Investigating Frame Strength: The Case of Episodic and Thematic Frames

Abstract

An impressive body of research shows that the framing of an issue affects citizens' attitudes, but also that some frames are more influential than others. Yet, we have surprisingly limited knowledge of the factors that affect the strength of a frame, that is, the frame's capacity to influence citizens' opinions. Therefore, this study investigates the relative strength of episodic and thematic frames to argue that our understanding of the dynamics of frame strength can be advanced through a better incorporation of citizens' emotional reactions. Based on experimental data, I demonstrate that the relative strength of episodic and thematic frames depends on the intensity of citizens' emotional reactions. When there are no or weak emotional reactions, thematic frames are stronger than episodic frames, whereas the relative strength is increasingly reversed when intense emotional reactions are inflamed in the audience. I conclude by discussing the implications of the findings.

Keywords:

• frame strength
• episodic
• emotions
• opinion formation
• appraisal theory

Acknowledgments

I thank Jamie Druckman, Leonie Huddy, Christoffer Green-Pedersen, Michael Bang Petersen, Rune Slothuus, seminar participants at Aarhus University, and discussants at the 2009 annual meeting of the Midwest Political Science Association and the 2008 meeting of the Nordic Political Science Association for helpful comments and advice.

Notes

1. Thus, pro and con frames provide more than a single argument or position on an issue because they also provide meaning to an issue and suggest how to understand and think about it. In this respect, the characteristics of pro and con frames are consistent with the general characteristics of frames as specified by Nelson and Kinder (1996), who with respect to the general notion of framing emphasize that “frames are more than simply positions or arguments about an issue. Frames are constructions of the issue: they spell out the essence of the problem, suggest how it should be thought about, and may go so far as to recommend what (if anything) should be done” (p. 1057; for parallel arguments, see Slothuus, 2008, p. 3, and Sniderman & Theriault, 2004, p. 136).

2. See Gross (2008, p. 172) for a parallel prediction.

3. The results from the pretest were replicated in the data from the main experiment. The findings in these data also demonstrated that the two pro and the two con frames did not differ in terms of the participants' subjective perception of their strength.

4. Compare Petersen (2010) for a similar selection criterion.

5. As a preliminary step, a randomization check was conducted. No evidence of systematic differences was found across conditions on the demographic and attitudinal background measures.

6. In their study of support for the EU, Vliegenthart, Schuck, Boomgarden, and de Vreese (2008, p. 431) also found that pro frames (a benefit frame specifically) worked better than con frames.

7. To further investigate the robustness of these results when controlling for variables identified by past research as influential moderators of framing effects (e.g., Haider-Markel & Joslyn, 2001; Zaller, 1992), the analyses in Table 3 were replicated controlling for (a) full three-way interactions between frame direction (pro/con), frame rhetoric (episodic/thematic), and the partisanship of the respondent (voters of center-right parties/center-left parties) and (b) full three-way interactions between frame direction (pro/con), frame rhetoric (episodic/thematic), and the political awareness of the respondents (high/low political awareness). In both of these tests of robustness, all three-way interaction terms between frame rhetoric (episodic/thematic), frame direction (pro/con), and each emotional reaction item remained significant (all ps < .05 or less, two-sided tests). This is consistent with the findings of Druckman and McDermott (2008, pp. 309–310) that emotions maintain statistical significance as moderators of framing effects when controlling for the moderating effect of these classical individual predispositions.

8. I also checked for complex four-way interaction effects between discrete negative and positive emotions and the direction of the frame and the rhetoric of the frame. All results were statistically insignificant.

9. The mean strength ratings of the episodic frames were .62 and .59, respectively.