While Mark Regnerus’s Social Science Research essay has been all the rage, another essay by LSU scholar Loren Marks published in SSR also deserves attention.
Marks’ contention is that the American Pyschological Association prematurely and inaccurately concluded in its 2005 brief on lesbian and gay parenting that “Not a single study has found children of lesbian or gay parents to be disadvantaged in any significant respect relative to children of heterosexual parents.”
Examining the 59 published studies that the APA relied on to make its conclusion, Marks finds the following.
1. “More than three-fourths (77 percent) of the studies cited by the APA are based on small, non-representative, convenience samples of fewer than 100 participants.” Further, many of these studies were racially homogenous, focusing on white gay couples. Furthermore, only eight of the 59 published studies focused specifically on outcomes of children from gay fathers. Of those eight, four did not include a heterosexual comparison group. Of the four that did include heterosexual comparison groups, one of them relied on a heterosexual comparison group of two single fathers.
2. Of the 59 studies relied on by the APA, 26 (or 44 percent) did not include a heterosexual comparison group, which as Marks notes, “In well-conducted science, it is important to have a clearly defined comparison group before drawing conclusions regarding differences or the lack thereof.
3. Of the 33 studies that did include heterosexual comparison groups, at least 13 of them used single-parent families as the heterosexual comparison group. The remaining 20 studies with heterosexual comparison groups are ambiguous about the nature of the comparison group, referring to them only as “mothers” and “couples,” without specifying if they were biologically intact families, or stepfamilies, or cohabiting families, etc.
4. Contrary to the APA’s assertion that “Not a single study has found children of lesbian or gay parents to be disadvantaged in any significant respect relative to children of heterosexual parents” — there was at least one notable exception: Sarantakos’s 1996 study. That study had a sample size of 174, the seventh-largest sample size of the 59 published studies listed by the APA. However, the other six with larger sample sizes relied on adult self-report studies, whereas Sarantakos’s study specifically examined children’s developmental outcomes, making it the largest study to specifically study children’s developmental outcomes. What did Sarantakos find? “Overall, the study has shown that children of married couples are more likely to do well at school in academic and social terms, than children of cohabiting and homosexual couples.” Why did the APA not take this study into consideration, particularly if it had the largest sample size that specifically addresses children’s developmental outcomes? They dismissed it because (a) the Sarantakos study was based, in part, on “subjective reports by teachers” (which is inferior to subjective reports by parents, as is frequently done in the same-sex parenting literature upon which the APA relied?), even though, as Marks points out, some of the assessment was based on “tests” and “normal school asssessments”; (b) the APA concluded that “[Children in Australia, journal where the article was published] cannot be considered a source upon which one should rely for understanding the state of scientific knowledge in this field, particularly when the results contradict those that have been repeatedly replicated in studies published in better known scientific journals.” (The latter dismissal sounds suspiciously to me like “We don’t like what this study says, and it contradicts what the other studies we like says, so we’re not going to give it serious consideration.”)
5. In regards to the children’s outcomes that were studied, 20 of the 59 studies examined gender-related outcomes, but there was “a dearth of peer-reviewed journal articles from which to form science-based conclusions in myriad areas of societal concern,” including integenerational poverty, serious criminality, incarceration, early childbearing, drug/alcohol abuse, suicide, etc. A 2002 review of the literature by Anderssen and colleagues captures the absence of these measures:
Emotional functioning was the most often studied outcome (12 studies), followed by sexual preference (nine studies), gender role behavior (eight studies), behavioral adjustment (seven studies), gender identity (six studies), and cognitive functioning (three studies).
Further, there was at least one book-length empirical study that did examine developmental measures of concern to society — again, this time by Sarantakos, and this time published by Harvard Press (in 2000). There, Sarantakos concluded the following:
“If we perceive deviance in a general sense, to include excessive drinking, drug use, truancy, sexual deviance, and criminal offenses, and if we rely on statements made by adult children (over 18 years of age)…[then] children of homosexual parents report deviance in higher proportions than children of (married or cohabiting) heterosexual couples.”
For whatever reason, the APA chose to ignore Sarantakos’s study (maybe Harvard Press isn’t prestigious enough?).
6. None of the studies cited by the APA track long-term outcomes of children into adulthood — which, as Marks points out, is important because it is possible that, as Judy Wallerstein and colleagues found with children from divorced families, it is possible that “the ‘major impact’ of same-sex parenting ‘might not occur during childhood or adolescence…[but that it will rise] in adulthood as serious romantic relationships move center stage.”
7. His seventh point is highly technical and drawn-out, and if you want the full import of this point, I’ll direct you to the article itself. For here, I’ll quote Marks, who addresses this question:
Have the studies in this area committed the type II error and prematurely concluded that heterosexual couples and gay and lesbian couples produce parental outcomes with no differences?
