Randomized Ballot Order Can Increase Invalid Votes: Evidence from Australia

Abstract

Randomized ballot order of electoral candidates minimizes unfair advantages for candidates with favorable positions on a ballot. However, it may also encourage voter errors because it minimizes information cues otherwise provided to voters, such as the order of candidates by their strength (e.g., vote shares in the previous election) or by their names (e.g., alphabetical order). We examine this previously uninvestigated effect of ballot-order randomization on voter errors using nearly 60,000 polling-place level data from seven Australian House of Representatives elections held between 1996 and 2013. Specifically, we show that the larger the deviation of the actual ballot order from a hypothetical ballot order that could be designed to help voters accurately fill out their ballot, the larger the percentage of invalid votes. The estimated effects are relatively small, however, compared to ballot order effects for candidates listed first on the ballot (King and Leigh, 2009). We call for further inquiry into the desirable and undesirable effects of ballot-order randomization.

Keywords: ballot order, informal vote, invalid vote, compulsory voting, Australia, natural experiment

JEL Classification: D72, C99