Framing Effects on Voter Attitudes and Behavior: A Meta-Analysis

• In Event: Framing Effects in the News

Abstract

One of the most widely studied phenomena in political science and communication research are framing effects, which occur ״when (often small) changes in the presentation of an issue or an event produce (sometimes large) changes of opinion” (Chong & Druckman, 2007, p. 104). Even though a large and growing body of research has examined the impact of framing on a wide range of attitudinal and behavioral outcomes—such as policy support, voting preferences, and political trust—the literature remains largely indecisive with regard to fundamental questions such as the overall strength of framing and the political and psychological conditions that facilitate or hinder a framing effect from occurring.
The main source for this indecisiveness is the lack of conceptual clarity in the literature, as different researchers define the process of framing in very different ways (Cacciatore, Scheufele, & Iyengar, 2016; Matthes, 2009). Specifically, while some scholars focus on “equivalency framing” and examine, for instance, the impact of gain versus loss frames (Tversky & Kahneman, 1981), others focus on “emphasis frames” (Druckman, 2001) and examine the impact of positive versus negative arguments. A third theoretically important focus in this literature is the effect of frame competition, defined as instances in which voters are exposed to conflicting perspectives on the same issue at the same time.
To address the open questions in this literature, we conduct a comprehensive meta-analysis covering the attitudinal and behavioral effects of the most widely used and theoretically meaningful types of framing. By statistically integrating the results of 110 research reports, we find that different conceptual paradigms lead to different conclusions regarding the impact of framing in the political domain. For example, we find that across 41 research results, emphasizing positive rather than negative aspects of a policy increases support for that policy significantly and substantially (d = 0.42; p < .001). However, the mean effect of this type of framing on citizens’ actual political behavior (i.e., their willingness to take action to promote the issue) is practically zero (d = 0.08; p = .05, k = 7).
Another notable finding emerging from our meta-analysis is that framing a policy in terms of potential gains (i.e., achieving a desirable outcome) rather than in equivalent terms of potential losses (i.e., avoiding an undesirable outcome) has no effect whatsoever on people’s support for that policy (d = –0.01; p > .1, k = 16). Yet, when examining the more specific case of citizens’ willingness to take risks when evaluating prospective policies, we find that gain versus loss framing has a large and consistent effect in the political domain. When presented with a gain frame, citizens are significantly and substantially more risk-averse compared to when presented with a loss frame (d = 0.46; p < .001, k = 17). The latter result not only points to the conditional effect of this type of framing but also confirms that prospect theory (Tversky & Kahneman, 1981) is broadly applicable to the domain of politics.
A third major finding our analyses point to relates to frame competition. One of the main criticisms against framing studies is the fact that most of them are conducted in a non-competitive experimental environment, in which respondents are exposed to a single, one-sided frame without opposition or counter-arguments (Chong & Druckman, 2007; Sniderman & Theriault, 2004). However, in the real world, citizens are typically exposed to opposing arguments regarding political issues. We examine how competitive framing, in which citizens are exposed simultaneously to two or more competing frames on the same issue, affects their attitudes. Analyzing 29 study results, we find that competing frames cancel each other’s effect out, leaving citizens’ attitudes unchanged (d = –.09; p > .05). This result is a strong indication that framing may not be as influential in causing real attitude change as often argued.
Taken as a whole, our results represent two major contributions to the literature. First, while previous studies have described and analyzed conceptual inconsistencies in the framing effects literature (e.g., Druckman, 2001; Scheufele & Iyengar, 2012), an integrative empirical account of framing effects studies has not been available to date. We show that different conceptual paradigms yield substantively different conclusions regarding the efficacy of framing. For example, when examining effects on policy support, emphasis (i.e., positive vs. negative) framing affects it substantially, while equivalency (i.e., gain vs. loss) framing does not. Second, our use of a meta-analysis to address ongoing debates in the framing effects literature is advantageous because of the highly powered statistical tests this method allows and due to the possibility to conduct further moderator analyses specifying the conditions that facilitate or hinder framing effects.

Authors

• Eran Amsalem, The Hebrew University of Jerusalem

• Alon Zoizner, Hebrew University of Jerusalem