Democratic Erosion, Partisanship, and Election Observers: Evidence from a Survey Experiment

Abstract

To what extent do voters support election observation missions (EOMs) monitoring their elections? Building on but departing from prior work, we argue that citizens’ support for election observation missions likely depends on their partisanship. To test our argument, we conduct a survey experiment with a national sample of about 3, 500 Americans in the run-up to the 2020 United States presidential election. Respondents read an adapted news article about the deployment of common interventions in the upcoming election. To test our theoretical expectations, we randomize the partisan source endorsing the interventions. We find that respondents who self-identify as Democrats and Independents/Others are much less likely to support election observers when endorsed by Trump. In addition, Democrats are more likely to support election observers if they read that Biden approves of them. In contrast, Republicans are less likely to support an intervention if they read the same. Our results contribute to ongoing debates about how to promote and defend free elections in backsliding or democratizing countries and, more generally, to the literatures on election observation and public opinion.

Resumen

¿Apoyan los votantes el despliegue de misiones de observación electoral con el fin de monitorizar las elecciones? En este artículo, abordamos estas cuestiones, las cuales resultan oportunas en la actualidad. Argumentamos, sobre la base de trabajos anteriores, pero al mismo tiempo apartándonos de estos, que el apoyo a las misiones de observación electoral por parte de los ciudadanos depende, con toda probabilidad, de su partidismo. Con el fin de poner a prueba nuestra hipótesis, llevamos a cabo un experimento de encuesta con una muestra a nivel nacional de, aproximadamente, 3.500 ciudadanos estadounidenses durante el período previo a las elecciones presidenciales de Estados Unidos de 2020. Los encuestados leyeron un artículo de noticias, oportunamente adaptado, referente al despliegue de intervenciones comunes en aquellas elecciones. Con el fin de poner a prueba nuestras expectativas teóricas, aleatorizamos la fuente partidista que respaldaba estas intervenciones. Concluimos que los encuestados que se identifican a sí mismos como demócratas, independientes o de otros partidos son mucho menos propensos a apoyar a los observadores electorales cuando estos están respaldados por Trump. Además, resulta más probable que los demócratas apoyen a los observadores electorales si leen que Biden los aprueba. Por el contrario, resulta menos probable que los republicanos apoyen una intervención si leen esta misma noticia. Nuestros resultados contribuyen a los debates en curso en materia de cómo promover y cómo defender elecciones libres en países en retroceso democrático o en proceso de democratización y, de manera más general, a la literatura relativa a la observación electoral y a la opinión pública.

Résumé

Les électeurs soutiennent-ils le déploiement de missions d’observation pour surveiller les élections ? Dans cet article, nous traitons ces questions fort à-propos. En nous fondant sur des travaux antérieurs, puis en nous en écartant, nous affirmons que le soutien des citoyens aux missions d’observation des élections dépend probablement du parti qu’ils soutiennent. Pour tester notre argument, nous menons une expérience de sondage à l’aide d’un échantillon national d’environ 3 500 Américains à l’approche des élections présidentielles américaines de 2020. Les participants ont lu un article de presse adapté quant au déploiement d’interventions courantes dans les élections à venir. Pour tester nos attentes théoriques, nous randomisons la source partisane de soutien aux interventions. Nous remarquons que les participants qui s’identifient comme démocrates, indépendants ou autres ont beaucoup moins de chances de soutenir les observateurs d’élections soutenus par Donald Trump. De plus, les démocrates ont plus de chances de soutenir les observateurs d’élections après avoir lu que Joe Biden y était favorable. Par contraste, les républicains ont moins de chances de soutenir une intervention s’ils lisent cette information. Nos résultats contribuent aux débats actuels sur la façon de promouvoir et de défendre des élections libres dans les pays où s’opère un retour en arrière ou qui se démocratisent, et, de façon plus générale, aux littératures sur l’observation d’élections et l’opinion publique.

Keywords:

• Democratic promotion
• election observers
• survey experiment

Acknowledgements

Thanks to Dawn Brancati, Nic Cheeseman, Annekatrin Deglow, Susan Hyde, Tore Wig, and other participants at the American Political Science Association, Peace Science Society Elections and Violence Workshop, the Folke Bernadotte Academy Research Workshop (The Pursuit of Peaceful Polling), the Midwest Political Science Association meeting, and the Peace Research Institute Oslo Series for their very helpful comments on the study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For example, in their study of American public reactions to revelations of foreign electoral intervention, Tomz and Weeks (2020) find that even modest forms of intervention polarize the public along partisan lines. They find that “Americans are more likely to condemn foreign involvement, lose faith in democracy, and seek retaliation when a foreign power sides with the opposition, than when a foreign power aids their own party” (Tomz and Weeks 2020).

2 Our study was pre-registered with OSF. The pre-analysis plan is available here: https://osf.io/rvmtz.

3 From a more critical perspective of EOMs, Nganje and Nganje (2019) argue that political elites might even support a potentially flawed election rather than embrace the unrestrained exercise of US power that might accompany many international missions.

4 See “Declaration of Global Principles for Nonpartisan Election Observation and Monitoring by Citizen Organizations” (https://www.gndem.org/declaration-of-global-principles/) and “Declaration of Principles for International Election Observation,” (https://bit.ly/3bsLSe0).