Marks’ explanation:
In research, incorrectly concluding that there is no difference between groups when there is in fact a difference is referred to as a type II error….All one would have to do to come to a conclusion of ‘no difference’ is to conduct a study with a small sample and or/sufficient levels of random variation….It must be re-emphasized that a conclusion of ‘no significant difference’ means that it is unknown whether or not a difference exists on the variable(s) in question (Cohen, 1988). This conclusion does not necessarily mean that the two groups are the same on the variable being studied or on all other characteristics. This point is important with same-sex parenting research because … the 2005 APA Brief seem to draw inferences of sameness based on the observation that gay and lesbian parents and heterosexual parents appear not to be statistically different from one another — thereby becoming vulnerable to a classic Type II error.
There is more, but as I said, you can look at it for yourself for the full story (see pp. 745-748).
In conclusion, Marks suggests that, based upon his examination,
To make a generalizable claim, representative, large-sample studies are needed–many of them…. Some opponents of same-sex parenting have made “egregious overstatements” disparaging gay and lesbian parents. Conversely, some same-sex parenting researchers seem to have contended for an “exceptionally clear” verdict of “no difference” between same-sex and heterosexual parents since 1992. However, a closer examination leads to the conclusion that strong, generalized assertions, including those made by the APA Brief, were not empirically warranted. As noted by Shiller (2007) in American Psychologist, ‘the line between science and advocacy appears blurred’ (p. 712).
If Marks’ critique is correct, then the following is true.
One cannot make generalized assertions, in the name of science, that children with same-sex parents fare worse than children with heterosexual parents. It may be true, and one may make reasonable arguments for it, but before one grounds a statement in science, we need many more large-sample studies.
Judge Walker’s following assertion in his Prop 8 ruling — based in part on a 2008 APA statement – was just that: an assertion that has no basis in scientific evidence:
Children raised by gay or lesbian parents are as likely as children raised by heterosexual parents to be healthy,successful and well-adjusted. The research supporting this conclusion is accepted beyond serious debate in the field of developmental psychology. [emphasis mine]
It may be true, and one may make reasonable arguments for it, but one cannot appeal to supposedly indisputed scientific research to say it is true.
Also, same-sex marriage proponents, like John Corvino in Debating Same-Sex Marriage, who appeals to a 2004 APA statement as evidence that ”mainstream professional opinion resoundingly supports the conclusion that, on average, children in same-sex households fare as well as children in heterosexual households,” must revisit their appeal to authority. Again, it may be true and one may make reasonable arguments for it, but one cannot appeal to science to say it is true. For if Marks is correct, the experts at the APA drew hasty conclusions unwarranted by the scientific literature.
Finally, if Marks’s critique is correct, it means that the APA — to whom the general population trusts is making decisions based on science, and not on their own individual biases — needs to explain itself. In the meantime, people like me are left wondering if the APA is masquerading its own bias in the name of science.
[Note: The link to Marks' paper currently says the site is experiencing technical difficulties, and is unavailable for the time being. But yesterday it was publicly accessible, so hopefully the problem will be resolved soon and we all can read to our heart's content.]
Categories: The Future of Parenthood
So, it seems worth noting three things.
1. This article, published in 2012, is about an APA policy brief written in 2005 and the studies it relied on. Is anyone else bothered by the fact that this appears to omit the last seven years of social science on this topic, which includes, e.g., a lot of research from the National Longitudinal Lesbian Family Study?
2. SSR is still down, but here is what the APA parenting brief actually says about the Sarantakos study (p. 6, n. 1):
(my emphasis)
3. Relatedly, your third take-away from Marks’ article misses the point. Same-sex couples raising children (as a group) are not comparable to married biological parents raising children (as a group), any more than married biological parents raising parents are comparable to the broad class of different-sex couples raising children. Social conservatives often complain that studies of same-sex parenting do not compare same-sex parents to married biological parents. But this is not an apples-to-apples comparison.
Let’s say Mark Regnerus is right and we can’t, at this point, get both rigorous random samples and good comparison groups. What do we do? Do we go with the overwhelming weight of the social science on this topic, which gets stronger with each year as more and more studies tell basically the same story, or do we instead rely on the unscientific intuitions of many people about the centrality of sex difference to parenting? Why don’t all these studies, at minimum, establish a presumption that same-sex couples should be treated equally with respect to adoption and parentage, until someone can come up with good evidence showing otherwise?
