Regnerus responded to and interacted with his methodological critics, and did some re-analysis in a follow-up SSR piece in November, but without backing down. And as retired Catholic University sociologist Paul Sullins recently recounted, the same month “he publicly posted his entire dataset, inviting [his critics] to analyze it for themselves and to try to overturn his findings,” though he was not required to.
None of this stopped the media and professional mud-slinging too numerous and vitriolic to fully recount here. For example, the Huffington Post, which just the year before had positively publicized a study (of 78 kids) based upon the NLLFS that claimed a zero percent child physical and sexual abuse rate in lesbian households, posted an anti-Regnerus diatribe that August.
Based on the Sherkat draft, and while the UT-A inquiry was still ongoing, the author demanded apologies from a list of people who had favorably cited or defended Regnerus’ work. Then in March and April 2013 Huffington unleashed two investigative pieces supposedly further uncovering the truth about Regnerus and the study’s backers. This was followed by a 2014 post by a graduate student in Regnerus’ department expressing angst about things such as that he fraternized with the wrong political and religious sorts, spoke at conservative places, said gay marriage would lead to more heterosexuals having anal sex (while not appreciating how great that could be), used media talking points, and called same sex relationships less stable without understanding all the of the reasons that might be true.
In October, the New York Times published a piece by Mark Oppenheimer that, while indicating the intensity of the attacks on Regnerus, noted that Regnerus “was immediately called a fraud, a charlatan, a shill for the religious right” as soon as the study was published. But Oppenheimer goes on to claim that Regnerus “won’t talk about his research” simply because Regnerus had declined to be interviewed for the Times article, despite Regnerus’ very public defenses of his research and his transparency.
Focusing the article on Regnerus’ faith and its implications for his scholarship, Oppenheimer said, “Dr. Regnerus was a proud Christian witness, once upon a time” but “these days, he won’t discuss his faith.” What was the evidence of that? In one recent interview with Christianity Today, Regnerus had “said nothing about his religious beliefs.”
The professional coup de grace were two replication studies using Regnerus’ data. The first, by sociologists Simon Cheng (University of Connecticut) and Brian Powell (Indiana University), questioned the classifications Regnerus had used, cutting out cases they believed he miscategorized. When they “fixed” this, as Sullins pointed out, “all of the important differences … became statistically insignificant.” Of course, reducing sample size makes it much harder to establish statistical significance.
The second, by Stanford sociologist Michael Rosenfeld, “corrected” Regnerus’ research by controlling for family instability. Voila, most of Regnerus’ statistically significant outcomes were reduced to insignificance. As Rosenfeld put it, “simple control for family transitions accounts for most of the negative outcomes for subjects who had ‘gay fathers’ or ‘lesbian mothers.’” Of course, if less stability is more common among same sex parents, as Regnerus argued, that is a clear indication that such parenting hurts kids.
As Sullins tells us, Regnerus “responded to both critics” but, again, did not back down, leaving things in a “standoff.” Nevertheless, most social scientists firmly sided with his critics, with some attacking any who cited it favorably or had the temerity to try to publish their own research identifying negative outcomes of gay parenting.
The alleged non-difference of same-sex versus opposite-sex parenting outcome is now widely treated as settled science. For example, in an otherwise excellent, rigorously empirical 2023 book, The Two Parent Privilege, University of Maryland economist Melissa Kearney dismissed the idea of any such differences out-of-hand. She did so by citing American Sociological Association’s (ASA) amicus brief in the Obergefell case (the ASA had strongly sided with Regnerus’ critics, and that brief refuted him at length), plus adding three other studies in a footnote, two of them from well before 2012. Said Kearney, “there is no empirical evidence showing that children raised by married same-sex parents have different outcomes than children raised by similarly situated married opposite sex parents … once socioeconomic status and family stability are accounted for.” Despite the fact that, as we know, tiny numbers and the recency of widespread gay marriage makes such an absolute, categorical assertion hard to defend.
That is where things are. We must say that there are no differences, except perhaps that gay couples produce better outcomes than heterosexual couples. Having two parents matters, but having both a mother and a father does not.
Breaking the False Narrative
Now this comfortable but rigid narrative has a fly in its ointment.
