A Methodological Study of the Torrance Tests of Creativity

The sensitivity of creative thinking as measured by the Torrance Tests with regard to an experimenter‐induced response set was investigated. The 198 subjects were divided into four groups. Before performing the Unusual Uses Activity (Verbal Form A), each group was given a unique set of instructions. Group I received standard instructions and acted as a control. Other treatments were varied in terms of the types of responses that were encouraged. Group II was encouraged to be “practical and reasonable”; Group III was encouraged to list “as many ideas” as possible; and Group IV was encouraged to include all “unusual, weird, or illogical” ideas. Discriminant analysis yeilded two significant functions, suggesting that univariate analysis of the Torrance scales of Fluency. Flexibility, and Originality can be misleading. A multivariate Dunnett test showed that Torrance Tests were highly sensitive to brief experimenter manipulation. A very modest induced attitude shift resulted in a shift of up to approximately a standard deviation in the creativity scores.