Revenge of the Sociologists
The perils of politically incorrect academic research
Jul 30, 2012, Vol. 17, No. 43 • By ANDREW FERGUSON
On the phone, Mark Regnerus sounds a little shellshocked. Professional sociologists hardly ever sound shellshocked.
“I knew it would be controversial,” he says. “But this is worse than I ever could have imagined.”
It refers to a scholarly paper Regnerus published last month. This is the hell that broke loose as a result.
Regnerus is an associate professor of sociology at the University of Texas. His paper, appearing in the journal Social Science Research, carried the title “How different are the adult children of parents who have same-sex relationships? Findings from the New Family Structures Study,” which is not a lapel-grabber. The subject was “gay parenting.” (One of the unpleasant aspects of social science is that you fall into the habit of using “parent” as a verb and silly participles like “parenting” instead of “child-rearing.”) For more than a decade now the unchallenged view among social scientists has been that there is no difference between children brought up—I mean parented—by lesbian and gay couples and those brought up in households where Ma and Pa are married. The “no difference” view has been certified by judges in gay-marriage decisions and repeated as received wisdom in the popular press, where you’re likely to see references to
In fact, this passage, from a news story not long ago in the Pittsburgh Post-Gazette, is more responsible and hedged than most on the subject. The reporter admitted that the body of research is small, and the “for better or for worse” is a nice touch. And she didn’t bother to mention the most recent stage in the evolution of the “no difference” view. Some researchers have begun saying there may be differences after all: By some measures, they maintain, children brought up by lesbian couples fare better than children from traditional families.
And now here comes Mark Regnerus. As a researcher, he says, “I’ve dabbled around.” He’s written well-received articles and books on religion, sex, and marriage. He has no particular expertise in the study of gay and lesbian relations, but the conventional view about the children of same-sex parents struck him as odd on its face. For starters, it’s widely known that children of divorce are more likely to show negative results on a number of “outcomes” in later life, such as depression, drug use, alcoholism, and so on. It’s also known that a high percentage of children being brought up by lesbians and gays are simultaneously the children of divorce. Logic suggests that some negative outcomes would show up in a study of those children too, placing them at a disadvantage next to children raised by heterosexuals in intact families.
Regnerus went through the literature—that often-cited “relatively small but conclusive body of research.” If a social scientist hopes to produce a statistical finding that applies to the larger population, probability theory requires that he study a relatively large number of people who are chosen at random. Most of the researchers in the literature on same-sex parenting used samples that were too small to produce generalizable results. All but a handful of them drew “convenience samples” from gay activist organizations or by word of mouth. These methods came up with samples of lesbian and gay parents who were whiter, richer, and more likely to be city dwellers than the national population. There were few control groups against which the results could be compared. Regnerus found the same methodological weaknesses in study after study.
Regnerus is a controversialist, though not a crude one—as a bull in the china shop of mainstream sociology, he’s more likely to pirouette than rampage. He decided to test the “no difference” consensus with a study of his own. With the approval of his university, Regnerus received funding from the deep-pocketed Witherspoon Foundation, a group dedicated to “scholarly research and teaching that enhance understanding of the crucial function that marriage and family serve in fostering a society capable of democratic self-governance.” Which is a long way of saying it’s conservative. Regnerus himself is a Catholic convert who says he’s never voted for a Republican presidential candidate.
Witherspoon executives say they sought matching funds from liberal groups to guard against charges of ideological trimming, but the foundation ended up paying most of the study’s cost on its own—nearly $700,000, a kingly sum in the world of social science research. Though several liberal sociologists declined his invitation to participate, including the well-known UCLA demographer Gary Gates, Regnerus was able to recruit a range of scholars to serve as consultants and advisers, paid and unpaid, in the design and execution of the study. He brought a team together to help plan the study over a two-day period in January 2011.
