Our findings provide the first experimental evidence that reframing voting choices can change voter opt-out behavior. We hope that this inspires further work on the applications of this for improving polling, campaigning, and GOTV efforts.
Reject-worse choices should better reflect the population by uncovering the preferences of these “double-haters.” We confirm this prediction by surveying Independent voters: Vote FOR Biden or Trump? 33% responded undecided. Vote AGAINST Biden or Trump? Only 20% undecided! Same for Harris vs. Trump!
We used these data to simulate election/polling outcomes at the population level. For typical (choose-better) choices, we show that these outcomes will be least representative of the population when the majority has a clear preference (A>B) but likes both less than the minority.
The other half of our participants performed the same task, but we asked them to choose the lesser candidate (to reject). This group opted out of fewer ballots overall, and opt-outs were now especially rare for a particular choice: Lose-Lose!
We asked half of our participants to vote for the candidate they liked better. These people chose to vote most of the time, but on ~40% of ballots they opted out. What kinds of choices did they opt out of? Ones with two bad options (Lose-Lose)!
We had participants make a series of voting decisions between hypothetical candidates. We tailored these choices to the participant's political positions, creating candidates ranging from ones they would like least to most. For each ballot, participants could vote or opt out.
With election stakes as high as they are, a lot of people want to know what makes someone a “non-voter.” We find that the answer lies not only in the person but in how they are being asked to decide (who they like most vs. least).
