Who prefers guessing to admitting They Don't Know? Measurement error in financial literacy surveys

A propensity to guess randomly rather than to admit ignorance answering "Don't know" is a plausible reason why frequent wrong answers are given to survey questions that aim to assess competence. We model this source of measurement error and assess its empirical relevance in two consecutive waves of a survey of financial literacy. Misclassification of standard financial literacy indicators is very likely, especially in some demographic groups. Respondents who answer correctly in both waves of the survey are less likely to have guessed in the first wave, and have a lower probability of reporting financial difficulties than those who guessed and were lucky enough to appear literate.