Regnerus study (cont.)

10.30.2012, 11:21 AM

At Witherspoon’s Public Discourse today, Matthew J. Franck posts the first of two articles defending the Regnerus study.   To me, the article disappoints.  There is a fair amount of glib cheerleading.  And much is made over the fact that many dumb, politically-correct things have been said about this topic in the past, and that previous studies of this topic have been deeply flawed (all of which in my view is true, but aren’t we talking about THIS study?). 

But to me the real kicker is that Franck acknowleges briefly, but then declines to devote even one sentence of his article to discussing, the core scholarly flaw (in my opinion and in the opinion of many many others) in the study – which is that, in a study that announces itself as a study of  ”new family structures,” which will for the first time with a valid data set compare those  ”new family structures” to other “family structures,” Regnerus establishes as the pivot point of his analysis, the very centerpiece of his methodology, a phenomenon (did a parent have a same-sex romantic relationship while you were growing up?) that absolutely, by any reasonable standard, simply cannot be called a family structure.  

This is hardly a small point. It’s what almost all of the (legitimate) fuss has been about.  But Franck has nothing to say about it.  Maybe he’ll discuss it in part two.

Update:  Here’s a new interview for Citizen Link/Focus on Family, in which Regnerus does (unlike Franck) discuss the main alleged problem with the study, attributing it primarily to inherent problems of the sample itself (not enough gay/lesbian households), while also saying that, if he was doing it over again, he’d be clearer about all of this.  (Actually, in my view, in the study’s fine print, he IS pretty clear.)  But even in this interview, in my view, he does not squarely talk about the serious problems connected to calling something that is not a family structure a “family structure,” and then comparing it as the study’s centerpiece to other (actual) family structures. 

Update II:  The first comments on this post had to be deleted.  Please abide by our civility policy in your comments.  If all you want to do is browbeat and name-call and vent, please don’t waste your time or ours — such comments won’t be allowed.


29 Responses to “Regnerus study (cont.)”

  1. Even if one overlooks the flaws in methodology and the sampling bias, we are still left with a study that answers a question that nobody asked.

    Franck seems more interested in portraying Regnerus as a victim than defending his research. Matthew J. Franck is the Director of the Center on Religion and the Constitution at the Witherspoon Institute. While much research is funded by people with a vested interest in the outcome, it seems improper for the funders to defend the scientific flaws.

  2. BTW David, with respect to the sampling bias, someone forwarded an email to me from Mark Regnerus. The headers confirm its legitimacy (i would be happy to provide it to you).

    We oversampled respondents who said their mother or father had a same sex relationship (as well as kids who were adopted). Under sampled some other groups that occur more frequently in the population. Then weighted the data to account for this. Standard procedure in survey data collection and analysis.

    That does not seem to comport with the study’s findings and obviously exaggerates what seems to be the key finding!

  3. David Blankenhorn says:

    DHC: I’m no expert on survey sampling procedures, and maybe because of that I don’t follow your point on survey bias. Where is the alleged bias in this case? In my experience, one over-samples in order to get larger numbers of a group that in the population is small (I’ve done that before); it’s standard and often useful and there is nothing per se wrong with doing it, as a matter of methodology. Or maybe I’m missing something?

  4. Kevin says:

    Witherspoon paid $750,000 for this study. Not surprising they are defending it to the end. While there may be few data collection problems, the problem is that the findings in Mr. Regnerus initial article (in Social Sciences Research) are unsupported by his data. Using children from broken homes/marriages, and concluding that their troubles are due to having a parent who had a same-sex affair, is fatally flawed. It’s like saying that the marriage breakup isn’t the source of the child’s problem but rather the affair is.

    The good news is that publicly discussing this “study” further exposes it for what it is. What’s that saying about the sanitizing effect of sunlight??

  5. Alana S. says:

    The time will come when the sample size is big enough.
    We just have to be patient.

  6. You are correct. Oversampling is used to correct the data – not to induce a bias. In this case, 8% of the net respondents claimed to have a parent who, at some time, engaged in a same-sex relationship.That percentage is at the very core of the study and it doesn’t seem possible. Keep in mind that the Regnerus sample is composed of the children of people in traditional marriages.