“We wanted to have a spectrum from left and right,” he said, “and we accomplished that.” Witherspoon handed over the check and then backed off. The foundation, Regnerus says, had nothing to do with the study’s design or implementation, or with his interpretation of the findings.
“I told them at the beginning that I would just report the results, whatever they were,” he says. “I really didn’t know what we would find, and I’m surprised by what we did find.”
Regnerus hired the public polling firm Knowledge Networks to interview a random sample of 15,508 Americans. From these another sample of nearly 3,000 was taken of young adults born between 1972 and 1992. Roughly 60 percent of the respondents said they spent their entire childhoods with both their biological mother and father. The rest were identified according to the type of family they grew up in: single-parent, adoptive, “blended” or stepfamily, divorced. Another category comprised those who said that one of their parents had a same-sex relationship before the respondents were 18. The group was very small—175 said their mothers had been involved in a same-sex relationship, 73 said their fathers had. Still, it was large enough, according to Regnerus and his consultants, to yield to statistical manipulation.
Only one large nationally based sample had been used before in gay parenting research. The Stanford sociologist Michael Rosenfield looked at how the children of gay parents compared with their counterparts from straight families on one outcome: whether the kids performed at an “age-appropriate” level in school. Rosenfield found no difference between the two groups. Regnerus and his colleagues were more ambitious. They checked for 40 different long-term outcomes that would carry over to adulthood. Are you happy in your current romantic relationship? Are you on government assistance, or were you as a child? Any thoughts of suicide in the past 12 months? Respondents were asked to classify their sexual orientation, whether they’d ever been in therapy, whether they’d been convicted of a crime, and to list their income, educational level, and employment history. Several questions explored whether they had been bullied in school or sexually abused as children.
One basic finding immediately leapt out—how few Americans between the ages of 19 and 39 say their father or mother had ever had a same-sex relationship: 1.7 percent. It was also clear that the nature of gay parenting has changed quite a bit from the ’70s, ’80s, and ’90s, when these young adults were children. Typically, Regnerus said, they were born from heterosexual unions that went bust; nowadays the children of homosexual couples are often “planned”—brought into a family through adoption, in vitro fertilization, or surrogate motherhood.
Regnerus found much to contradict the “no difference” view. On 25 of the 40 outcomes, young adults who said their mother had a lesbian relationship (he calls these respondents LMs) differed in a statistically significant way from young adults reared by their parents in intact biological families (IBFs). Among those whose fathers had a gay relationship (GFs), 11 outcomes were different from the IBFs.
And the differences were almost always negative for LMs and GFs. LMs in particular were far more likely to be on public assistance and to have received public assistance as children; to suffer depression; to be cohabiting; and to describe themselves as unhappy in their romantic relationships. Their income on average was lower, and so were their educational attainments. More of them were unemployed. And they were far more likely to report that they’d been abused by an adult as children. The differences between the GFs and the IBFs were smaller and less significant—there was no difference, for example, in reports of childhood sex abuse. And GFs were much more likely to have voted in the last election. In case you were wondering.
Regnerus wrote up his findings and submitted them to the editor of Social Science Research, who in turn submitted the paper to a panel of peers for approval. Three other scholars wrote critiques to appear alongside Regnerus’s paper. He also turned over the findings to the Witherspoon Foundation, which prepared a publicity campaign to unveil the new research: press interviews with Reg-nerus, op-eds by him and others, and background briefings for reporters and friendly scholars.
Then he sat back waiting for publication, expecting not much more than heck to break loose.
