The effects of sex, verbal ability, test anxiety, ideational fluency and training procedures on Formulating Hypotheses test performance were studied. Training consisted of presentation of models of "acceptable" responses that stressed either quantity or quality performance. Both the quantity and quality models were effective in modifying behavior in the expected direction. Ideational fluency was related to number of responses, and verbal ability was related to scores reflecting quality. Females were in general superior to males with respect to scores reflecting quantity of responses. Test anxiety was not significantly associated with performance. Weak evidence of treatment-anxiety and sex-vocabulary interactions was found. (Author)