zlacker

>>geox+(OP)
> odds were tripled for autism (OR = 3.1; 95% CI: 1.02, 9.7)

This is scientific malpractice! The most ridiculous confidence interval I've ever seen! 1.02 to 9.7, reported as "tripled", seriously? And of course the data is non-blinded, self-reported survey responses recalling events that occurred many years ago, and the analysis is not preregistered and splits the cohort in an arbitrary way to eke out so-called "statistical significance" (by the slimmest imaginable margin, 1.02 > 1.00, just barely).

How can this dreck be published? Everyone involved should be sanctioned. And everyone who took this headline at face value should seriously reconsider their approach to consuming science news.

>>modele+fd
With an 95% confidence interval which doesn’t include zero, doesn’t it mean that it’s statistically significant?

Assuming data is valid and unbiased of course.

Not a statistician, just curious.

>>jtaft+Sk
Nope, you're thinking of regression coefficients, where you'd be correct that usually the null hypothesis is $\beta = 0$. In this case, what's being reported are odds ratios, so the null hypothesis would be that OR = 1.

The parent comment's point is that although the reported effect is significant at $\alpha = 0.05$ (the usual "95% CI" you mentioned), there are other problems that render their test of this hypothesis less than valid.