Fall 2012

Postscriptum: BAAF vs Regnerus on Gay Parenting

Paul Sullins Download Article

The Catholic Church, like many traditional religions and cultures, recognizes that “the absence of sexual complementarity in [homosexual] unions creates obstacles in the normal development of children.”[1] In the past twenty years advocates for gay adoption have produced a spate of “research” studies which claim, on the basis of scientific evidence, to demonstrate the opposite, i.e., that children in same-sex households fare no differently on developmental outcomes than those in heterosexual ones.

To date, these gay parenting studies have lacked one or both of two essential features of method which are necessary for their findings to be dispositive on the question: a sufficiently large random sample and information on child outcomes. The March 4, 2013 study by the University of Cambridge’s Centre for Family Research for the BAAF (British Association for Adoption and Fostering) exemplifies both deficiencies in arriving at its conclusion that children in gay and lesbian families suffered no significant disadvantages compared to those in heterosexual families.

In contrast, the recent (2012) study by Mark Regnerus of the University of Texas, specifically designed to address both methodological problems, is the first to demonstrate, although not without limitations, that children raised by gays or lesbians are subject to a range of negative social, emotional, and relational outcomes.

The BAAF study compared 131 adoptive families: 40 gay male, 40 lesbian, and 41 heterosexual.

Statistical research samples estimate the presence of differences in the population they represent according to a strict calculation of probabilities; the larger the sample, the more precise the estimate. In a typical telephone survey of about 1,000 cases, differences between two categories (say, those for or against a particular policy proposal or political candidate) greater than about 6 percentage points probably represent true differences in the underlying population; this is commonly expressed as a “margin of error” of plus or minus 3 percentage points for each of the data points. With only 131 cases, the BAAF study has an ideal margin of error of 8.6 percentage points, meaning it can at best only distinguish differences greater than 17 percentage points. (This limitation, by the way, is never reported in the study.) Because it compares three groups, rather than the usual two, its precision is reduced even further.

Consequently, the BAAF study reports findings of “no difference” in the population even when there are substantial differences in the sample. For example, the authors report, “No significant differences between family types were found for any of the subscales” of a psychological assessment, even though the subscales differed by 10 to 18 percentage points in the sample. On feelings about contact with the birth family (“open adoption”), sixty percent of heterosexual parents reported positive feelings about this, compared to only forty percent of lesbian mothers. The conclusion? “No difference in feelings about contact was found among family types.” Eighty percent of heterosexual fathers, but only 59% of lesbian non-primary mothers, were satisfied with the household division of labor for childcare; this means, we are told, that “no differences were found.”

By contrast, Regnerus collected a sample over 20 times as large as the BAAF study – 2,988 cases – which, with a margin of error of only 1.8 percentage points, could distinguish differences as small as only 3.6 percent, though most of the differences he found were substantially larger than that. Because his sample was capable of finding differences with precision, if Regnerus had found no differences it would have been a highly meaningful finding. But because the BAAF sample was not statistically able to “see” even large differences, the fact that it repeatedly found no differences means very little.

The second weakness of the BAAF study, and corresponding strength of Regnerus’ study, is in the use of outcome measures.

Social scientists have long understood that child development can only be definitively assessed by looking at outcomes, that is, how well children thrive as adults once they leave the family on their own. What matters most for wellbeing and fulfillment is not how children – or parents – feel about things at the moment, though that is certainly important in other ways, but how well the onetime children function as adults. And the former is a poor predictor of the latter. Children denied primary education, for example, seldom are unhappy about it at the time, but find themselves at a great disadvantage as adults compared to persons who had the opportunity to learn to read and write. It doesn’t take a trained social scientist to recognize that conclusions about child development based on interviews with small children are simply premature. Many problems, if they are there, will not show up until much later.

Here the difference between the two studies is stark: the BAAF study considers no adult or even teen outcomes at all, whereas Regnerus examines forty adult outcome measures that have been validated by long use in other studies of child development. The children in the BAAF study average only six years of age; none is older than nine. Regnerus gathered data from adults about their entire childhood and current success in life.

