Aestu wrote:
In the hundreds or thousands, across a broad variety of social and national groups.
I was asking about significance level, not sample size, but OK. The results are already statistically significant at a reasonable level. Larger sample sizes are always nice, but they don't seem necessary here.
Quote:
Since the thesis is, after all, connected to human physiology, it stands to reason it can't be adequately corroborated unless it is tested in many different populations.
This is a fair point, although I suspect there wouldn't be strong inter-population differences in this trait (unlike those that are subject to natural selection based on the local environment). The results are still valid for the population they were conducted on.
Quote:
The problem with such a small study is not only the potential for a single, or a very few, participants to massively skew the overall probability, it is also that many small studies give rise to the probability of a single study getting outrageous, headline-grabbing results.
If you look at the graph in the original article, it's clear that there's a lot of variation, but the observed effect isn't being caused by outliers.
Quote:
Which is, of course, for undistinguished researchers looking for ten minutes of fame, writing about about OMFG GAY PPL, the point. The very choice of topic implies they are looking for headlines; there's substantially more interesting and relevant issues to look into, many of which are also much more conducive to scientific methodology.
It's bad science. This sort of really bad approach is a major factor in why we don't have decisive answers to many very interesting and controversial questions.
Repeatedly "exploring" the issue with repeated, uselessly small studies is like endless Mulligans.
I can see why some people don't appreciate this type of study, but I'm interested in the statistical criticism you originally had. What did you mean by " a margin of about one individual"?
