"DON'T KNOW" BOXES IN FACTUAL QUESTIONS IN A MAIL QUESTIONNAIRE: EFFECTS ON LEVEL AND QUALITY OF RESPONSE

This article summarizes the findings of an experiment to evaluatethe effect of "Don't Know" (DK) boxes in a mail questionnaireconsisting of factual questions. In a split-panel test, halfof the respondents received a questionnaire with DK boxes andthe other half received a questionnaire without DK boxes. Theoverall response rates were the same for the two versions. Thequestionnaire version without DK boxes obtained an appreciablyhigher rate of substantive responses for many items. However,there were few large substantive changes in response distributionsbetween the two versions, and there was no appreciable differencein the response error rates.