    He told Focus on the Family:

    We completed interviews with 3,000 people. And while 248 by itself isn’t a lot, you have to understand we’re talking about very, very few people in the first place. Finding someone whose parent had some sort of same-sex relationship as they were growing up is like trying to find a needle in a haystack. I got taken to task for leaning on young adults’ assessments of their parents’ relationships.

  7. Diane M says:

    @David Cary Hart – it doesn’t seem impossible to me, but maybe I’m missing something. “In this case, 8% of the net respondents claimed to have a parent who, at some time, engaged in a same-sex relationship.That percentage is at the very core of the study and it doesn’t seem possible.”

    Say 3-6% of the population self-identify as gay or lesbian. Given when these kids were growing up, many or most of them went ahead and straight-married anyhow. They didn’t marry each other, so I think you would have more than 3-6% of families where someone was really gay or lesbian. Add in a few people who were bisexual or just had a same-sex relationship because they were questioning and it sounds possible.

    Am I missing something?

  8. Diane M says:

    @David Blenkenhorn – I think your analysis is spot-on.

    “Regnerus establishes as the pivot point of his analysis, the very centerpiece of his methodology, a phenomenon (did a parent have a same-sex romantic relationship while you were growing up?) that absolutely, by any reasonable standard, simply cannot be called a family structure.” Franck never discusses this problem. Maybe his part two will, but I think in a subject this contentious it would help if at least let us know he would get to it.

    Similarly with Regnerus, he never quite acknowledges the flaw. The problem is not that he called someone a lesbian parent based on behavior not self-identification. I am fine with calling a woman who has a female lover a lesbian or bisexual. And even if she were a straight woman who had a lesbian affair, it would probably affect her parenting in the same way.

    The problem is with comparing someone from a divorced family with someone from a never-divorced family that just happens to be a heterosexual couple and then concluding that sexual orientation was what mattered. We already know that divorce is hard on kids – the rest of his data even confirm that. And none of the lesbian or gay couples in his study even had the possibility of marrying.

    Regnerus still doesn’t seem to see that this is inaccurate. I think he must really want to be able to say that being raised by heterosexual parents is better for you and he is blinding himself to reality.

    Franck seems to be doing the same thing, but he could still say something in part II.

  9. David Blankenhorn says:

    On “did your parent … ?”, my understanding is that he got the number he got (8 percent of respondents) precisely because he over-sampled for that group — and in my experience that is a normal procedure; nothing wrong with it (that I can see).

    For example, if you are surveying a city, and you know that only 2 percent of the residents are Latino, and yet for your own reasons you want to look meaningfully at Latino residents as a subgroup of city residents, you will tell the survey firm to over-sample for Latinos, so that you can get a large enough number of this subgroup to make meaningful statements about …

    That doesn’t mean that 8 perecent of the population answers “yes” to that question; in fact it means that 8 percent won’t say yes to that (unless you oversample for that group) ….

  10. Tristian says:

    I’m not sure the problem with Regnerus’ definition of “gay households” is as severe as it is being made out to be.

    The problem is this. We want to compare the right two things, one of which is heterosexual parenting , or a heterosexual family structure. One unquestioned assumption is that the benchmark is a child being raised together by its two biological parents, and that seems right, because in a sense everything thing else is a departure from that. So if we’re going to compare heterosexual households to gay households we need to look at the gay counterpart to a child being raised by its biological parents. Except there’s virtually no such thing. (Literally this would be a gay man and a gay woman having a baby and raising it together, which I take it is very rare). So what is the “gay family structure” we’re supposed to be looking at as the counterpart?

    Well, here there’s a dilemma. We might insist that we compare heterosexual parenting with the next closest thing, which is maybe a gay couple raising an adopted baby from infancy. The problem is a) there are still comparatively few such cases, hence Regnerus’ difficulty in getting a decent sampling; and b) this seem very unfair to gay single parents, gays who bring a child from a heterosexual marriage into a later gay relationship, etc. , gays who adopt older children, etc. Aren’t they gay parents? If not, why not, especially if this more typically how gays come to be raising children? But if you include those in your understanding of a gay family structure, you end up defining things in a way not all that removed from what was actually used in the study.

