> When you discover that some "race" is underrepresented in some field, it's completely possible that a major cause is that the median in that arbitrarily defined racial category is below the median in another arbitrarily defined racial category.

It's possible, yes. It's also possible that perceptions of race influence the subjective decisions that drive those outcomes. How would you determine causality?

Eg, you could try sending employers identical resumes with "black sounding" and "white sounding" names and measuring callback rates. This has been done: https://www.politifact.com/punditfact/statements/2015/mar/15...).

There are likely many factors in racial disparities, but this appears to be one of them.