As of mid-July, a month after his paper was published, these are some of the things that have happened to Mark Regnerus. Three of his colleagues in the sociology department at UT joined with a fourth to -publish a widely distributed op-ed in the Huffington Post accusing him of “besmirching” the university through his “irresponsible and reckless misrepresentation of social science research.” Led by Gary Gates, the UCLA demographer who had declined Regnerus’s offer to help design the study, more than 200 “researchers and scholars” signed a letter to the editor of Social Science Research. The letter demanded that the editor “publicly disclose the reasons” why he published the paper and insisted that he hire scholars more sensitive to “LGBT parenting issues” to write a critique for the journal’s next edition. UT’s Director of Research Integrity sent Regnerus a letter informing him that a formal complaint of “scientific misconduct” had been lodged against him. The complaint, made by a gay blogger/activist/“investigative journalist” called Scott Rose, triggered an official inquiry into Regnerus’s research methods and his relationship with the Witherspoon Foundation; he’s now preparing to appear before a panel of faculty investigators. Requests have been filed with the Texas attorney general’s office demanding that Regnerus, as an employee of a state-run institution, make public all email and correspondence related to his study. And he has hired a lawyer.
A large number of his fellow social scientists—members in good standing of the guild of LGBT researchers—would like to destroy his career.
The good news for Regnerus, as far as I can tell, is threefold. For one thing, he has tenure. For another, 18 of his fellow researchers wrote a public letter in his support. And much of the major news media, including the New York Times, has written fairly, if not copiously, about his study—perhaps as a result of the careful rollout orchestrated by Witherspoon. (The rollout failed to move the explicitly ideological press, such as the New Republic, which said Regnerus “gets everything wrong,” and the New Yorker, which called his work “breathtakingly sloppy.”)
Time magazine published a brief and sober description of Regnerus’s methodology. His paper’s great strength was the large and nationally representative sample, so that groups drawn from it could be compared against one another with statistical confidence. The great weakness was that the group of stable gay couples was minuscule, making a meaningful comparison between stable heterosexual households and stable gay households impossible.
As the study entered the heated debate over gay marriage—which is, after all, the reason Witherspoon paid for it—this sampling weakness was overemphasized by gay activists, who said it discredited the entire study, and -underestimated or ignored by their counterparts in the “pro-family” world, who were often too excited to qualify the numbers accurately. But the weakness is real. Out of the original pool of 15,000 respondents, only 2 young adults reported living with their gay parents for their entire childhood.
It isn’t Regnerus’s fault that so few gay and lesbian -couples were raising children in the 1970s, ’80s, and ’90s. He tried to boost the numbers by expanding his definitions. He divided respondents into eight family types, including stepfamilies and single-parent families, along with LMs and GFs and IBFs. If a respondent said his mother had a same-sex relationship at some point in his childhood, he was counted as an LM—even if he was also the product of a divorce or raised in a single-parent family. That’s how Regnerus got the number of LMs up to 175 and the GFs up to 73.
To the extent that the dispute over Regnerus’s study is scientifically serious and not brute cultural warfare, it largely turns on whether this way of boosting the sample is legitimate. Many reputable social scientists, including the three commenters recruited by the editor of Social Science Research, say it is. His critics say Regnerus was stacking the deck. Social science is unanimous that children raised in unstable families—divorced parents, for example, or stepfamilies—have worse outcomes later in life than children from stable families. By counting these people as LMs or GFs (assuming their parents had a same-sex relationship), Regnerus ensured that those categories would show less favorable outcomes. With so many children of gays also being children of divorced or single parents it could hardly have turned out otherwise.
Again, it’s not Regnerus’s fault that gay and lesbian relationships were so unstable when today’s young adults were children. But the complication should have tempered the overenthusiastic pronouncements of his popularizers. As the conservative Family Research Council put it:
This is not only breathless but inaccurate. We may concede that Regnerus’s study could rightly be called “historic”—the data set he collected is unique and likely to yield interesting findings for years to come. But it is not a study of “children raised by homosexual parents.” Regnerus did not ask respondents to give their parents’ sexual orientation; merely whether they knew if their parents had at some point engaged in a homosexual relationship, for however long. The parents may or may not have considered themselves gay, then or now. And many of these children were not raised by a homosexual parent: There were GFs who never lived with their father at all. As a close reading of its title suggests, this is a study of adult children of parents who had same-sex relationships. And the Family Research Council’s use of the present tense is jumping the gun. The study is retrospective—a picture of the nation during the last 40 years, much of it before the gay rights movement and the widespread social acceptance of homosexuality. For all we know, and as Regnerus readily admits, the instability, and hence the bad outcomes, could be largely traced to trauma caused by the antihomosexual prejudice of an earlier time.
