If you measure some quantity using biased sampling, you're invariably going to see biased results. The assumption that the biased results in this scenario are due to lack of opportunity, racism, etc are unfounded. Correlation is not causation.