Disparate impact of results is NOT proof of racism. People are complicated - and some racial correlations are perfectly valid - at least until enough members of a racial group decide to change them...