It was inevitable that someone would claim that children raised by adults who have or who act on same-sex attraction would be better off than children raised by normal adults, or by parents.

And so it has come to pass in the peer-reviewed paper “Psychology of Sexual Orientation and Gender Diversity: No Differences? Meta-Analytic Comparisons of Psychological Adjustment in Children of Gay Fathers and Heterosexual Parents” by Benjamin Graham Miller, Stephanie Kors, and Jenny Macfie in the journal Psychology of Sexual Orientation and Gender Diversity.

From the Abstract:

… The current study applied … meta-analysis to 10 studies … to evaluate child psychological adjustment by parent sexual orientation. …[R]results indicated that children of gay fathers had significantly better outcomes than did children of heterosexual parents in all 3 models of meta-analysis.

The emphasis on “better” was in the original — a word that was noticed in the popular press.

If the results are true, then surely if we want what is best for the nation’s children, they should be placed in the households of men who enjoy non-procreative sex-like activities. (Actual sexual intercourse can only take place between males and females.) Leaving kids to fester with their own parents dooms them to lesser outcomes.

That prescription might to your ears sound absurd, but it does follow if Miller and his co-authors are right. Are they?

The authors used a controversial technique,badly applied and in the service of confirmation bias.

The trio used a statistical technique called “meta-analysis,” which I jokingly define as a method to prove a hypothesis “statistically” true which could not be proved to be actually true. Actually, it is a way to glue together results from disparate studies, so that one needn’t be troubled by the hard work of investigating the disparate studies. In other words, it is a controversial technique, often badly applied and in the service of confirmation bias. I suspect that is true here.

Miller et al. gathered 10 studies culled from “a list of over 6,000 citations of published and unpublished studies from 2005 and later based on the search terms same sex, same gender, gay, child, and parent in any combination.”

Somehow — it is a mystery — in their diligent search, the researchers did not turn up the remarkable 2012 study known by all sociologists,  “How different are the adult children of parents who have same-sex relationships? Findings from the New Family Structures Study” by Mark Regnerus. That study made national headlines!

Regnerus’s study came to the unwanted conclusion that kids did better when raised by adults who did not have same-sex relationships. Regnerus’s work also showed that the rate of same-sex attraction of kids growing up under same-sex attracted adults was higher than for other kids, a finding which goes against the conventional wisdom that all those who have same-sex attraction are “born that way.” Doubtless, Miller and co-authors will correct the oversight of forgetting Regnerus in their next paper.

Back to the point of cobbling disparate studies together for the purposes of statistical modeling. The (alphabetically) first paper examined by Miller was the 2009 work “An Evaluation of Gay/Lesbian and Heterosexual Adoption” by Paige Averett, Blace Nalavany and Scott Ryan in Adoption Quarterly.

This study asked two groups of kids, 1.5 to 5 years and 6 to 18 years, sets of questions with arbitrary numerical answers about behavior (unfortunately an exceedingly common practice; see Chapter 10 of this book). Averett then reported on the differences in summaries of the numerical answers, and concluded that “child internalizing and externalizing behavior was not contingent upon adoptive parent sexual orientation.” In other words, it didn’t make any difference in outcomes whether kids had gay or non-gay minders.

This seems to be in Miller’s favor. But what is unusual is the nature of the children studied by Averett.

For example, for the 1.5 to 5 years old group of kids, the gay adults who raised them were all white, whereas the normal parents represented a mix of races (close to matching actual racial differences in the USA). The gays were much better educated; nearly 3 out 5 had Masters Degrees. Yet over 70% of normal parents only had high school educations. Not surprisingly, the gays made twice as much money as the normal parents. Only 1 out of 10 adoptions by gays was “transracial,” and it was about 4 out of 10 by normal parents. A little more than 3 out of 10 kids adopted by gays suffered previous abuse, whereas twice as many, some 7 out 10, of kids adopted by normal adults were abused.

And so on for other probative, obviously relevant differences. Conclusion? Averett stacked the deck. The statistical measures they derived were therefore meaningless, and thus should not be included in any list of studies, except in a list of papers which show How Not To Do Research. Miller and his co-authors should not have given this study any weight, but they did.

We could go through the other nine papers and make similar criticisms, but it would take too long, and besides, the point about the inadequacy of the meta-analysis wouldn’t change. What’s worse is that we’d miss the real error, which is this: “outcomes,” which is to say the lives of actual human beings, cannot be quantified so easily as Miller and the other authors contend.

Most important of all, no scientist can measure the spiritual well-being of any child (or adult), which, in the end, is the only metric that matters. And it is this well-being that, as all history teaches, is imperiled by these fashionable social experiments.