JHW -
Thanks for drawing my attention to the full APA bit about the Sarantakos study. I looked at Marks’ article again, and he doesn’t mention the fact these kids had experienced overt hostility from other parents, and that many of them came from divorced households. Sure seems like it would have been helpful if Marks would have mentioned that, as it certainly complicates any finding from that study that children with same-sex parents look different on the outcomes compared to bio, intact families.
So it seems that APA applied its due diligence to the Sarantakos study. What puzzles me is, why didn’t they apply the same rigor to the other 58 studies?
And why did they offer the sweeping conclusion that no differences exist, when they had so little evidence on important children’s outcomes? Look back at Table 1 in Marks’ study and look at the “outcome studied” column — I find very few studies that examined important objective child outcomes like integenerational poverty, serious criminality, incarceration, early childbearing, drug/alcohol abuse, suicide, etc
Also, according to Regnerus, whatever the merits of the National Longitudinal Lesbian Family Study, it employs a convenience sample, “recruited entirely by self-selection from announcements posted at ‘lesbian events, in women’s bookstores, and in lesbian newspapers’ in Boston, Washington, and San Francisco.”
According to Regnerus, Rosenfeld in a 2010 study was the first scholar to employ a large, random sample comparing outcomes from children in same-sex parents with those of heterosexual married parents. Importantly, he apparently limited the sample to include only children from households “exhibiting at least five years of co-residential stability,” and found that there were no significant differences between the two groups when it comes to assessing kids’ progress through primary school.
It appears to me that what we need before we can call anyting settled science is more random studies like Rosenfeld that specifically compare stable gay families and intact, hetero families, and which look at a variety of important child outcomes.
For the record, the Rosenfield study has numerous flaws in it. Thus, a paper soon to be published (i.e. in 2012) http://www.ruthinstitute.org/ITAF12/faculty/readings/SSParentingHJLPP.pdf observes:
“….Unfortunately, the Rosenfeld study is seriously flawed. First, his results are misreported. What he actually finds is that child performance in same-sex households is not statistically different from any type of household — including ones that are known to be unhealthy for children. Second, the reason why his study lacks the precision to find any statistical effect is because he decided to drop 680,425 observations from his sample (45% of the data) in order to eliminate children who had not lived in the same location for five years. This turns out to introduce a strong bias in the results since this variable is strongly correlated with grade failure. Price, Pakaluk, and Allen (forthcoming) show that once the sample is restored and location is properly controlled for through the use
of dummy variables, children from same-sex households are 35% more likely to fail a grade compared to children in opposite-sex two parent families…The point is that a proper control of bias can lead to a dramatic reversal of an empirical finding….”
See also footnote 15.
Steven: Let’s move through the problems with that excerpt, working with the actual text of the Rosenfield study.
1. Rosenfield uses two models to compare children in same-sex couple households to children in different-sex married households. First, he does a standard regression analysis, and he finds that with controls, the effect of being in a same-sex couple household, relative to being in a different-sex married household, is slightly negative but not statistically significant. (See Table 3 and accompanying discussion.) Second, he compares the children raised by same-sex couples to children raised by different-sex married couples who are similar to the same-sex couples in the control characteristics, and he actually finds that with the most extensive set of controls, in this comparison same-sex couple households do slightly better, though this result is also not statistically significant. (See Table 4 and accompanying discussion.)
Douglas Allen claims Rosenfield finds that “child performance in same-sex households is not statistically different from any type of household.” Perhaps (in the paper, Rosenfield himself only undertakes this analysis for children raised by different-sex cohabitating couples.) But this strikes me as rather burying the point. The constant refrain of opponents of same-sex parenting has been that if only we had a large, representative sample, we would be able to detect that same-sex couple households, like other households that are not married different-sex couple households, do worse. Here we have a large, representative sample, and same-sex couple households are not shown to do worse.
2. How does Allen propose to “fix” the data? By getting rid of the restriction that guarantees that the children in question were actually raised by same-sex couples for a long period of time. Rosenfield did not invent the five-year rule out of a biased intent to make same-sex couples look better. Rather, he wanted to ensure that the family structure categories he constructed were actually meaningful. Here’s what he says on the subject:
So, Allen tells us, if you include lots of children in the “same-sex couple” category who may not have been raised by same-sex couples for a long period of time, and who indeed may have had a recent disruption in their household circumstances, you find that children in this newly-constructed category are 35% more likely to fail a grade (he doesn’t say whether or not this is statistically significant, but I assume he mentions it because it is.) So?
It’s worth mentioning that even Rosenfield’s method doesn’t fully correct for this factor. It’s likely that the children in the same-sex couple group in his sample have had more instability, in undetectable ways, than the children in his different-sex married group, simply because the ways in which same-sex couples end up raising children are more likely to involve instability (typically, the dissolution of a different-sex relationship); as long as this happened far enough in the past that there could have been five years of the same-sex couple living together with the child, Rosenfield doesn’t account for this. He finds that there is no statistically significant difference despite that fact.
