TIME magazine also had a recent piece hailing the lesbian “mother” as the parental ideal. The study that was the basis for this glowing report, as well as the one cited in a recent article in The Atlantic, involve statistically insignificant samples and highly questionable methods of evaluation.  The study in Time, which first appeared in Pediatrics, the official journal of the American Academy of Pediatrics, looked at 84 “families.” Its findings were based on interviews with the mothers, their lesbian partners and the children, including an “online questionnaire” addressed to the children at age 17. Alice Park writes:

Data on such families are sparse, but they are important for establishing whether a child’s environment in a home with same-sex parents would be any more or less nurturing than one with a heterosexual couple.

In other words, there is not reliable data to draw any conclusions but we are going to draw some anyway, and use them to influence public policy.

One of the authors of the study, Nanette Gartrell, openly admits her bias. Park writes:

“We simply expected to find no difference in psychological adjustment between adolescents reared in lesbian families and the normative sample of age-matched controls,” says Gartrell. “I was surprised to find that on some measures we found higher levels of [psychological] competency and lower levels of behavioral problems. It wasn’t something I anticipated.”

As one reader who sent the above link to the study stated, the document virtually “advertises its incompetence and speciousness on the very first page.” It states the following:

“Results: According to their mothers’ reports, the 17-year-old daughters and sons of lesbian mothers were rated significantly higher in social, school/academic, and total competence and significantly lower in social problems, rule-breaking, aggressive, and externalizing problem behavior than their age-matched counterparts in Achenbach’s normative sample of American youth.”

Lesbian mothers are the source for the study’s conclusion that lesbians make both great mothers and great fathers.

                                                — Comments —

J. Lancaster writes:

While I’m certain that any conceptual response I make to your post about lesbian mothers and research done on them will meet with complete disdain, I would like to ask how you can justify claiming that this study is “statistically insignificant”? Are you aware of the connotations of that term? As a statistician in developmental psychology, I can assure you that the measures used in developmental research to measure significance take into account the numbers involved–that’s the whole point. Whether or not you agree with the interpretation of the data, calling it “statistically insignificant” is not only erroneous, it’s completely incomprehensible from an intellectual point of view.

Laura writes:

I hope I don’t meet any arguments made in the spirit of open debate “with disdain,” but judge them on their merits.

Let me ask you this. Given the magnitude of the findings, the conclusion that a child does not need a father, a conclusion that defies not just common sense but most of human psychology, which recognizes the child’s interactions with both father and mother as fundamental to the inner drama of development, do you think 84 families is statistically significant? How so?

Sage McLauglin writes:

J. Lancaster is playing a semantic game with you and trying to bully you intellectually over tertiary issues. You used the term statistically insignificant to describe the sample size when what you were driving at was unrepresentative. He (or she) knows exactly what you meant, and did not directly address that question, instead preferring to take a smarmy and dismissive tone, while preemptively accusing you of the same. Anyone who reads your blog knows that you do not treat your readers with contempt. Lancaster appears to be motivated by a knee-jerk impulse to defend the study because he likes its conclusions, and a similar impulse to make accusations of bad faith against you, because he does not like your conclusions. This is not the attitude of a dispassionate researcher, and hiding behind the title of “statistician” can’t obscure the prejudicial attitude with which he is approaching this study and your discussion of it.

And for the record, the word Lancaster is looking for is denotation, which indicates the precise definition of a term, not connotation, which means the implied sense or nuance of a term. Or isn’t Lancaster aware of the denotation of the word connotation? As a qualitative researcher who deals in words for a living, perhaps he can meekly accept my admonishment, since it’s being offered with such dispassion and an obvious generosity of spirit.

Seriously, though, now that we’re done substituting posturing for arguments and bullying for debate, let’s hear Lancaster defend the study on the merits.

Laura writes:

Yes, I meant the numbers were not representative.

Lancaster is a woman, by the way, as are both of the journalists who reported these “studies” in national magazines, and I am eager to hear her response in good faith.

Jesse Powell writes:

On the issue of “statistical significance,” I think I can say a few things. Now I will admit the idea of a “statistically insignificant sample” doesn’t make much sense. A sample of 10 can be statistically significant depending on what you are claiming. 

If you take an object shaped like a coin, where the weight of the coin is unevenly distributed, and you want to do a study to determine how often the coin ends up “heads,” you can then flip the coin 10 times, for a sample size of 10. If the coin ends up being “heads” all 10 times then you have “statistically significant” evidence, with a “p-ratio” of less than 1%, that the coin has a true probability of coming up heads 63% of the time or greater. If you flip the coin 50 times, a sample size of 50, and the coin comes up heads all 50 times, then you have statistically significant evidence, with a 1% p-ratio, that the coin comes up heads 91% of the time or greater. 

The reason why larger sample sizes are better than smaller sample sizes is because the margin of error, the range of the confidence interval, is smaller with large sample sizes. A larger sample size allows you to detect smaller differences in the underlying data. 

In the actual description of the study provided in the link, the term “statistically significant” is never used. The word “significantly” is uttered but we don’t know if this is merely an impressionistic judgment or if some kind of statistical test was done showing some result. We don’t know what the null hypothesis was, we don’t know what the p-ratio used was, we aren’t given the underlying data, and the phrase “statistically significant” is never used regardless. 

In the conclusion offered in the study it is asserted that “Adolescents who have been reared in lesbian-mother families since birth demonstrate healthy psychological adjustment.” No evidence is offered to support this conclusion other than people’s assertions and impressions. If there is a statistical basis for this claim the data underlying this assertion is not given or even described except in the most vague and broad terms.

Laura writes:

One interesting aspect of the articles in which these studies appeared is that they offered no contrary investigations. Yet studies that offer opposing conclusions indeed exist. According to a  study that examined the life experiences of two million adults in Denmark, where same-sex marriage has been legal since 1989, the incidence of homosexuality was significantly higher in men raised in a home without a father. The study is discussed here.

The psychologist Trayce Hansen reviews studies that show children raised by openly homsexual parents are four to ten times more likely to engage in homosexual behavior themselves as adults here.

Returning to the study in Pediatrics that was discussed in Time, I am not a statistician, but questionnaires asking 17-year-olds, who are still economically and emotionally dependent on their parents and guardians, about their psychological health and the parenting they have received seem to be of limited value. I realize many of the questions would have addressed the general mental states of the teens, whether perhaps they had suicidal thoughts or difficulty making friends. But how could these questionnaires probe one of the most important issues: whether these teens would ultimately be able to form real families of their own? Given the health risks of homosexuality, the remarkably shorter life span and incidence of psychological maladjustment among adult male homosexuals, it is an important question even if one does not regard homosexuality as immoral. Furthermore, how could these questionnaires have measured any buried desire for contact with one’s biological father?

Miss S. writes:

One of the issues that the article didn’t take into account is any factors besides gender. What were the economic levels of the families for example? Did they look at the gender of the children? If the gender of the parents is that important so must the gender of the children for constancy. Also how did they select the sample? Self-selection may play a big role in these things and if it was online that is a red flag.

Please follow and like us: