End-to-end verifiable (E2E) voting systems provide voters with (privacy preserving) receipts of their ballots allowing them to check that their votes were correctly included in the final tally. A number of recent studies and field tests have examined the usability of ballot casting and receipt checking. Simply checking receipts, however, is not enough to provide strong assurance that the election outcome is correct; voters must also be counted on to report any discrepancies between their receipts and the official record when they occur. In this paper we designed and ran a study examining the frequency and conditions under which voters (a) check their receipts, and (b) report discrepancies when they occur. Participants were recruited online and were asked to vote in a survey on charitable giving. Similar to previous work, we found that the proportion of voters performing a receipt check was low. More importantly, within this group, we found that the proportion of voters reporting discrepancies was also low. We did however observe that the incidence of receipt checking was significantly higher when the election outcome was unanticipated or unexpected by voters. In the condition with an adverse election result we observed that, while 7.5% of voters checked receipts, only 0.5% filed a dispute when shown an incorrect receipt. With such low reporting rates, E2E voting systems will struggle to detect fraud with high confidence, especially in elections with narrow margins of victory. We posit, therefore, that improving the usability of the receipt check component in E2E systems is an important open problem.
