I can see that you care about this and want to get it right. So, you should understand that it is the percentage within each group that is being evaluated. Here's the way this works: Let's say for a simplified example that there are a million whites in a particular location, and 60 percent of them graduate from college. In the same location, there are half a million non-whites, and 20 percent of them graduate. The 20 percent is not as good as the 60 percent and (along with differences in long-term unemployment percentages within each group and income disparities within each group) does indicate racism. Please understand I didn't just make up this list; it is the result of studies, some of which have been going on for years and were conducted by researchers who really do know what they're doing and really were interested in getting it right.