I think the point is that there's a difference between 'obvious statistical evidence' and 'non-obvious statistical evidence.' Often the results of statistics can be counterintuitive due to issues like sample size, confidence intervals, etc.