To Know What Social Science Says About Gay Parenting, Look At Studies Of Gay Parenting.

Sherif Girgis’ rebuttal to Richard Chappell’s withering critique of “What Is Marriage?” covers more than I can respond to in a blog post.

But I did want to look at Girgis’ discussion of social science and same-sex parenting. because the errors he makes are extremely common among opponents of marriage equality.

Regarding social science and same-sex parenting, Chappell criticized Girgis and his “What Is Marriage?” co-authors, writing:

Throughout the paper, the authors make reference to “sociological evidence [that] children fare best… when reared by their wedded biological parents.” (p.257) More specifically, a cited study tells us: “Children in single-parent families, children born to unmarried mothers, and children in step-families or cohabiting relationships face higher risks of poor outcomes.” (258) But the unsurprising fact that children tend to do worse in broken or unstable families is no evidence at all that gay stable families do worse at raising children than heterosexual stable families. […] More relevant studies would directly compare straight parents with gay parents (as opposed to comparing stable straight families with dysfunctional straight families).

Chappell is obviously correct about this. To understand how the ocean works, we have to study the ocean (or to consult published scientific literature about the ocean, I suppose, to make the analogy closer). You can’t get accurate answers about the ocean by reading statistics about Lake Placid.

Girgis responds by saying that studies that directly compare straight parents with same-sex parents do not “meet the acknowledged gold standard of social-scientific research, by drawing on large, random, and representative samples observed longitudinally.” He goes on:

Every parenting arrangement that has been examined in high-quality studies has consistently been shown less effective than parenting by married biological parents: this is true of single- and step-parenting as well as parenting by cohabiting couples. Moreover, and more relevantly here, studies also suggest that mothers and fathers foster—and their respective absences impede—children’s healthy development in different ways. So, as we tried to show, it would be surprising if same-sex and opposite-sex parenting were equally effective.

1. Girgis is illogical.

Girgis is wrong to imply that there are no high quality studies of same-sex parenting (see below). But even if he were right, his logic would still be atrocious.

His argument is identical to someone saying “I don’t have any good quality studies of the ocean, but I’ve got studies that compare the lake to the desert, to the mountain, and to the plains, and in every case the lake was wetter. Isn’t it better to draw conclusions from irrelevant but high-quality studies, than from a relevant study? Besides, I have a theory which tells me that lakes must be the wettest place. Therefore, it’s safe to assume that the lake is wetter than the ocean.”

Just as it’s illogical to assume that the ocean must be just like deserts, mountains and plains, it’s illogical to assume that stable same-sex parents are just like divorced families, single parents, and step-parents. To find out what social science says about same-sex parents, it is obligatory to read studies that examine same-sex parenting.

2. Girgis is Inaccurate.

Girgis implies that there is no such thing as a “high-quality” study that examines same-sex parenting. What is a “high-quality” study? Girgis doesn’t define his terms, but presumably it is related to the “gold standard” for studies he also mentioned:

What we need, however, are studies that meet the acknowledged gold standard of social-scientific research, by drawing on large, random, and representative samples observed longitudinally. […] none of the studies comparing children reared by same-sex couples in sexual partnerships to children reared by their married biological parents has these features…

Girgis seemingly assumes that if studies do not meet this narrow “gold standard,” then they are useless and their findings should be ignored. No authority or logic is cited to support this remarkable conclusion.

A question for Girgis: Why do leading social science journals — which use rigorous standards and peer reviewers who are (one suspects) far more qualified to recognize good study design than Mr. Girgis — routinely publish studies which are, Girgis implies, worthless?

Judith Stacey, responding to a similar argument made by Stephen Nock, commented (pdf link):

…To accept Professor Nock’s primary criticisms of these studies, [we] would have to dismiss virtually the entire discipline of psychology. The vast majority of research in child development, and in the field of psychology more broadly, would be invalidated as unscientific. The research design of the studies on lesbian and gay parenting that Professor Nock criticizes is by no means peculiar to or below the generally accepted scientific standards of the field. On the contrary, most of the research designs used in these studies characterize predominant methods employed throughout the entire discipline of psychology. In essence, Professor Nock appears to be claiming that psychologists would have to become demographers in order for their work to have scientific value. By that standard, almost every study published in the premier journals Child Development and Developmental Psychology would be rejected out of hand. […]

By Professor Nock’s standard, we would throw out a good portion of research in medicine that uses inferential statistics (e.g., all of those that conduct t-tests or chi-square tests for treatment effects on small non-probability samples). We would also have to reject the preponderance of research in psychology, psychiatry, social-psychology, anthropology, clinical research, and so on. In fact, some of Professor Nock’s own published research fails to meet his unreasonably narrow methodological standards.

I agree, of course, that for many social science questions (although not all of them) “large, random, and representative samples observed longitudinally” are ideal. I also agree that some study designs are so bad as to be worthless. But there’s territory between “worthless” and “ideal,” and that territory is where most good social science research happens. And when many good studies, using various methodologies and samples, all reach the same conclusion, that conclusion should not be dismissed because it does not meet some contrived and dubious ideal.

When Girgis writes that “Every parenting arrangement that has been examined in high-quality studies has consistently been shown less effective than parenting by married biological parents,” he is simply, baldly wrong. There are many high-quality social science studies of same-sex parenting; this research is peer-reviewed, meets the scientific standards of the field, and published in high-quality journals. But, inconveniently for Mr. Girgis’ views, this research has all found that same-sex parenting doesn’t measurably harm children. (See, for example: Brewaeys, Ponjaert, Van Hall and Golombok (1997); Golombok, Tasker and Murray (1997); Wainwright, Russell, and Patterson (2004); Chan, Raboy, and Patterson (1998); Golombok, Perry, Burston, Murray, Mooney-Somers, Stevens, and Golding (2003).)

3. What about the studies Girgis relies on?

I thought of checking the research Girgis relies on, to see if it meets his “gold standard.” Unfortunately, his fourthhand citations make checking his research effectively impossible.

When Girgis says “Every parenting arrangement that has been examined in high-quality studies has consistently been shown less effective than parenting by married biological parents,” he links to a report from the Witherspoon Institute, a respected but ideologically right-wing organization.

But that report relies on review articles, not on research. I looked up this research brief from the Institute for American Values (chosen because it’s cited multiple times in the Witherspoon report, and because its conclusions support Girgis’ claims). (Full disclosure: I’m a guest blogger on “Family Scholars Blog,” which is run by IAV.)

The IAV brief, in turn, does not directly cite any research; it primarily references (in a note at the end) this review article, “Family Matters,” published by the Alabama Policy Institute.

It is only with “Family Matters” that we finally reach a document that reviews research literature, rather than reviewing reviews of literature. However, “Family Matters” opens with a lengthy, admirable introduction in which the authors frankly discuss the many ways the research they rely upon for their findings is not perfect; they don’t claim that only “gold standard” research is worth considering.

This maze of indirect citations makes it unreasonably difficult for Girgis’ readers to find out what research Girgis is actually referring to. For the sake of intellectual accountability, Girgis ought to either cite research directly, or at least cite reviews of research, rather than reviews of reviews of reviews of research.