When Do Partisans Stop Following the Leader?

ABSTRACT

Evidence of public opinion blindly following political leader rhetoric has important implications for the scope of elite influence and normative democratic concerns. Past research, however, does not test the strength of real world leader cues amid signals that conflict with a leader’s policy message, and thus has not gauged sthe robustness of the “follow-the-leader” dynamic. The current study explores whether two different conflicting signals – 1) opposing intra-party Congressional elite cues and 2) negative policy information that gives compelling reasons to oppose a policy – attenuate leader influence in support of a realistic counter-stereotypical policy. A national survey experiment with two parallel partisan designs shows that individuals follow their leader to a substantial degree whether or not conflicting signals are present. Conflicting co-party elite cues do not attenuate leader influence among Republicans. For Democrats, although they weaken amid opposition, leader cues still shape mass opinion sizably. Providing substantially more information about the policy at hand does not make either partisan group much less likely to follow their leader, a finding that holds regardless of individuals’ preexisting ideology in the policy area. Results demonstrate the broad conditions under which “follow-the-leader” behavior holds and reveal a stronger nature of elite influence than previously understood. Party elites and information fail to effectively constrain the sway of prominent leaders, who have considerable latitude in positions they can take without losing mass support.

KEYWORDS:

• Public opinion
• elite cues
• intra-party conflict
• policy information

Disclosure Statement

No potential conflict of interest was reported by the author.

Data Availability Statement

The data described in this article are openly available in the Open Science Framework at https://osf.io/9n3cr/.

Open Scholarship

  

This article has earned the Center for Open Science badges for Open Data, Open Materials and Preregistered. The data and materials are openly accessible at https://osf.io/9n3cr/.

Supplementary Material

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/10584609.2020.1772418.

Notes

1. Brian Resnick, “Trump is a real-world political science experiment.” Vox, July 19, 2018. https://tinyurl.com/ybk4cfc5 (last accessed: December 18, 2019).

2. “Conflicting signals” refers to cues and information that pull partisans toward a policy position that diverges from the position of a co-partisan leader (see Boudreau & MacKenzie, 2014, for similar use).

3. “Republican Party on Free Trade Party Platform.” On the Issues, January 20, 2018. https://bit.ly/2zdA7aJ (last accessed: December 14, 2018).

4. Bradley Jones, “Support for free trade agreements rebounds modestly, but wide partisan differences remain.” Pew Research Center, April 25, 2017. https://pewrsr.ch/2q1I1ys (last accessed: December 14, 2018).

5. Todd Krainin, “Trump, Reagan, and Why Republicans Flip-Flopped on Free Trade.” Reason, May 4, 2018. https://bit.ly/2ArtqiX (last accessed: December 14, 2018).

6. Ariel Edwards-Levy, “Trump Voters Recognize GOP Lawmakers Are Backing Away From Him.” Huffington Post, August 25, 2017. https://bit.ly/2ApmYss (last accessed: December 14, 2018).

7. “Generic elite cues” refer to cues coming from anonymous and abstract party elite sources, such as fictional candidates (Arceneaux, 2008), state legislators (Bullock, 2011; Ciuk & Yost, 2016), or members of Congress (Druckman et al., 2013).

8. Outside of a survey experiment, it is likely that individuals more often receive cues from specific figures than generic elites, which, for example, consist of “most Democrats” or “62 out of 69 House Democrats” supporting a certain position (Bullock, 2011). Future research should probe this descriptive question.

9. Naftali Bendavid and Christopher Conkey, “Obama, Democrats in Congress Clash on Spending.” Wall Street Journal, June 27, 2009. https://on.wsj.com/2UjnWQv (last accessed: December 14, 2018).

10. Michael Fletcher, “House Republicans clash with Bush on immigration reform.” Washington Post, January 3, 2005. https://bit.ly/2zIpZ7T (last accessed: December 14, 2018).

