-1

When the study "Invisible Victims: Delayed Onset Depression among Adults with Same-Sex Parents" was released, a rebuttal was soon posted. Sullens posted a counter which stated at the end:

I think I have addressed enough errors in Frank's critique to establish that his criticisms of my study are unfounded and that my findings are well justified. However, I doubt this will be convincing to him or those sharing his perspective, because what appears to disturb them is not the study methods but the findings. I suspect no evidence will convince Frank that children with same-sex parents may face unique and heightened struggles and difficulties. It is right to be appalled at that thought, but the most useful response is to try to understand the problem better, so as to address the conditions or provide support necessary to ameliorate the problem, not deny the evidence.

Looking at his study and justification, was there nothing wrong with his methodology? Or is he lying?

Jacob Blaustein
  • 107
  • 3
    Welcome to Cross Validated! Questions here need to be fairly self-contained - certainly they mustn't require reading external sources just to understand what the question is - & well focused - a review of all aspects of a particular study is too broad. If you have questions on specific methodological issues raised by these papers, you're of course welcome to ask them here - possibly over more than one post if they're not especially related. – Scortchi - Reinstate Monica Apr 20 '22 at 08:02

1 Answers1

1

From a very brief scan of the initial article (but not the response and rebuttal) it looks like the evidence is quite weak. First and most importantly, this analysis involves a sample of 20 adolescents. That's very small and far too small to support logistic regressions with multiple covariates. Minimum sizes for logistic regressions have been discussed extensively on this site. See here and here.

Second, the analysis here doesn't provide any causal evidence, only correlational evidence. Third, as I understand it, the data for this is taken from a much larger survey. Thus, it would have been entirely possible for Sullens to simply look for all possible correlations between the group of adolescents raised by same-sex parents and report correlations that were significant. This is a form of p-hacking though because as long as you look at enough possible correlation, by chance, some will be statistically significant.

num_39
  • 1,454