How Explicit “I Don’t Know” Options Affect Survey Responses: An International Methodological Experiment

Should researchers offer survey participants explicit options not to respond to certain questions? While some scholars provide such “I don’t know” (DK) choices out of fear to illicit “non-attitudes,” others refrain from doing so to limit “satisficing.” We conducted survey experiments on diverse samples of more than 32,000 respondents across ten countries worldwide in 2021. We find that explicit DK choices increase non-substantive responses as cognitive demands rise – with the largest increases for knowledge questions. Examining the latter more closely, we find that correct answers among all responses decrease when explicit DK options are offered, while right answers among substantive responses increase. This indicates that DK choices make participants less likely to cheat and/or guess, and that the remaining respondents are more likely to be truly knowledgeable. We conclude that while DK options induce some satisficing, their potential benefits for data validity increase as cognitive demands of questions rise.