Someone doesn’t know how science is actually done. It’s a series of surveys over 35 years using sample sizes between 1,500 people and 18,693 individuals, from the limited methodology publications I read through in the last 15 minutes. That’s far more than is required to be considered trustworthy, reliable, and accurate as long as their other methods were sound, which they appear to be.
The study uses binary and ordinal logistic regression models that appear to be highly reliable with large sample sizes, and their accuracy appear reliable based on the study meeting standard core statistical assumptions and they seem to avoid rare-outcome traps.
This is a sound study, and you should probably look into statistical modeling in producing such studies before using language you’ve likely read in comments sections that like to dismiss studies because they don’t like the results or findings for political or personal reasons.
Their methodology: