Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
We develop and experimentally test a model of voter information acquisition in nonpartisan elections, both with and without abstention. We theoretically demonstrate that allowing for abstention can increase information acquisition, provided the cost of information is not too low. Our experimental data find that voters are less responsive than predicted to increases in the cost of information. As a result, the cost required to yield higher levels of informedness when abstention is allowed is higher than predicted. Our data are well explained by agent quantal response equilibrium, which accounts for the fact that uninformed voters are often observed to vote, even when it is not rational to do so.