A new study of 6-to-17-year-old children of female same-sex households has been rushed to publication and is now making the rounds at the typical outlets, which are proclaiming that now the social science here is truly, genuinely, totally, finally settled. The problem is that the study doesn’t really accomplish anything near what its adoring fans claim it does. In fact, it all but undermines their wish for consensus.
Here’s what the new study claims: “No differences were observed between household types on family relationships or any child outcomes.”
Here’s what the study actually signals (and it didn’t take a PhD to see it): female same-sex parents report more anger, irritation, and comparative frustration with their (apparently misbehaving) children than do opposite-sex parents.
The study in question was published recently in the Journal of Developmental and Behavioral Pediatrics, and is based on data from the National Survey of Children’s Health, a 2011-12 effort that yielded nearly 96,000 completed surveys of parents in households with children under the age of 18.
In the pecking order of good study qualities, it has several things going for it, and I am happy to give credit generously where it is due. First, it focuses on “continuously coupled” households, which were profoundly rare in my 2012 study of 18-to-39-year-old adults answering questions about the households in which they had grown up. That is optimal, no doubt about it.
Second, it originates with a nationally representative sample—another big plus. However, when you start with tens of thousands of eligible cases but whittle down to comparing 95 female same-sex households with 95 opposite-sex ones, you quickly arrive at territory where statistical significance is going to be hard to locate. (Indeed, reducing sample size further and further from my original study is exactly how analysts came to proclaim that there was little statistical difference after all.) Basically, (sample) size matters. Yet this one is over twice as large as Charlotte Patterson and Jennifer Wainwright’s matched pattern studies of 44 same-sex households compared with 44 opposite-sex ones.
Locating an ample sample of same-sex households with children in population-based studies remains a challenge, inflated assumptions about their real numbers in the population aside. Hence it is still hard to randomly find stably-coupled same-sex households with children almost anywhere except in newspapers and on TV. Despite these common limitations, this study would seem to be an improvement. But it dismally fails to deliver what it proclaims, and it’s no stretch to say that. Why? Several reasons, the first of which is rather stunning.
Despite claims to the contrary, same-sex-couple moms display a problem in the study on a measure the authors oddly decided to label “parental stress.” That is, (presumably) lesbian mothers display notably more of it than do opposite-sex parents. The oddity I speak of is why they call the measure “parental stress” in the first place. It is not a measure of stress, and it doesn’t take a psychometrician to see it. Each parent respondent was asked how often in the past month they have:
- Felt that their child is much harder to care for than most children his/her age
- Felt that their child does things that really bother you a lot
- Felt angry with their child
The authors label as “stress” what is far more obviously a three-measure index of irritation and anger (at the child). Why are female same-sex parents more angry at their children than opposite-sex ones? I confess I don’t know. But this study unwittingly reveals that they clearly are. The effect size, moreover, is a “moderate” one, meaning it’s not tiny.
The authors even make overtures toward blaming the absent father for the irritation female same-sex parents feel at their children’s behavior. They don’t cite his absence, though. (That cannot matter, right?) Rather, they question his unknown genes and their possible influence on their child’s behavior:
The NSCH did not collect information about the source of the sperm used for conceiving the children of same-sex mothers. It is conceivable that there might be differences in family relationships, parenting stress, or child outcomes associated with whether the sperm donor was known (i.e., a friend, acquaintance, or relative of the mother) or unknown.
In reality, we don’t know if these children were the product of assisted reproductive technology at all. Nor do we even know if the women self-identify as lesbians, or are even in a same-sex romantic relationship. (At least my study verified the latter.) We must presume they are.
Instead of acknowledging that same-sex parents are irritated and angrier at their children more often than comparable opposite-sex parents, the authors framed the finding as noting simply that same-sex parents acknowledge more “parenting stress” than opposite-sex parents. It comes across sounding noble.
Additionally, the authors employ regression models, seeking to emancipate same-sex households from any responsibility for the health problems, emotional difficulties, coping behavior, and homework habits of children. To be sure, they find no direct effect of household “type.” What they do find, while failing to admit its implications, is a strong deleterious effect of “parental stress” on each of these outcomes except general health. That, in the social science world, is called an indirect (negative) effect of female same-sex households on child well-being by way of the former’s influence on so-called “parental stress.”
That’s the key takeaway from what is a modestly improved effort at evaluating same-sex households with children. Several other things catch my eye, though.
First, the rest of the measures they employ seem all but carefully selected. The NSCH survey, available online from the CDC, has lots of measures to evaluate child well-being. But the reader of this study is treated to a small number, including a general health measure. (Why would the average 8-year-old in a same-sex household be in worse health? I have no idea. Nor does it make sense to even evaluate it.) The authors overlook—that is, they do not include—measures of school progress, problems in school, participation in sports and recreational activities, volunteering, sleep, exercise, media consumption, reading, depression, bullying behavior, and all but one of five different measures of flourishing. A good question to ask them is: why not include all these measures?
Second, despite the improvements the study is still on the small side, leading to judgments of “no differences” in coping behavior (when, according to a parent, the child “stayed calm and in control when faced with a challenge”) that are—when you eyeball the study’s Table 3—quite different at face value. But when you only have 95 cases, the “power” to detect real differences in the population is limited. In other words, you can see that there is difference (and in almost every study I know it’s typically in more harmful directions). But unless you have a big sample, you’re left impotent, able only to state that there are “no statistically-significant differences.”
Observers have been hammered by this clever ploy for a decade now in the study of same-sex households. Judges and legislatures have been spoon-fed it and led to believe that statistics cannot deceive. They can, which is why I always report basic associations first. (To their credit, the authors of this study did, too.)
Third, the longstanding practice of discerning child outcomes by talking to parents about their children is getting old. Rather than asking questions of the children themselves (as I did in my 2012 study, and after they had left their parents' household), we still go to the parents as their spokesperson. I am a big fan of speaking to independent sources—that is, the children—and doing so anonymously.
The children of divorce, which has been legal in the US now for many decades, have never appeared comparable—on average—to the children of stably intact households. (The same is true of adoption.) Nor did their advocates insist we agree that they are comparable. In reality, there are kids who navigate all manner of household upheaval and diversity, often emerging scathed but resilient, going on to live productive and emotionally healthy lives as adults. I know lots of them; we all do. They have lived in straight and gay households, as well as those of the rich and poor, black and white.
No diligent scholar I know of has stated that same-sex couples make uniformly terrible parents whose efforts at childrearing are doomed to failure. No, what is new here is not the revelation of difference and the tacit acknowledgement that a stable, loving, married mother and father remains an optimal scenario. What’s new is that we are learning that legalized civil same-sex marriage and adoption laws are not enough. We have to agree that “the kids are fine.”
People think I have it in for the LGBT community(ies). I do not. I have it in for a science that refuses to proceed honestly, and instead shelters privileged groups—as it currently is doing—with a protective shell of administrators, grant-makers, and editors. Hence the Regnerus bashing will continue until further notice. So be it. I may be unpopular—there are more important things in life than that—but about the comparative advantages of stably-married households with mom, dad, and children, I am not wrong. It will take more than smoke, mirrors, and shifty rhetoric to undo the robust empirical truth.
Mark Regnerus is associate professor of sociology at the University of Texas at Austin, research associate at its Population Research Center, and a senior fellow at the Austin Institute for the Study of Family and Culture.