Of course! When studies compare 2 distributions and see a significant difference in a summary statistic, that says nothing of the distributions or their overlap. The distributions themselves are rarely Gaussian and usually have significant long-tailed behavior. For example, a study may find group A is better at math than group B. A lay-person, usually a journalist, will take it to mean ALL group A's are better at math than ALL group B's, when that's clearly wrong. Because of the usual long-tailed behavior of these kinds of metrics, discriminating against group B because of the study will be inefficient.