Measures of (explicit) prejudice and discriminatory behavior usually show only modest correlations. On this background, I was wondering if anybody here can point to studies that explicitly estimate the proportion of extremely prejudiced people that engages in discriminatory behavior of any kind? I am more concerned with explicit measures of prejudice but would also appreciate if you name studies investigating this question only for implicit measures.
Example: Let's, for the sake of an example, define a person "extremely prejudiced" if she gives extreme answers to all (or a vast majority of) items of a scale measuring prejudice against a particular social group. I am interested in studies that have looked at the proportion of such extremely prejudiced people engaging in discriminatory behavior (e.g. preferring a white over a black job candidate).