    Now, someone might say in response, the problem here is making the two biological parents model normative for heterosexual parenting, as many—maybe by how most—children don’t grow up with that even if neither parent is gay. So we should put that aside and compare gay single parents to single mothers and fathers, gay divorced parents to straight divorced parents, etc. And that does seem right. But in so far as the explanation of the worse outcomes of Regnerus’ samples are explained away by pointing to the bad effects of broken homes, single parenthood, and the like, this also seems like a damning admission. Aren’t we still conceding then that gay parenting is more typically analogous to (statistically) less successful forms of heterosexual parenting?

  11. Kevin says:

    Regenerus’ fatal flaw isn’t how he collected his data, it’s the conclusions he drew that weren’t based on his data, such as the absurd notion that children raised by gay fathers are more likely to commit suicide. His data simply don’t say that. Common sense would tell us that one’s sexual orientation or practices have little to do with one’s parenting skills.

    His data DO show that children from broken marriages, where a parent had an affair, and the affair was with someone of the same sex, have worse outcomes than children whose parents didn’t get divorced. Well, duh!

    Newsflash: children from broken homes are more often messed up than children from intact homes. That Regnerus chose to blame the problem not on the the divorce, or on the affair, but on the sexual nature of the affair, is offensive, to say the least. That’s why his study conclusions are being widely rejected: his data don’t support his conclusion.

  12. David Blankenhorn says:

    Tristian: For what it’s worth, my own guess is that child outcomes in homes headed by gay/lesbian couples are likely to resemble child outcomes in stepfamilies (for reasons suggested in your comment).

    And I agree with you that figuring out how to make valid comparisons here is quite difficult and may in fact be impossible without spending far more money and time than anyone could possibly spend.

    But I still want to insist that, whatever the RIGHT way to compare “new family structures” to (other) “family structures” might be, the way that this study chooses and seeks to do it is NOT the right way.

  13. I am over-quota: David: I don’t buy the oversampling. You still end up with 250 out of 3,000 no matter how it is sampled. However, I will review this again on the Witherspoon site. If I am incorrect I will find the connect thread to say so. I want to be intellectually honest above all else.

    Kevin: You are correct. Although it’s not just the conclusions. He summarized his study incorrectly (as he admits in the Focus on the Family piece). He also failed to speak out when an organization related to Witherspoon misused the study to the point of claiming that apples are oranges.

  14. Mont D. Law says:

    (Aren’t we still conceding then that gay parenting is more typically analogous to (statistically) less successful forms of heterosexual parenting?)

    I have no problem with this as long as you pass a law that says no one is allowed to legally marry if their relationship leads to statistically less successful forms of parenting. So no more stepparents.

  15. Roger says:

    Mont D. Law makes, as usual, a very pertinent comment. Deadbeat dads and moms are not prohibited from marrying. Neither are (repeatedly) divorced persons. Neither are child molesters. Neither are felons, not even murderers. I suspect that none of these are ideal parents, yet no one is campaigning against them marrying or doing studies on whether they are less than ideal parents as an intellectual exercise.

    But somehow it is seen as appropriate to concoct an intellectually dishonest study for the express purpose of discrediting gay and lesbian parents so that it can be used in legal proceedings about the right of gay and lesbian people to marry. The study is corrupt both as social science research and as a legal tactic.

    There is no way that Regnerus can explain away the defects of his ill conceived “research.”

  16. Phil says:

    But somehow it is seen as appropriate to concoct an intellectually dishonest study for the express purpose of discrediting gay and lesbian parents so that it can be used in legal proceedings about the right of gay and lesbian people to marry.

    Roger, I’m not sure you’re correct. My recollection is that Regnerus specifically stated in his study that his research was not applicable to the question of whether SSM should be legal. I think he also wrote that anyone attempting to cite his study as a reason to oppose SSM should be slapped, repeatedly, but I may be misremembering his exact words.

    Aren’t we still conceding then that gay parenting is more typically analogous to (statistically) less successful forms of heterosexual parenting?

    Tristian,

    You’re saying that if we compare gay divorced dads to straight divorced dads, we’re conceding that gay parenting is more typically analogous to less successful forms of parenting. I’m not sure how you arrive at that conclusion. If we compare gay apples to straight apples because we don’t have any gay oranges to study, then I think we’re just being reasonable.

    You could just as easily say that heterosexual parenting is more typically analogous to less statistically-successful forms of parenting.

