New Study on Same-Sex Parenting Raises Questions

There's a new study out next month, by Dr. Mark Regnerus, about same-sex parenting.  The study relies on a national probability sample -- and that is a big deal.  Nobody has ever done that before.  Since any general sociological finding not based on a probability sample has approximately the same weight as cheesecake on a serving plate made of cavorite, we have not, to date, really known anything solid about same-sex parenting and its long-term effects on children.

Now, to be fair, the American Psychological Association argues otherwise.  Its apologists are rather more vehement:

...all credible research over the past 20 years has shown conclusively that children of same-sex couples perform at the same rates as children of opposite-sex couples. Strong, unchallengeable research which compares children of single parents to children of other single parents, long-term couples to long-term couples, and comparable home lives. Science is, by nature, unbiased, honest, and fact-driven. To call Regnerus’ work “science” is just laughable. (Source)

This bit of rhetoric is factually incorrect.  Not even the APA would call its announced consensus "unchallengeable," given the lack of large sample sizes, good sampling techniques, or solid longitudinal data on same-sex parenting.  Critics of the APA consensus (this blogger included) consider the APA's consensus to be driven more by politics and personal desire than by science, given the manifest gaps in the available data.  (This is not the APA's fault; same-sex parenting is rare today, and was vanishingly rare just a few years ago.  It is hard to study populations that barely exist.)

The state of the science on same-sex parenting today is similar to the state of the science on divorced parenting in 1970, just as the divorce revolution was beginning.  Then, like now, there were no really good studies of the effects of divorce on children.  Then, like now, there was an emerging sociological consensus that divorce was certainly not bad for children, and might even be a positive good in most cases.  Then, like now, that consensus was based almost entirely on studies of well-to-do, white, upper-middle-class families, and it assumed their conclusions would apply equally to other social classes and racial minorities. Then, the consensus was proved absolutely, overwhelmingly, catastrophically wrong about twenty years later, after ten years of longitudinal studies of widespread divorce demolished the theory that the children of divorce turned out just fine (or better).  Divorce, we now know, is very bad for children, and, had we been a little more cautious about it jumping to conclusions in 1969, we might have avoided a lot of human misery. Social scientists should take great caution before walking down that garden path again.  So I restate: to date, we have not known anything solid about same-sex parenting and its long-term effects on children.  A large probability sample, like the one Mr. Regnerus publishes next month, has the potential to change that.

Now for the bad news: it doesn't.  Yet.  The presumed object of Mr. Regnerus's study was to determine whether, all else being equal, there are real, observable, important differences between the children of same-sex parenting and the children of two-sex parenting.  His study reaches a much more modest conclusion: yes, there are real, observable, important differences between the children of same-sex parenting and the children of two-sex parents... but the study has not yet screened out the very important variable of household stability.  The way Mr. Regnerus's study is set up, he groups every family into one of eight categories:

1. IBF: Lived in intact biological family (with mother and father) from 0 to 18, and parents are still married at present (N = 919).

2. LM: R reported R’s mother had a same-sex romantic (lesbian) relationship with a woman, regardless of any other household transitions (N = 163).

3. GF: R reported R’s father had a same-sex romantic (gay) relationship with a man, regardless of any other household transitions (N = 73).

4. Adopted: R was adopted by one or two strangers at birth or before age 2 (N = 101).

5. Divorced later or had joint custody: R reported living with biological mother and father from birth to age 18, but parents are not married at present (N = 116).

6. Stepfamily: Biological parents were either never married or else divorced, and R’s primary custodial parent was married to someone else before R turned 18 (N = 394).

7. Single parent: Biological parents were either never married or else divorced, and R’s primary custodial parent did not marry (or remarry) before R turned 18 (N = 816).

8. All others: Includes all other family structure/event combinations, such as respondents with a deceased parent (N = 406).

Families who fit into more than one group were sorted into the highest applicable group.  Thus, the biological child of a married gay man and a lesbian woman, who then put that child up for adoption at age 6, into a family which then went through a divorce, and where the father subsequently remarried, and the mother finally died, would be classified as belonging to an LM family.  Detailed data on the prevalence of this "forced exclusivity" in the dataset was not available in Mr. Regnerus's article -- but the obvious effect would be to make the samples for gay and lesbian households disproportionately unstable.

On top of that, for the purposes of the study, an otherwise intact biological family where the mom had a three-month affair with another woman in 1996 would be coded a "Lesbian Mother (LM)" family -- exactly the same as a faithful, committed lesbian couple who had a child by anonymous sperm donor and raised that child in an intact household from birth to age 18.  Again, the statistical effect is to actually encourage more instability to show up in the gay/lesbian household samples, while hiding them from other groups.  In this study, "all else" is certainly not equal.

So, at this point in the game, same-sex marriage advocates would be well within their rights to defend the null hypothesis against Mr. Regnerus's new study, ascribing all the negative effects of same-sex parenting found in this study to household instability, rather than to an inherent defect in the parenting capabilities of homosexuals.  (Jim Burroway does a fine job of that at Box Turtle Bulletin, making many of the same points this blog is making.)  Of course, same-sex marriage advocates, if they are being intellectually honest, would have to admit that this is almost identical to the flaws pointed out by anti-redefinition campaigners in virtually every other study on this question, and that, if this study can't be trusted, no study out there can be trusted.  No conclusions can currently be drawn on same-sex parenting and its effects on kids.  Trying to draw such conclusions from it would be akin to drawing conclusions from the Guttmacher Institute study on religion and contraception I mocked in these pages a few months ago.

Now for the good news: although the way this data is being analyzed is not as useful for the marriage debate as culture warriors on either side might hope, that's going to change.  The New Family Structures Survey (NFSS), which provided the data for this study, was well done, and captured a great deal more information than was presented in this study (much of it relevant to answering the very questions we are raising).  In particular, there were a series of "calendar questions" that will allow scientists to closely examine -- and screen for -- varying levels of household instability in various sample groups.  Mr. Regnerus will be continuing to study that data, and specifically promised to take a closer look at that part of the data.  In general, Mr. Regnerus seems well aware of the limitations of his data, and that's always encouraging in a researcher.  Most important of all, the complete NFSS dataset is scheduled to be made public in Fall 2012, after Mr. Regnerus is done with it.  That will allow we, the public, to check his work, while digging into the data for further insights and alternative explanations.

Soon, we may have our first (maybe second) study of same-sex parenting that has enough statistical power to be generally trustworthy.  I'm excited, as ever, to see what the research uncovers -- but, for now, I'll be keeping my powder dry, even if that means I have to miss out on the blog-o-rumpus that generally follows a study like this.

For now, this study can be treated as a compelling reminder that household instability is a very bad thing for kids, regardless of whether same-sex relationships are involved.  There hasn't been any serious debate about that for twenty years, but it's always nice to have a consensus reaffirmed.

A final disclaimer: I've only had a few hours to think about this study, so I'm more open than usual to correction and comment, should my understanding of the study or my reasoning based on it be proved flawed.