The analysis of racial differences has been a popular undertaking for social scientists, perhaps because it seems to be easy to do and it yields interesting differences. This study raises the possibility that the frequent finding of higher self-esteem scores among blacks compared with whites may be attributable, at least in part, to black-white differences in response styles. Blacks are more likely than whites to use the extreme response categories in Likert-type questionnaire items. This general tendency has important implications for black-white comparisons along self-esteem dimensions. Analyses of six large-scale nationally representative surveys of high school students revealed that blacks score significantly higher than whites when the full scale range is used in computing self-esteem scores, but the black-white discrepancy disappears when a truncated scoring method is employed to control differences in use of extreme response categories. Any firm conclusions about racial differences in self-esteem lie beyond reach with current data. Race differences in general response styles, and in responses to particular self-esteem items, leave any specific self-esteem measure and scoring open to dispute. (Author/ABL)