How do you define mistake though? Part of the problem with a lot of these social surveys is that they have an agenda before writing the study, so they frame the surveys to fit their conclusion. To just lump all black faces as bad and all white faces as good means they will "show you have your racially biased" based on unfair criteria. Good social surveys have the biases as minimized as possible and will ask the same questions in different ways to try and identify consistent or inconsistent answers to weed out biases.