With traditional measures of children's racial preference multiple requests, all presumed to measure racial preference, are made. However, children change their racial preference across the requests. Two interpretations of these changes are offered. One, suggested by Piaget, views these changes as self-contradictory and due to cognitive incapacity. The second is that children intend to express a liking for dolls of both races. The data support this latter interpretation suggesting the use of the doll technique as a measure of racial ethnocentrism.