Regnerus’s handling of the data led to the further objection that he was comparing apples and oranges: children raised by a biological mother and father in stable families, on the one hand, and on the other, children raised in families that were by definition unstable. If Regnerus had wanted to isolate the effect of sexual orientation on child-rearing, he would have had to compare like with like: stable heterosexual families with stable homosexual families, one-parent heterosexual families with one-parent homosexual families, and so on. That he didn’t do so is taken by his pitiless critics as a sign of either incompetence or bad faith.
“Here’s the way I put it,” said Gary Gates, the demographer. “It’s like he took a group of men who never smoked and compared them with a group of women who smoked three packs a day. Then he checked lung cancer rates. And he concluded that being a woman puts you at greater risk for lung cancer. But of course the cancer rate has nothing to do with being a woman.” In the same way, he says, we can’t tell from Regnerus’s data what role homosexuality—as opposed to divorce, welfare, single-parenthood—played in the bad outcomes.
Gates is best known for his finding a few years back that only 3.8 percent of Americans are self-identified homosexuals, as opposed to the 10-plus percent routinely cited by gay activists (and created ex nihilo by the creepy zoologist Alfred Kinsey). He was heavily criticized as a traitor to the cause. “The question to me has always been why Gates . . . wants to punish us so,” said the radical activist Larry Kramer. Gates says he turned down Regnerus’s offer to help with the study because of the same scrupulousness. The design was flawed from the start, he says—and Regnerus, moreover, is not a member of the guild that studies LGBT families.
“He told me what the design was,” Gates says. “I said you’re designing this to get bad outcomes for gay and lesbian parents. I could see it. I asked him, Why are you leading this study? On the most basic level, this subject matter is about family structure and the interplay with sexual orientation. And he has no—no—background in that.”
The criticisms of Regnerus’s paper would be more impressive if they weren’t anticipated and in many cases acknowledged by the author in the same paper being criticized. Regnerus notes explicitly that the study did not identify the sexual orientation of the parents being reported on, and that some of the “gay parents” had little or no contact with their children. He admits that the categories into which he divided respondents were hardly exhaustive: “There are far more ways to delineate family structure and experiences—and changes therein—than I have undertaken here.”
He also addresses the charge of an apples-to-oranges comparison. Measuring children from divorced GFs and LMs against children from intact families, he concedes, is “arguably unfair.” Other sociologists will be free to make comparisons they deem more appropriate. His data set, he says, “enables researchers to compare outcomes across a variety of other types of family-structural history.” And he never speculated on causation—nowhere does he suggest that homosexual parenting or orientation was responsible for the lower outcomes of the children of GFs and LMs.
Whatever its faults, Regnerus’s study has unique strengths, even beyond the size and randomness of its sample, that his critics ignore altogether. His commendable attempt to include a diversity of views among his advisers is rare within the guild, where the leftism is unrelieved. So too were his willingness to immediately publish his research materials online and his pledge to make all his data digitally available this fall. Rather than a study of monochromatic and well-to-do lesbians or gay men, he managed to capture the full ethnic, socioeconomic, and geographic range of gay America. And his study is one of the first to systematically measure outcomes from the children themselves, rather than simply through the reports of their parents.
The limitations of Regnerus’s study compare favorably with the shortcomings found routinely in the same-sex literature. It does no credit to the guild that researchers have choked on Regnerus’s paper while happily swallowing dozens of faulty studies over the last 20 years—because, you can’t help but think, those studies were taken as confirming the “no difference” dogma. “If the Regnerus study is to be thrown out,” wrote the Canadian family economist Douglas Allen in a statement supporting Regnerus, “then practically everything else [in the literature] has to go with it.”
