This is a group-level inference, there are probably a considerable number of individual exceptions. So it's perfectly reasonable that the average results don't

This is a group-level inference, there are probably a considerable number of individual exceptions. So it's perfectly reasonable that the average results don't apply to every person. That alone doesn't negate the research - but I haven't read it in enough depth to say if there are bigger problems.