My point was simply there is no “clear” answer; feel free to forward your comments to Doug.
JHW,
Two points.
1. In table 4 of Rosenfeld, the smaller number is better, so in model 5, same-sex couples still do slightly worse (indeed, it is a restatement of table 3)
2. What if Rosenfeld’s study included in the “most extensive group of controls” a dummy variable for stability? would that be a balance between ignoring and acknowledging the issue?
I think what matters most is that the child has two parents that love him/her and provide a nurturing environment for the child. Does it really matter if they are same sex? Encouraging good study habits is vital and it all begins with a study area for the kid. http://www.squidoo.com/study-table-for-kids
The American Psychiatric Association (APA) is run by a homosexual activist caucus called the AGLP…
http://en.wikipedia.org/wiki/Association_of_Gay_and_Lesbian_Psychiatrists
The American Academy of Pediatrics gets all their info about homosexuality from the APA. See here…
http://www.clgs.org/official-statement-concerning-homosexuality-american-academy-pediatrics
Junk science begets junk science
Mainstream psychiatry and psychology has been thoroughly compromised by homosexual activists and sympathizing ideologues. “Homosexuality” was removed from the DSM list of disorders in 1973 as a result of to homosexual activist coercion, bullying, threats in 1973, and not based on any scientifically objective decisions. Then at the 1978 annual APA (American Psychiatric Association) meeting, a formal structure was adopted for the Gay, Lesbian, and Bisexual Caucus of the American Psychiatric Association, which is now named the Association of Gay and Lesbian Psychiatrists (AGLP). In 1978, an official APA Task Force on Gay, Lesbian, and Bisexual Issues was established, which was given formal status as a standing committee in 1981. The following year the Assembly, APA’s legislative branch, approved adding elected representatives from a group initially designated as the Caucus of Homosexually Identified Psychiatrists, and later renamed the Caucus of Gay, Lesbian and Bisexual psychiatrists. This activist caucus is still active in the APA Assembly.
http://www.aglp.org/pages/chistory.html
Since its founding, AGLP with a membership of over 600 psychiatrists, has been active in helping to shape the dramatic conceptual shift in the cultural understanding and significance of homosexual behavior within psychiatry and within society. With the support of AGLP members, the APA has issued Position Statements supporting same sex unions and the adoption and co-parenting of children by same sex couples (http://www.care2.com/causes/apa-unanimously-supports-gay-marriage.html), as well as a position statement opposing “Reparative Therapy” as unethical.” (http://www.citizenlink.com/2011/07/27/therapy-for-unwanted-homosexuality-%E2%80%93-part-2/) Nothing objectively scientific or neutral here folks. Any APA accredited therapist who specializes in issue of “gender” and “sexuality” are nothing but ideologues of the homosexual activist bias.
http://www.aglp.org/pages/chistory.html
APA Psychiatric Divisions are also homosexual activists groups…
http://www.apadivision44.org/
Proof that American Academy of Pediatrics gets all their info about homosexuality from the APA:
http://www.clgs.org/official-statement-concerning-homosexuality-american-academy-pediatrics
Judge Walker’s following assertion in his Prop 8 ruling — based in part on a 2008 APA statement – was just that: an assertion that has no basis in scientific evidence
Judge Walker’s assertion was based entirely, not “in part”, on the evidence in the record, not on an independent review of social science. That record was developed by the evidence and arguments of both sides. If you have read his entire opinion, David, you’d know that this was not a decision that turned around a judge being fooled by what is in your opinion a misstatement of the state of social science.
Jason,
“Mainstream psychiatry and psychology has been thoroughly compromised by homosexual activists and sympathizing ideologues. “Homosexuality” was removed from the DSM list of disorders in 1973 as a result of to homosexual activist coercion [blah blah blah].”
Ah, but why was homosexuality put into the DSM in the first place?
fannie writes:
Historian Margot Canaday, in The Straight State makes a fairly compelling argument, grounded in the historical evidence, that the DSM was altered to include homosexuality as a mental disorder under pressure by the military to have a method for screening out recruits whom they suspected of being gay.
Thanks Anna.
I always find it telling when people whinge about homosexuality being removed from the DSM due to purported gay pressure and “bullying,” as opposed to acknowledging that homosexuality isn’t, actually, a mental disorder and that it’s placement in the DSM was an error to begin with.
[...] to be disadvantaged in any significant respect relative to children of heterosexual parents.” Here’s what the district court claimed in the Proposition 8 case: Children raised by gay or lesbian parents are as likely as children raised by heterosexual parents [...]