We almost always get a more accurate picture by digging a little deeper and looking not just at the averages but at the spread: not just the group all bundled together, but the individuals. Then we often see that apparently distinct groups are in fact very much overlapping.