Matthew Graham (Yale University)
Abstract: Survey measures of the public's factual beliefs suggest widespread misinformation on politically relevant matters of fact: not only do many Americans not only choose incorrect responses, but they claim to be certain that their answer is correct. This paper evaluates measures of certainty against two measures of belief strength: the probability of choosing the same response again in a subsequent survey, and the respondent's willingness to bet on their answers. Both measures suggest that survey measures exaggerate the extent of misinformation. Conditional on their stated certainty level, respondents who answer incorrectly are less stable in their responses and less willing to bet on their answers than are respondents who answer correctly. Even when respondents claim to be certain about incorrect answers, observers of politics should view these responses as incorrect guesses, not as evidence that the respondent was misinformed.