Response Effects in Computer-Administered Questioning

Responses to questions administered by computer, personal interview, and self-completion methods were assessed in a between-subjects experimental design. For yes/no and pick-one question types, responses were similar for all three methods of questioning. For multiresponse knowledge questions, subjects in the computer-administered treatment group picked more answers, which were more often incorrect. Responses to integer-scale questions in the computer-administered treatment group had slightly higher means and lower dispersion, but these effects are not significant.