Chiefs and Pre-Election Violence in Sub-Saharan Africa: The Precolonial Legacy of Centralized States

Abstract

This article examines how African chiefs shape pre-election violence (PEV) locally. I argue that chiefs with greater capability of voter coordination, which stems from precolonial centralized states, reduce local risk of PEV. Utilizing existing and original data, I find that precolonial centralized states have a negative effect on PEV in Anglophone Africa, where precolonial institutions have been better preserved. This relationship is robust to a series of tests, including two-stage least-squares regressions. The findings contribute to our knowledge of how local political actors shape violent campaign strategies and of the diverse (pre-)colonial legacies in Sub-Saharan Africa.

Este artículo estudia las maneras en que los gobernantes africanos dan forma a la violencia preelectoral (PEV, por sus siglas en inglés) a nivel local. Argumentamos que aquellos gobernantes que tienen una mayor capacidad de coordinación de votantes, la cual se deriva de los Estados centralizados precoloniales, disminuyen el riesgo de PEV a nivel local. Concluimos, mediante el uso de datos ya existentes y de datos originales, que los Estados centralizados precoloniales tienen un efecto negativo sobre la PEV en el África anglófona, donde las instituciones precoloniales se han conservado mejor. Esta relación se mantiene fuerte cuando la sometemos a una serie de pruebas, que incluyen regresiones de mínimos cuadrados en dos etapas. Nuestras conclusiones contribuyen a nuestro conocimiento, tanto en materia de cómo los agentes políticos locales dan forma a las estrategias de campaña violenta como en materia de los diversos legados (pre)coloniales en el África subsahariana

Cet article analyse la manière dont les chefs africains façonnent les violences pré-élection (VPE) au niveau local. J’affirme que les chefs qui sont davantage en mesure de coordonner les électeurs, grâce à la centralisation de l’État précolonial, réduisent localement le risque de VPE. À l’aide de données existantes et inédites, je constate que les États précoloniaux centralisés s’accompagnent d’un effet négatif sur les VPE en Afrique anglophone, qui a mieux préservé les institutions précoloniales. Cette relation résiste à une série de tests, y compris des régressions de moindres carrés à deux étapes. Les résultats enrichissent nos connaissances quant aux acteurs politiques locaux et leur influence sur les stratégies de campagne violentes et quant aux divers héritages (pré-)coloniaux en Afrique subsaharienne.

Keywords:

• Chiefs
• precolonial States
• pre-election violence
• sub-Saharan Africa

Acknowledgments

I would like to thank Jessica Maves Braithwaite, George Bob-Milliar, and Alhassan Ibn Abdallah for their invaluable support for this research. I also thank the reviewers and editors for their insightful feedback. Finally, thanks are also due to discussants and attendees at Elections, Violence, and Politics (EVAP) Workshop 2021 and MPSA 2022 for their helpful comments. All errors are my own.

Notes

1 This article uses “voter coordination” rather than “voter mobilization” to describe chiefs’ role in elections. While both terms involve actions to encourage bloc voting, “voter mobilization” suggests a more active involvement. In contrast, “voter coordination” portrays a more reserved position. My fieldwork reveals that “voter coordination” better captures the chief-politic dynamics, especially considering that chiefs, as per the laws of many Anglophone African countries, are non-partisan. Despite privately influencing voters, most chiefs maintain a public distance from politics.

2 In this article, I differentiate “traditional” institutions from “precolonial” institutions. I use the former term to signify the native authorities created by the colonial government (Mamdani 1996). The latter, on the other hand, denotes the institutions that had existed before the colonial conquest. For example, the warrant chiefs invented by colonial government are part of the traditional institutions, even though it is groundless in precolonial institutions.

3 The field research has been reviewed and approved by the institutional review board of the University of Arizona, with Approval No. 2106880638. A detailed description of my fieldwork is in Appendix A (Supplementary material).

4 While acknowledging the ongoing evolution of traditional institutions, this article primarily delves into the precolonial and colonial roots of chiefs’ authority. It does not disregard the impact of post-independence states, which successfully curtailed chiefs’ power at the national level (Ubink 2008). Nevertheless, chiefs often maintain significant influence locally, particularly in land custody and dispute resolution, despite challenges from nationalist elites (Logan 2013; Mengisteab 2019; Ubink 2008). The article focuses on this local influence, emphasizing how chiefs coordinate voters within their communities.

5 For a more detailed discussion regarding how political parties and party elites interact with chiefs, please refer to Appendix D (Supplementary material).

6 As Ekeh (1990) points out, the lack of state penetration in African societies renders kinship networks critical in contemporary African politics, including elections.

7 These includes the Armed Conflict Location & Event Data (ACLED) Project (Raleigh et al. 2010), the Deadly Electoral Conflict Dataset (DECO) (Fjelde and Höglund 2022), the Electoral Contention and Violence (ECAV) dataset (Daxecker, Amicarelli, and Jung 2019), and the Social Conflict Analysis Database (SCAD) (Salehyan et al. 2012).

8 An incident of electoral violence is the actual or threatened use of physical violence intended to affect any section of the electoral processes (e.g. party primary, voter registration and transfer, electoral campaign, and voting) and/or the electoral outcomes, prior to or on the election day. Appendix A (Supplementary material) details the data collection strategies.

9 Appendix E (Supplementary material) further compares the death counts across data sets, which indicates a notable underestimate of casualties by existing data sets.

10 “During election campaigns in this country, how much do you personally fear becoming a victim of political intimidation or violence?”

11 Further assessments of the validity of this measure can be found in Appendix D (Supplementary material).

12 A detailed explanation of this algorithm is included in Appendix B (Supplementary material).

13 Appendix D (Supplementary material) details how the overall sample is split.

14 A major concern is that the negative correlations between state centralization and fear of PEV result primarily from the ethnic majority groups, whereas the ethnic minorities surrounded by the centralized groups, who are more likely to be targeted before PEV, have significantly higher fear. However, I find that ethnic minorities are not significantly more fearful than ethnic majorities. See Appendix D (Supplementary material) for detailed results and discussion.

15 Voter solidarity has been used in the existing studies on the effects of traditional institutions on electoral competitions (Conroy-Krutz 2018; Nathan 2019). It is a Herfindahl index of vote share at the constituency level, scaled from 0 to 1, with higher score meaning higher probability that two randomly selected voters voted for the same political party.

16 Author interviews, Upper West, April 25 and 26, 2022.

17 Author interview, Upper West, April 25, 2022.

18 Author interview, Upper West, April 27, 2022.

19 Author interview, Upper West, April 25, 2022.

20 Author interview, Upper West, April 8, 2022.

21 Author interview, Upper West, April 15, 2022.

22 Author interview, Upper West, May 17, 2022.

23 Author interview, Upper West, May 17, 2022.

24 Author interview, Upper West, April 15, 2022.

25 Author interview, Upper West, April 25, 2022.

26 Author interviews, Upper West, April 15 and 16, 2022.

Additional information

Funding

This research was made possible in part by a grant from the American Political Science Association. It was also partially supported by grants from Social & Behavioral Sciences Research Institute and Graduate & Professional Student Council at the University of Arizona. The statements made and views expressed are solely the responsibility of the author.