The inmates were randomly assigned to three response set groups: fake-aggressive group, a fake-nonaggressive group, and a standard-instruction control group. The MMPIs were scored for five obvious and five subtle aggression or hostility research scales. The BDHI is a totally obvious test, whereas the DAP is a very subtle measure. A paradoxical relationship between response set and subtlety of the scales was hypothesized, such that the obvious scales could be successfully faked, but the subtle scales would show scores in the opposite from the intended direction. It was found that the inmates did correctly manipulate the obvious scales, but the subtle scales did not consistently show the hypothesized trend. 24 references, 4 tables, and 1 figure. (Author abstract modified)