Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

John Northup


Several studies have examined the relationship between self-report of behavior and actual behavior (Baer, Williams, Osnes, & Stokes, 1985; Bernstein & Michael, 1990; Risley & Hart, 1968). Results from such studies indicate poor correspondence between verbal self-report and actual behavior. This finding is particularly true for children (Guevremont, Osnes, & Stokes, 1986). However, few studies have examined conditions under which the accuracy of self-report procedures can be increased. The current study evaluated whether the accuracy of children's verbal self-report of reinforcers could be increased by modifying antecedent conditions. Specifically, the current study examined the effects of two antecedent variables (i.e., mode of preference assessment and type of instruction) on the accuracy of verbal identification of reinforcers. Participants were thirty children in grades one through four, who had been referred to an outpatient pediatric behavioral health clinic for non-compliance. Participants were administered two preference assessments: (a) survey and (b) stimulus paired-choice questionnaire with three different types of instructions (contingency unspecified, vague contingency specified, and exact contingency specified). Subsequently, participants were allowed to choose actual reinforcers for completing easy math problems (token coupons). The coupons that were chosen represented the true criterion for comparisons with the results of the verbal self-reports. A main effect was found for type of instruction and mode of preference assessment. An interaction was also found such that the unspecified survey resulted in significantly less accurate responses than any other group (i.e., vague survey, exact survey, unspecified paired-choice questionnaire, vague paired-choice questionnaire, and exact paired-choice questionnaire). Results indicated that accuracy was higher when contingencies that were either vague or exact were specified in the instructions, and accuracy was higher for the paired-choice method than the survey. Overall, accuracy of the survey was increased when instructions specified vague or exact contingencies. The accuracy of the paired-choice method of assessment did not vary with instructions. Finally, when vague or exact contingencies were specified in the instructions, responses on the survey were comparable to those of the paired-choice.