The “no difference” thesis was legitimized in a decree issued by the American Psychological Association in 2005. The issue of Social Science Research in which Regnerus’s paper appears coincidentally contains a study of the 59 studies the APA researcher cited in issuing its decree. Its author, Loren Marks, a sociologist at Louisiana State University, quantifies the weaknesses that Regnerus noticed in his reading of the literature.
“More than three-fourths (77 percent) of the studies,” writes Marks, “are based on small, nonrepresentative, convenience samples of fewer than 100 participants.” Nearly half did not use a heterosexual comparison group against which the study group could be measured. Many of those that did have a comparison group measured intact, well-to-do lesbian couples against single-parent heterosexual families. Outcomes were in most cases ill-defined and impossible to quantify: “socioemotional development,” for example, and “sex-role behavior.”
Most of these shortcomings were acknowledged by the researchers themselves in their respective papers, just as Regnerus points out the limitations of his own methods. APA acknowledged the shortcomings too—and then issued its decree anyway, in the most confident terms. But the accumulation of methodological errors calls into question whether any plausible conclusion can be drawn from gay parenting research.
Marks sums it up: “In response . . . to any question regarding the long-term, adult outcomes of lesbian and gay parenting we have almost no empirical basis for responding.”
And now, with the publication of Regnerus’s study . . . we still don’t.
Will we ever? The guild says no study of the kind Regnerus attempted would be acceptable unless it used a random sample of intact homosexual parents, drawn from a national sample, to measure against the intact biological families. No other comparison could be legitimate. But as Regnerus points out, the number of such parents in the general population is infinitesimal right now. A survey would have to take a national sample of hundreds of thousands of people (and cost millions of dollars) before it could randomly find the 500 or so stable homosexual couples necessary to make an ideal sample of their group.
Of course, with the dawning acceptance of homosexual adoption and homosexual marriage, the number of those couples will presumably increase. Which places policymakers in a double-bind. If they want to decide whether gay marriage and gay parenting are a good idea on the basis of widely accepted scientific studies, they will need a large population of homosexual parents to study, and that population won’t exist until we legalize gay marriage and wait 20 years. Until then, any deficiencies in gay parenting can be blamed on the fact that gays can’t marry.
The guild gets you coming and going.
On the phone, Gary Gates sounds much happier than Mark Regnerus. And why wouldn’t he? His side’s winning. As several of Regnerus’s allies point out, the professional intimidation of Mark Regnerus isn’t about Mark Regnerus—it’s about the next researcher who might attempt a study of gay parenting. The guild has put that poor fellow, crouching under his desk, on notice: Only some findings will be acceptable. (“That’s a nice little tenure-track job you got there. We’d hate for something to happen to it . . .”)
Needless to say, the intimidation of Regnerus is also about Regnerus, as Regnerus, facing a future of open-ended inquiries into his professional ethics, no doubt agrees.
I mentioned to Gates that Regnerus wasn’t the only one who was astonished at the vehemence of the reaction to his study. Gates assured me that no one should have been surprised.
“Let’s look at the nature of what was studied,” he said. “For LGBT people, this is about whether or not they can have children. Think about that. These are core, core things about being a human that are at stake. This goes right to their experience as human beings.
“And then you’re surprised that they react when an article appears that says if they have children, those children are more likely to be sexually abused? Of course they’re going to react.”
But as a researcher, I asked, doesn’t it make you uneasy to see a fellow academic hauled up on charges because he produced findings that somebody didn’t like?
“Look, I’ve been eviscerated by lots of people for some of the things I’ve written,” he said. “Larry Kramer called me a ‘horse’s ass.’ They were brutal. It comes with the territory.”
But that’s not quite what’s happening to Regnerus.
“If you’re asking did they try to bring me up on ethics charges, no, they didn’t go that far,” he said, laughing.
Andrew Ferguson is a senior editor at The Weekly Standard.
Recent Blog Posts