  17. Tristian says:

    Phil, let me work backwards. We routinely do say heterosexual parenting is analogous to less successful parenting. Or, more accurately, that much of it departs from the most successful model (statistically speaking). But of course that model is supplied by a kind of heterosexual parenting. So no one can use that to argue the inferiority of straight parenting. But if indeed there are no ‘gay oranges’–no models of gay parenting that can match the best model of heterosexual parenting–we seem to be saying that there’s something about gay parenting that keeps it from reaching that level. That’s what strikes me as a serious concession.

    For the record I am not tempted to draw any legal conclusions from any of this. Also, it’s worth noting that Regnerus did compare his (misleadingly labelled) gay and lesbian parents to divorced straight parents, and found some unfavorable differences there as well.

  18. JHW says:

    Tristain: You seem to think that the question is whether the average situation where a gay person is parenting is as good as the average situation where a straight person is parenting. But that’s not the case. The question is whether sexual orientation, or the fact that the parents are two men or two women rather than a man or a woman, has relevance to parenting quality in itself, beyond the other sorts of information we might know about the family’s structure. I don’t see how pointing this out is an important concession at all.

  19. Roger says:

    Phil, Regnerus has said different things to different people, depending on his audience. He has told conservative publications that the study indicates that gay and lesbian parents are significantly worse than heterosexual parents and is relevant to the marriage debate. He tells more mainstream and liberal interviewers that it is not relevant to the marriage debate. What is noteworthy is that the study has already been cited in the filings to the U.S. Supreme Court in regard to the Prop 8 and DOMA cases that are pending for writs of certiorari. That, I suspect, was the real purpose of the Witherspoon money, to find some “peer-reviewed” article that could be used to say that Congress (or in the case of California, voters) had a rational, non-biased reason to deny equal rights to gay and lesbian couples. Luckily, by now the study has been exposed as the absurdity that it is and attorneys arguing against Prop 8 and DOMA will point that out.

  20. Tristian says:

    JHW, there are indeed two questions here, but I think I’m focusing on the right one. Let’s consider an easy case for illustration. If I’m looking to find the cause of a disease—let’s say Tay Sachs disease—ethnicity will prove itself to be irrelevant. Being Ashkenazi doesn’t cause Tay-Sach’s disease. However, if I’m looking to predict patterns of Tay-Sach’s disease in a population ethnicity does matter—having the gene that causes it is significantly correlated with being Ashkenazi. For that kind of a question the correlation is more important than the cause.

    Now, we can make the same move with parenting. For some questions ethnicity is irrelevant—being Asian is not itself going to be an important cause of any parenting outcomes. But being Asian may correlate with parenting practices that are causally relevant to outcomes. So we can meaningfully ask whether children of Asian parents are more likely to be X, Y or Z than children of white parents, etc.

    It works the same with gay parenting. One question is whether sexual orientation itself is causally implicated in parenting outcomes. I take it that’s the question most people here think is most important. But there’s another perfectly good question here: is gay parenting as it actually happens correlated with certain outcomes? That’s the question I’m asking, and this is the right question to ask if what were interested in are questions like what will be the likely results of more and more gay parenting? And here I’m surprised to see people conceding so quickly that in fact gay parenting correlates with conditions that seem to predict comparatively negative outcomes. To make that point more forcefully, consider the following familiar argument:

    1) Gay marriage will mean more gay parenting
    2) Gay parenting is typically inferior to the best model of parenting we have (children being raised by both biological parents)
    3) Therefore gay marriage is bad for children

    I think this argument fails for many reason, but it doesn’t help in making that case to deliver the second premise on a platter. But the thing to note is that whether or 2) is true does not turn on the cause of the negative outcomes—a significant correlation is all the premise needs. Note too that Regnerus himself explicitly denies that he has shown sexual orientation to be the cause of the negative outcomes, or that it is even a plausible candidate as a cause of the negative outcomes. His conclusion is in fact the one that people here are suddenly happy to reach: gay parenting reliably correlates with negative outcomes.

  21. Mont D. Law says:

    [ gay parenting reliably correlates with negative outcomes.]

    So what. Lots of parenting reliably correlates with negative outcomes. And when you break it down the reasons for the negative outcomes are identical. Gay/straight – white/black/Asian – family instability creates negative outcomes for children. Are there identifiable groups of people that are more likely than others to have problems forming stable families? Yes – but again so what. The group identification is irrelevant to the negative outcome. Being Asian doesn’t protect a family from instability. Being black doesn’t guaranty instability. You don’t really seem to have a point beyond playing the contrarian.