The BAAF sample and interview procedures, moreover, were highly biased to exclude problem reports. Unlike Regnerus’ study, the BAAF sample was not random; it was, in fact, recruited through adoption agencies, the same organizations that collectively commissioned the study. Some gays and lesbians were also recruited through gay and lesbian parenting groups. In other research settings such relationships have so consistently led to biased results that they raise ethical questions, as when, for example, studies on the health effects of smoking funded by tobacco companies consistently found little or no health damages to smokers.

Obviously, it is not in the interest of BAAF or its member adoption agencies to report problem placements, so it’s unlikely they would recruit difficult cases into the study. Members of gay and lesbian parenting groups are clearly already supporters of gay and lesbian adoptions, and those who are most interested in encouraging more gays and lesbians to adopt are also more likely to volunteer to participate in the study. This is called “ascertainment bias”; it is a typical problem of studies of gays and lesbians, when samples are recruited from GLBT advocacy and interest organizations, often by advertisements or appeals to help demonstrate the desired results of the study, and has been shown to lead to not only biased, but completely inaccurate, results. Not surprisingly, the study does not report any details on how the sample was recruited. Regarding canonical measures of sample quality, it tells us only that calculating a response rate is “not possible” and that “working out the representativeness of the sample is problematic.”

The study report is published by BAAF, in a glossy booklet – with pictures on the cover of drawings of families having fun with their children, and an effusive foreword – with the express intent to encourage more adoptions by gays and lesbians. At this point we must recognize that the purpose of the BAAF study is not to advance scientific knowledge but the goals of the BAAF. It is, significantly, not a scholarly peer-reviewed study, nor could it pass peer review as published. It is a piece of public relations under the guise of scientific objectivity.

Regnerus’ study is not without its flaws and limitations, but, unlike the BAAF study, it is a serious attempt to address an important question of social science that results in clear, defensible empirical findings. The BAAF study presents no serious evidence whatever to challenge to Regnerus’ findings. This, of course, will not prevent the BAAF study from being celebrated, as it has not prevented Regnerus’ study from being excoriated, by cultural elites for whom the acceptance of homosexuality as normal, and thus the innocuousness of same-sex parenting, is an article of faith.

Although said cultural elites do not yet know it, Regnerus’ study is really the wave of the future, and signals that the high-water mark of approval for gay parenting is near, or may have already been reached. I say this because, regarding the effect of marriage innovations on the wellbeing of children, we have been down this cultural cul-de-sac before. In the 1970s, rising divorce rates were widely celebrated among academic elites as a new era of relational freedom. Studies at that time of young children in post-divorce households, of which today’s gay parenting studies are eerily reminiscent, purported to show that they were faring no worse than comparable children in stable married households.

It wasn’t until scholars could examine the actual empirical outcomes of children of divorce well into their adult lives, first through retrospective data through the 80s and 90s, and then emphatically in Judith Wallerstein’s 25-year longitudinal study, published in the book The Unexpected Legacy of Divorce,[2] that the full trauma and harm of parental divorce for children finally became starkly evident. Today, family scholars widely recognize that children of divorce, compared to those from stable marriages, are more likely to suffer reduced success and function throughout their lives. Slowly, even the most liberal scholarly proponents of divorce are advising increased restrictions on divorce and support for marriages, even (unthinkable even a decade ago) that parents consider staying together for the sake of the children.

We can have confidence that at some point, as we gain more experience with children raised in same-sex households, and as the first of them mature into adulthood in large enough numbers to examine them as a group more closely, the developmental obstacles they face, which are now hypotheses of religious faith and the natural law, will become fully, though perhaps slowly and grudgingly, recognized by those who are amenable to empirical evidence which, in the end, cannot be denied. [Article copyright 2013 Paul Sullins. All rights reserved.]


[1] J. Ratzinger and A. Amato, Considerations Regarding Proposals to Give Legal Recognition to Unions Between Homosexual Persons (Vatican: Congregation for the Doctrine of the Faith, 2003), n. 7.

[2] Judith S. Wallerstein, The Unexpected Legacy of Divorce: A 25 Year Landmark Study, 1st edn (New York: Hyperion, 2000).