Lies, Damned Lies, and Elections: Three Essays in Experimental Public Choice

Jared Barton

Advisor: Ragan Petrie, PhD, Department of Economics

Committee Members: Marco Castillo, Thomas Stratmann

Truland Building, #400A
April 26, 2012, 04:00 PM to 01:00 PM

Abstract:

This dissertation examines information transmission, framing, and deception in three field experiments in political environments. The first chapter is a field experiment embedded in a candidate’s campaign for office. The candidate varied whether he revealed his ideological position or explained how to vote to voters, as well as whether he delivered the message in person or left it in their door. These treatments allow us to identify whether political persuasion is due to direct transmission of ideas or indirect transmission through, e.g., costly signaling. The results are consistent with the latter: voters were persuaded by the candidate’s presence, but there is no difference across messages. Voter turnout differences across messages indicate voters did receive the messages, and the fact that voting occurs both privately and well after the campaign contact implies the result is not social pressure from the candidate.

The second chapter explores how message tone and framing influence contribution to political campaigns using field experiments embedded in two candidates’ campaigns. We vary whether two candidates’ partisan supporters receive a positive letter, negative letter, or no letter soliciting funds and support for the candidate. Negative messages may lead to greater giving of both money and votes (voter turnout) through emotional responses, loss aversion, or greater informational content of comparative messages. We find no difference in contribution behavior between treatments; both the negative and positive letter are equally effective at fundraising. Partisans receiving the negative message, however, are more likely to vote than positive message recipients. Further analysis reveals that negative messages are do not mobilize voters relative to the control group, but that positive messages depress them.

The final chapter is a quasi-field experiment conducted during a gubernatorial election to study misinformation. There are reports of voter misinformation in every recent election cycle, though their number, magnitude, and effect is currently unknown (U.S. Election Assistance Commission 2006). I constructed a mock election in which students who registered and voted could receive a cash prize, and varied whether subject-voters received a warning about misinformation, an untruthful email on Election Day indicating the mock election was canceled, both, or neither. There is a weak effect of misinformation on voters generally which is mediated by subjects’ prior knowledge: there is no effect on politically-related majors, but a demobilizing effect on non-politics majors. Overall, the results suggest misinformation plays a small role in voter turnout.