  22. JHW says:

    Tristian: I see what you’re saying, but I still think you’re missing the point.

    Generally speaking, the questions we care about are the questions that matter, the questions that ought to determine public policy or even just cultural norms and personal rules of conduct. If it were the case that parental sexual orientation or same-sex parenting, in itself, were harmful to children, that would probably be relevant to those questions. (It would not be as relevant as some people suggest, but it might, for instance, justify as a general principle that different-sex parents ought to be preferred over same-sex parents in adoption, all else being equal.)

    But if people with a parent who had a same-sex relationship are on average worse off because most of them were the product of a broken different-sex relationship, it makes no sense at all to argue that this undermines the case that equal treatment of same-sex couples in marriage and parentage. The whole point of the critique of the Regnerus study is that the gay parenting “in practice” that Regnerus’s study observes is not at all akin to the gay parenting that is actually at issue. There is no one who is an advocate of gay people entering into different-sex marriages, having children, and then ending the marriage.

  23. Kevin says:

    Although I continue to believe that Regnerus’ study was designed with a propaganda purpose, even Regnerus had to admit that his (faulty) conclusions supported legalizing same-sex marriage, in order to avoid the family instability that is the primary cause (whether he wants to state it or not) for the negative outcomes of the adult children he studied.

  24. Diane M says:

    @Tristan – “Aren’t we still conceding then that gay parenting is more typically analogous to (statistically) less successful forms of heterosexual parenting?”

    The children of the gay parents in this study were 18-39 years old. They were probably born between 1973 and 1994.

    At that time NO gay parents had access to the most successful form of family for heterosexual parents, marriage to each other.

    Adoption was generally not available to them. Technology for assisted reproduction was also difficult to get.

    Social pressure was such that many gay and lesbian people still went ahead and entered heterosexual marriages.

    So yes, I am willing to concede that in a society without acceptance and support for same sex relationships, then same sex families will end up looking like heterosexual families that involve adultery and divorce. For one thing, many of them will start that way!

  25. Diane M says:

    “But there’s another perfectly good question here: is gay parenting as it actually happens correlated with certain outcomes? That’s the question I’m asking, and this is the right question to ask if what were interested in are questions like what will be the likely results of more and more gay parenting?”

    @Tristan – It depends what causes the outcomes. I maintain that gay parenting as it actually happened in the past was correlated with certain outcomes precisely because the parents were not allowed to marry each other or adopt.

    If I had to define the family structure I would call it:

    heterosexual-parents-where-one-or-both-parent-had-a-same-sex-relationship

    I would break the groups down by whether or not the parents got divorced and whether or not the parents had affairs before the divorce.

    More gay parenting in the future doesn’t have to look like the past. It doesn’t have to start with adultery and divorce. It can start with marriage.

  26. Tristian says:

    One last post and then I should quit. Diane, yours are the next set of questions that need to be asked, because they get at the critical issue: is the correlation between gay parenting and negative outcomes one that we can break? For what it’s worth, Regnerus concedes his study can’t answer this question, but he does in the more recent paper argue that gay marriage probably won’t much change the circumstances under which gay parenting actually happens.

  27. [...] the FamilyScholars.org blog, Blankenhorn rushed to comment yesterday, after the first of my two installments went up, that I had not addressed what he regarded as the [...]

  28. Kevin says:

    “….is the correlation between gay parenting and negative outcomes one that we can break?”

    There is no known correlation between gay parenting and negative outcomes, and even if there were, so what? What are the social or public policy implications? We don’t make public policy on potential parenting abilities or outcomes. If we did, poor blacks might be denied the right to parent. If the goal is to decide that gay people make bad parents, and then decide to deny them the right to marry because of it, you might find yourself opening a can of worms you don’t really want to open. Why stop at judging only gay people as unfit parents?

  29. Kevin:

    Simply stated, the agenda of this study and the organization that paid for it was to make a case against marriage equality. Secondarily it is to support the Pope’s 2003 statement that gays do violence to the children they adopt.

    Otherwise, this study answers a question that nobody asked in the first place. As David Blankenhorn correctly points out, this is not an honest exploration of family structures.