Contested claims
Every claim on this site that is disputed, limited, or unresolved — collected in one place, on purpose.
This page exists because an argument collection that hides the strongest objections to its own evidence is propaganda. Each module is required to declare its weak points before it can be published. They are gathered here.
4 declared across 1 module.
The Regnerus study on same-sex parenting
Regnerus (2012) reported that grown children of parents who had been in a same-sex relationship did worse on many measures. The flaw: his 'same-sex parents' group was mostly people whose parent had at some point been in a same-sex relationship — typically after a straight marriage collapsed — compared against intact, never-divorced families. He was comparing family breakdown with family stability and attributing the difference to orientation. But the paper was NOT retracted and it was NOT fraud, and saying so to someone who knows the details hands them the argument.
What survives anyway: Reanalyses of his own data (Cheng & Powell 2015; Rosenfeld 2015) find the effect largely disappears once family instability is accounted for. Whole-population Dutch registry data (Mazrekaj 2020) finds no deficit at all.
Conversion therapy causes suicidality
Blosnich et al. (2020) found roughly double the odds of lifetime suicidal ideation among people exposed to conversion therapy. Sullins (2022) objected that because the measure was LIFETIME suicidality, some of it may have predated the therapy — and that restricting to what happened afterwards reverses the association. Blosnich et al. published a reply disputing this. The dispute is not resolved.
What survives anyway: Drop the suicide claim entirely and nothing important is lost. It remains a treatment that does not work, for a condition that is not a disease, delivered mostly by people with no medical training — and its most famous supporting study was withdrawn by its own author (Spitzer 2012).
Marriage equality reduced adolescent suicide attempts
Raifman et al. (2017) is a difference-in-differences design, not an experiment. States that legalised early were already different — more urban, more liberal, possibly already improving. Some of the drop may belong to those other differences. Critics said exactly this in print, and they were not wrong to.
What survives anyway: It remains strong evidence of a link, and the effect was concentrated among sexual-minority students, exactly as the stigma model predicts. Conceding the limit costs nothing and makes everything else more credible.
Religious liberty versus anti-discrimination law
The baker who won't make the cake. The church hall. The religious adoption agency. Evidence cannot settle these. Reasonable, decent people disagree — and this is the strongest ground the opposing side has left. It should not be waved away as obvious.