Just a few months ago, a groundbreaking reanalysis of Regnerus’ 2012 study, and of the replications by Cheng and Powell and Rosenfeld that supposedly drove a stake in its heart, appeared. It convincingly vindicates Regnerus and refutes these replications.
This was published in the most unlikely of places — a chapter inside a research methodology tome explaining and demonstrating how to do “multiverse analysis.” Released in March and authored by Cornell sociologists Cristobal Young and Erin Cumberworth, the book is Multiverse Analysis: Computational Methods for Robust Results. The Catholic News Agency has posted the chapter.
Multiverse analysis is explained in laymen’s terms by Martin Schweinsberg, a business professor and psychologist with the European School of Management and Technology in Berlin. I draw on that here.
Ahead of actually analyzing any set of data, researchers make choices among a set of potentially valid alternatives they are aware of. Multiverse analysis, Schweinberg says, takes “each possible analytical option and crosses it with all other possible analytical options, resulting in a universe of possible analyses … you now run all possible combinations of these choices so you can explore the universe of possible analyses.”
Young and Cumberworth chose to try this on Regnerus’ study, and the two replications discussed above, because various factors made it, they said, “ideal for this analysis.” It took “more time and effort” than they expected, which is an understatement considering they identified over “2.6 million model specifications.” They undertook the analysis expecting to vindicate Regnerus’ critics, believing that they would “drive their point home in a powerfully conclusive way.” But, as Gomer Pyle would say, “Surprise, surprise, surprise.”
To be sure, almost all the models reduced the strength of Regnerus’ estimates. But nullify them? No. For example, this is what they said about their “full multiverse results”: “critics were only partly successful in challenging [Regnerus’] results … very few of the 2.65 million estimates are zero or opposite-signed (i.e., opposite Regnerus’).” Their point? “The most reasonable conclusion is that there is some negative effect of gay/lesbian parenting on children’s outcomes in these data, but it is probably smaller than suggested by Regnerus’s original study.” They said, “we were surprised by the robustness of the Regnerus finding,” and rejected Cheng and Powell’s and Rosenfeld’s main conclusions.
Does this mean that their analysis supported the claim that people raised by same-sex couples fare worse than those raised by opposite-sex couples in apples-to-apples comparisons? No. They do not seem to think so, and Regnerus was not testing that specifically. As indicated, doing so will require a lot more analysis and, especially, contending with the still miniscule numbers of children in same-sex married families.
But it is possible that Young and Cumberworth have made it acceptable to do competent and objective research on this critically important issue again.
But it still will not be easy. How likely is it that peer reviewers in today’s social science journals will let such studies into their publications unless the conclusions fit the established narrative? Meanwhile, as Sullins averred, as in the past, weak research with the right conclusions will probably continue to be accepted, if not trumpeted, a woeful but tiresomely predictable double-standard.
Moreover, as Sullins says, these days, “adverse, malicious findings tend to be suppressed much more quietly” than the shrill denouncements and attempts at methodological refutation that Regnerus dealt with. He continues:
For progressive partisans, it has become clear that there is no necessity to refute findings that are deemed unfavorable. Instead of noisy opposition that advertises the strength of their opponents’ arguments or stimulates thinking about both sets of ideas being debated, proponents of gay parenting have deployed their dominance in the academy to quietly disappear adverse ideas. Partisan editors and peer reviewers ensure that negative findings seldom get published; if published, they are seldom cited. Increasingly, they are subject to retraction for “ethical reasons.”
The reception of Young and Cumberworth that I have seen so far supports Sullins’ depressing observations. Google searching shows publicity of their findings in conservative venues, but little if anything in the mainstream press, and certainly no mea culpas from Regnerus critics, the Huffington Post, the New York Times, the Human Rights Campaign, the Southern Poverty Law Center, the ASA, and so on.
But there are stirrings that suggest the American public, tired of Woke thought control, bullying, and intimidation may be open to such research. There are also thriving conservative colleges and universities, perhaps some willing to support faculty who conduct it, or scholars creating new journals ready to buck Left orthodoxy on such a sensitive issue. And there are now numerous other pathways for disseminating such studies.
Let us hope this research happens. We owe it to our children to ask such hard questions and try to honestly answer them.
READ MORE from David Ayers:
Paganism and LGBT: Kissing Cousins?