11. This term comes from Arceneaux (2008), who defines it as “positions that are inconsistent with partisan stereotypes.” Similarly, in the context of this study, “counter-stereotypical” position-taking means policy endorsements outside of the party mainstream – namely, in terms of being out of step with expected elite behavior and traditional elite stances within the party. See Section 4.4 for more.

12. See Appendix Section C for an expanded version of this table with more details on coding.

13. Preregistration materials can be found in Appendix Section A.

14. These distributions cannot be compared against national benchmarks because the population of interest is not all American adults but rather all partisan American adults. No reliable benchmark data exists for this latter group.

15. Tables D.1 and D.2 in the Appendix contain randomization checks for the Republican and Democratic designs, respectively. Despite some deviations, key demographics are well balanced across the six conditions in each design. When later estimating treatment effects, inclusion of pre-treatment controls to correct for any covariate imbalance does not change results. See Appendix Section F for more.

16. The term “party leader” describes especially prominent party figures referred to by name in the experiment to create a more realistic and involving cue (see Nicholson, 2012). See Section B in the Appendix for more discussion on the approach to leader selection and the suitability of the two figures chosen.

17. Exact wording of the survey instruments can be found in Section J.

18. The fictitious policy bill vignettes draw directly from news reports and documents regarding past actual policy bills in these same areas. The bottom of Section I in the Appendix contains links to these resources.

19. Section 1 contains full treatment contents, which show the use of “headlines” (larger sized, underlined text summarizing the basic policy description and circumstantial cues/information) as a way to more strongly administer the treatment. In addition to mimicking how people often encounter news communication and what they prioritize in news consumption (Ecker et al., 2014, p. 324), using headlines better administers the treatment to possible survey satisficers – those quickly taking the survey who might not fully read the vignettes – and thus more survey-takers should properly receive the treatment as a result.

20. The treatment contents can be found in Section I in the Appendix.

21. It’s worth noting the possible unintended bipartisan interpretation among respondents seeing a co-partisan leader support a policy that past out-party elites were inclined to support. At the same time, the potential for immediate bipartisanship would not exist, given the past nature of the out-party cue, and other research suggests this combination of cues does not fulfill broad two-party elite support that is necessary for bipartisanship to influence opinion (Bolsen et al., 2014).

22. For example, past work has tested how Democrats react to a Democratic candidate taking a pro-life stance (Arceneaux, 2008). This limits how meaningful the results are, as the situation remains largely non-credible (most Democrats do not take pro-life positions), especially at the national level, over which party leaders’ influence presumably extends.

23. “Trump is inclined to ‘just spend money’ on infrastructure, aide says, but House GOP could have issues with that.” Eamon Javers, NBC News. January 10, 2018. https://cnb.cx/2ShcD9k (last accessed December 14, 2018).

24. “Obama: U.S. will overcome Democrats’ opposition to free trade deal.” Dave Boyer, Washington Times. August 1, 2016. https://bit.ly/2Q2Ghm7 (last accessed December 14, 2018).

25. This specification deviates from the preregistered analysis, which was incorrectly specified for testing Hypotheses 2 and 3. Many thanks to an anonymous reviewer for clarifying this point. Appendix Section E details the original analysis, why it was unsuitable for testing the hypotheses, the correct specification, and how results changed after this correction.

26. See Appendix Table G.2. Importantly, when Trump’s message is absent, conservative Republicans decrease their policy support in the face of the information treatment. This result suggests the Issue Area Ideology question about government services captured a real underlying attitude in this issue realm.

27. Curiously, Democrats more favorable toward Obama appear a bit more inclined to reject his message when opposed by other elites in the party, though the relevant three-way interaction effect here is not statistically significant (see Table G.4).

Additional information

Notes on contributors

Alexander Agadjanian

Alexander Agadjanian is a research associate in the MIT Election Lab. He studies political psychology, political behavior, and race & politics, and has been published in Political Behavior and Research & Politics.