Download data and study materials from OSF
University of California, Berkeley
Sample size: 2071
Field period: 01/25/2013-05/24/2013
Studies on representation generally adopt one of two approaches, either an examination of the goods directed at constituents, or an analysis of legislative roll-call voting. Both activities figure prominently into the responsibilities of a representative, but neither gets at the central question of what qualifies as good representation, and what voters look for in their member of Congress. Do constituents prefer representatives with a record of localized goods or general policy achievements? To address these questions, I fielded a series of experiments using a simulated story about two anonymous incumbents running for a redistricted seat. By framing the experiment as a real news story, the study maintains a high degree of mundane realism while still regulating the information respondents use when choosing between the two candidates. This approach controls for important features about the member, such as partisanship and ideology, and allows for a more direct test of whether candidates who run on records of pork are more attractive to voters than representatives who focus on broad policy. In the end, I find that constituents often prefer policy-based appeals, but that the ideological content heavily drives this preference.
The overall experiment examines whether constituents prefer representatives with a record of distributive goods (pork) or broad legislation (policy). The TESS experiment tested two specific hypotheses: does the ideological content of the policy drive this selection, and how does effort moderate these effects.
H1 - Are constituents more likely to select the policy candidate when the policy highlighted is closer to their ideological position. In other words, are Democrats [Republicans] more favorable of the MC that takes credit for helping to extend coverage to the uninsured [lowering costs through professional responsibility] compared to the neutral and opposite party conditions?
H2 - Do additional claims of "committee work" on the issue's behalf increase the likelihood that constituents will select that candidate? Note: this ancillary hypothesis was particularly relevant for the policy candidate, as a common critique of policy based credit claims is that they lack credibility.
Subjects were randomly assigned to read one of several "news" stories. The introduction identified the candidates as male office-holders hailing from the same party. From there, the article then briefly describes the substantive focus of the two candidates' campaigns. In each vignette, one candidate highlights a record of securing localized projects, hence pork. The pork candidate's challenger focuses instead on a record of broader, national policy achievements.
The pork candidate highlights a hospital project. For the policy candidates, the study uses several issues that were common during the 112th Congress, such as a medical device reform, used in the "Neutral" condition. In the second and third conditions, the policy candidate takes credit for legislation that more closely resembles issues from the parties' respective platforms. For the "Democratic" leaning record, the candidate discusses legislation to expand health coverage among the uninsured. The "Republican" leaning message instead highlights the member's efforts to lower costs through individual responsibility.
In addition to testing the ideological content of the policy, the study consists of two additional conditions used to examine whether respondents prefer the Neutral policy candidate if he claims to have exerted additional effort in committee, "PolicyHE." For completeness, the study also includes a pork high effort condition, "PorkHE."
Three outcome measures.
1 - Which candidate would you prefer as your representative?
2 - Which candidate do you believe works harder in Congress?
3 - On a scale of 1-7, how much do you approve of Congress?
H1 - Both sets of partisans preferred policy over pork when the issue matched their PID, at least compared to the neutral condition. Moreover, Republicans really hated the Democratic-leaning policy candidate, selecting him over the pork candidate only 20% of the time. The Democratic respondents were a little less predictable. They actually liked the Republican-leaning policy candidate almost on par with the Democratic condition. This odd result may speak more to the fact that the Republican-leaning treatment addressed lowering costs, which sounds good to everyone.
H2 - No major significant findings here, particularly with respect to the first outcome measure on which candidate the subject preferred. There is some evidence that respondents saw the higher effort candidates as "working harder" (the second outcome measure), but these findings are fairly weak as well.