Well Mark Regnerus is in the news again. Why do I care? I work for a science agency that strives to provide the best available science to policy makers tasked with making public policy decisions. It’s like a kick in the nuts every time I see bad science (what’s even worse in this case, it’s bad science that apparently everybody knows is bad) trundled out into a policy forum.

There’s been a considerable amount of fair criticism of the Regnerus paper. A quick stroll through the inter web will turn up many summaries of these critiques for the lay person.
Here’s one.

The basics are this. He engaged in some analysis using the National Family Structure Study and claimed that his analysis provided unequivocal proof that children of same-sex parents were/are worse off then children raised by opposite sex couples. Not surprisingly, this earned him some popularity with certain groups with a certain social agenda. I make no claims to know statistically whether children of gay or straight couples are better or worse off. I know this however: Regnerus’ paper was crap. I’m not even saying anything inflammatory here…everyone knows it was crap. The journal that published it conducted an auditwhich concluded the paper was crap.

My brief synopsis for the econometrics homies: Regnerus purports to have a sample of children raised by: i) their biological (opposite sex) parents, ii) some other permutations of opposite sex couples, and iii) same-sex couples. He makes some pretty strong claims about how children raised by these various groups fared. He even goes as far as to say that children raised by same-sex couples not only fared worse than children raised by their biological parents but also faired worse than children of an opposite sex couple who experienced a family break-up. The problem is that his sample did not contain any (enough to make statistical inference) children raised by same-sex couples who did not experience a family break-up. That is, he combined the 2 observations where the child was actually raised by a same-sex couple with, for example, children who came from households originally led by an opposite sex couples that had suffered a family break-up (when, for example, the mother or father came out as gay later in life).

Regnerus conflates the effects of growing up in a broken home with the effects of growing up with one or more gay parents. Since his sample does not contain enough children that were reared by same sex parents (so he groups children of same-sex parents whether they suffered a family break-up at some point or not), his study is not capable of disentangling the marginal impact of gay parents from the effect of broken homes.

That, in itself wouldn’t be the end of the world. Every applied researcher knows your data are never going to be perfect. The code empiricists live by is:

1. Be transparent about your methods,
2. Be forthcoming about the limitations of your study,
3. Subject to 1 & 2, do the best you can with the data you can get.

Mark Regnerus broke rule number 2 and there’s sufficient evidence that he did so with mal intent. The fact that this guy parlayed some suspect analysis into a spot as an expert witness to a state legislature just seems like a move any think tank/consultant would be proud of. It’s politics, which I have a hard time getting too worked up over. But the paper’s bad science. I find that unforgivable.