Lookups and Measured Knowledge

• DOI: 10.31219/osf.io/amq82
• Publication Date: 2024-05-13T21:28:47Z
• Authors: Matthew H. Graham

How do survey respondents who look up the answers affect measures of knowledge? This paper derives "lookup bias'' in knowledge. Lookup bias has two key components: prevalence lookups and yield," i.e. average lookup's effect on measured Even when are common, lookup yield is small information scarce or inaccurate, as well use to confirm what they already know rather than fill ignorance. To test these insights, I develop an instrumental variables method for estimating apply it study politics. On standard political knowledge questions, about 0.5, meaning that each results 0.5 additional correct answers. Variation across questions appears be related variation availability accurate probability guessing correctly. For example, "catch" designed easy hard guess, close 1. The suggest most settings, considerably smaller proportion answer.