5 In the case of the US, there is subnational variation in whether domestic and international observers are allowed to monitor the electoral process. When states do allow observers, state authorities work with the US State Department to invite and designate official international election observers. Importantly, ahead of the November 3rd presidential election, the OSCE launched a limited delegation of long and short-term election observers.

6 Corstange and Marinov (2012) use a survey experiment to examine the role of foreign electoral interference in Lebanon, focusing on electoral process and partisan cues.

7 If we find effects among EOMs, we suspect these effects would be more pronounced among partisan poll watchers; however, we lack the data to test this hypothesis.

8 This logic only holds for credible IEOMs not international “zombie observers” who are meant to legitimate a flawed election. See “How Dictators use Zombie Election Monitors to Stay in Power,” National Endowment for Democracy, October 29, 2014 (https://www.ned.org/buzzfeed-how-dictators-use-zombie-election-monitors-to-stay-in-power/).

9 For a discussion of the relationship between international organizations and a new era of populist nationalism, see (Copelovitch and Pevehouse 2019).

10 Even more explicitly, Weaver and Prowse (2020) argue that the US commitment to democracy is highly racialized. They argue that in the US racial authoritarian governance has “deeply shaped our institutions, political arrangements, and state development” (Weaver and Prowse 2020).

11 As a personal example, Curtice worked with the Carter Center’s Democracy Program on several EOMs; one of which occurred under the Obama administration while the remaining three deployed during the Trump administration. While the Carter Center is a nonpartisan organization, one of their key funding partners is United States Agency for International Development (USAID). USAID operates under the president’s authority, even if the Carter Center is non-partisan in practice.

12 Asunka et al. (2019) note that Ghana’s elections have routinely suffered from allegations of fraud and malpractices” in addition to “low intensity violence and voter intimidation” were relatively widespread in each national election held since 1992.

13 See Peter Baker, Nick Corasaniti, Michael S. Schmidt and Maggie Haberman, “The Voting Will End Nov. 3. The Legal Battle Probably Won’t,” New York Times, August 17, 2020.

14 See, Marc Fisher “With Election Day looming, an anxious nation hears rumblings of violence,” Washington Post, October 31, 2020; Michael Martina and Andy Sullivan, “Fears of voter intimidation follow Trump’s debate call for his backers to monitor polls,” Reuters, September 30, 2020; and Chris Megerian and Arit John, “Activists brace for voter intimidation efforts on election day,” Los Angeles Times, October 20, 2020.

15 See Barbara Sprunt “FACT CHECK: Trump Falsely Claims That Votes Shouldn’t Be Counted After Election Day,” NPR, November 1, 2020. Such claims ramped up following the election. See David A. Fahrenthold, Elise Viebeck, Emma Brown and Rosalind S. Helderman’s article, “Here are the GOP and Trump campaign’s allegations of election irregularities. So far, none has been proved,” Washington Post, November 10, 2020.

16 OSCE “General Elections, 3 November 2020,” (https://www.osce.org/odihr/elections/usa/456787).

17 https://www.osce.org/files/f/documents/1/1/456169_0.pdf.

18 See OSCE “United States of America, General Elections, 3 November 2020: Final Report,” February 9, 2021 (https://www.osce.org/odihr/elections/477823).

19 The institutional review board at REDACTED approved this research.

20 The speed check involved dropping all respondents who completed the survey in one third of the median completion time. Our results are substantively the same if we do not drop these respondents from the survey.

21 Julian Borger and Sam Levine, “International body says voting problems ‘could harm integrity’ of US election,” The Guardian, August 13, 2020 (https://www.theguardian.com/us-news/2020/aug/13/osce-election-monitors-us-integrity-november).

22 In the lead-up to the 2020 election, Trump also publicly endorsed the idea of deploying his own partisan election observers. Partisan observers (i.e., poll watchers and challengers) are a commonly used tool by political parties. Poll watchers are often selected by political parties to ensure that their party has a fair chance of winning an election. While poll watchers perform a similar function to EOMs, their focus is the interests of the party rather than the general credibility of the election.

23 We acknowledge there is a double-barreled aspect of our design as it raises both concerns about election integrity and endorses election observers. We did this by design. To keep our vignette representative of things as they occur in the world and to maximize ecological validity, we did not try to unbundle the treatment here.

24 We note that deception was used in the endorsement experiment. We attributed quotes to actors who did not say them. We addressed this by debriefing respondents at the end of the survey. Complete debriefing text is included in the IRB protocol.

25 As we show in the Online appendix, we obtain substantively similar results if we examine treatment effects using an ordinary-least-squares model that regresses Agreement on a set of treatment indicators and a set of pre-treatment covariates included to increase precision.

26 Again, we obtain substantively similar results if we examine treatment effects using an ordinary-least-squares model that regresses Agreement on a set of treatment indicators interacted with respondent party identification. These models, reported in the Online appendix, also include a set of pre-treatment covariates to increase precision.

27 The results of a multiplicative interaction model suggest that these differences across groups are statistically significant (p < 0.05).

28 As we show in the Online appendix, we obtain substantively similar results if we examine treatment effects using an ordinary-least-squares model that regresses Margin of Victory on a set of treatment indicators and a set of pre-treatment covariates included to increase precision.