Extreme Policy Proposals and Public Opinion

Download data and study materials from OSF

Principal investigator:

Erik Peterson

Stanford University

Email: epete1@stanford.edu

Homepage: http://web.stanford.edu/~epete1/

Sample size: 5835

Field period: 09/21/2013-01/16/2014


Increasing levels of elite polarization mean that citizens now encounter divergent policy proposals from politicians (McCarty et al. 2006). How does exposure to extreme, as opposed to moderate, policy proposals affect public opinion? Previous research on the effects of elite polarization on public opinion (e.g., Levendusky 2010, Druckman et al. 2013) holds policy substance constant across experimental conditions and is unable to answer this question. In what follows I present a novel experimental design to examine how the public responds to extreme policy proposals on two issues, deficit reduction and abortion, and how partisan cues condition this responsiveness. The large sample of respondents allowed for in the Young Investigators Special Competition is crucial for this project as it will provide the statistical power necessary to detect whether partisan cues condition citizens' responsiveness to policy content and will allow the experiment to be analyzed with nonparametric methods that relax the assumptions about the functional relationship between policy extremity and policy support that would be required with a smaller sample.


1) A policy's degree of ideological extremity will have a limited influence on public support for that policy (e.g., Bullock 2011, Exp. 2).
2) Encountering an extreme policy proposal will pull a respondent's ideal points in a more extreme direction.
3) The public will respond to policy content less when also provided with information about the party supporting the policy.

Experimental Manipulations

1) Policy Content - manipulated by changing the ideological content of two policies (abortion policy and deficit reduction) in a fine-grained manner over a wide ideological interval.
2) Policy Sponsorship - manipulated whether the policy received an endorsement from Democrats, Republicans or did not receive a party endorsement.


1) Respondent support for a particular policy
2) Respondent ideal outcomes in a policy domain

Summary of Results

1) Policy responsiveness is greater here than in prior research. Large shifts in opinion occurred as the distance between a respondent's ideal policy and the policy they were asked to evaluate increased.
2) Extreme policy content did not pull individual's ideal policy preferences in a more extreme direction on these issues. Respondent's own preferred policies were largely stable no matter how extreme a policy they encountered.
3) Partisan cues largely failed to condition responsiveness to content. While there is some heterogeneity in responsiveness to content in the presence of partisan cues by subgroup/issue, respondents were as responsive to content when it was accompanied by partisan cues as when it was provided without a cue.