Campaign Funding Disclosures and Mass Attitudes

Download data and study materials from OSF

Principal investigator:

Michael Sances

Massachusetts Institute of Technology

Email: mwsances@mit.edu

Homepage: http://www.mit.edu/~mwsances/


Sample size: 3006

Field period: 3/23/2011-9/1/2011

Abstract

Campaign finance policy in the United States is founded on key assumptions about how voters use information about money to evaluate candidates and institutions. In one view, voters use information about campaign contributions as informative signals of a candidate's policy views, leading to better informed voters; in another view, contributions signal the potential for government corruption, leading to more cynical voters. Despite the prominence of these views in theoretical and policy debates, empirical evidence is scarce. To circumvent issues with existing observational studies, and to speak more directly to policy debates, I present the results of two survey experiments where I randomly vary voters' information about money in politics in mock election campaigns. My results support the view that campaign contributions allow voters to better place candidates on an ideological spectrum. In contrast, I find only limited evidence that contributions depress voters' trust in government.

Hypotheses

(1) Campaign contributions provide voters with information about candidates' ideological positions, increasing voter competence.

(2) Campaign contributions also provide voters with information about government performance, harming trust in government.

Experimental Manipulations

All study participants were shown a video clip of approximately 30 seconds in length. This video was an edited version of an actual campaign advertisement distributed by a political action committee in support of a candidate for the U.S. Senate in Ohio in the 2010 elections. While the candidate ran on the Republican ticket, the original advertisement did not mention the candidate’s party and emphasized the non-partisan issue of “job creation.” I obtained the original video from public sources and edited it to my purpose using free and open source software. The treatments in this study took the form of textual and audio messages inserted at the beginning or end of the original video. In the Disclaimer condition, these treatments informed the subject that “Americans for Change,” a fictional organization, was “responsible for the content of this advertisement.” This message mirrored the original ad, except for the fictional organization title. The Disclosure conditions added text that listed the “top five funders” of Americans for Change. In the Labor Disclosure condition, these funders were five labor unions that were listed among the most influential, in terms of total political contributions, by the Center for Responsive Politics web site at http://www.opensecrets.org as of January 2011. In the Business Disclosure condition, these “top five funders” were five corporations selected from the list of contributors to American Crossroads, a political action committee active in the 2010 elections. Finally, the Control condition received no disclaimer or disclosure information.

Outcomes

Measures of the likelihood of voting for the fictional candidate (for H1) and measures of trust in government and political efficacy (for H2).

Summary of Results

In the TESS-funded experiment, I found no evidence in support of H2, but strong evidence in support of H1. That is, respondents who were given information about contributions were no more or less likely to express trust in government. However, the ideological content of the contribution pushed different partisans toward more or less support for the candidate, in predictable ways. For example, Democratic respondents who were told the candidate received contributions from labor unions were more likely to vote for the candidate.

Additional Information

I supplemented the TESS study with an additional experiment fielded on Mechanical Turk.

References

Sances, M. W. “Is Money in Politics Harming Trust in Government? Evidence from Two Survey Experiments.” Working Paper, Department of Political Science, MIT.