Assessing the Breadth of Framing Effects

Download data and study materials from OSF

Principal investigator:

Daniel Hopkins

Georgetown University

Email: dh335@georgetown.edu

Homepage: http://people.iq.harvard.edu/~dhopkins/


Sample size: 3318

Field period: 9/19/2011-12/27/2011

Abstract

Issue framing is a central concept across the social sciences. But scholarship has yet to consider the breadth of framing effects, meaning the extent to which frames on a specific issue influence attitudes on other policy areas. Here I undertake a survey experiment to identify the breadth of framing effects for frames related to terrorism, criminal justice, health care, and the economy. In this experiment, respondents read a randomly selected frame about one of those issues or else are assigned to a control condition. Respondents then answer randomly ordered questions about several policy areas, including those that were previously discussed in the frame. The results provide evidence that frames are most likely to influence attitudes on the same issue, or else on closely related issues. Framing effects appear to be domain-specific, with only limited spillover into separate areas.

Hypotheses

In this experiment, I test the hypothesis that frames in one policy domain are able to active associations in analogous policy domains through “spreading activation.”

Experimental Manipulations

75% of the respondents are randomly assigned to be exposed to a paragraph-long frame on one of four policy issues: crime, terrorism, health care, and economic stimulus. The remaining 25% of respondents serve as a control group. To hold signals of partisanship or ideology constant, all four frames are associated with conservative positions in contemporary American politics. After answering two randomly ordered questions about spending preferences, respondents are then exposed to a second frame before answering three more questions about spending preferences on other issues.

Outcomes

Respondents are asked to assess five areas of federal spending in a randomly selected order. Those areas are: anti-crime spending; anti-terrorism spending; anti-poverty spending; health care spending; and economic stimulus spending. Each is measured using seven-point scales adopted from the National Election Study.

Summary of Results

As hypothesized, frames are most likely to influence attitudes on the specific issue addressed by the frame. The experiment identifies four strong effects, three of which are by frames on the same issue area. That is, the terrorism frame increases support for anti-terrorism spending, the health care frame decreases support for health care spending, and the stimulus frame decreases support for stimulus spending. The anti-crime frame also influences attitudes on the related issue of anti-terrorism spending. Yet the general conclusion is one of limited breadth: frames appear to primarily influence attitudes on the specific topic, with only limited spillover effects on other attitudes.