Is terrorism deadlier in democracies?

Abstract

A long literature examines the relationship between terrorism and democracy. However, little research examines the lethality of terrorist attacks across regime type. This article theorizes the terrorism that democracies do experience will be less deadly. Democracy increases the opportunity for nonstate actors to use terrorism to attract attention to their causes, which we argue also mitigates the need to carry out deadly attacks. Using cross-national data on domestic terrorist attacks committed between 1970 and 2013, a multilevel statistical analysis demonstrates that terrorist attacks in democracies are less lethal. A time-series cross-sectional analysis further reveals that consolidated democracies and harsh authoritarian regimes experience few deaths from terrorism. While democracies experience high volumes of nonlethal terrorism, strong autocracies experience low amounts of deadly terrorism. Thus, there is strong evidence that—in one important sense—democracies are safer from terrorism.

Existe una extensa literatura que examina la relación entre el terrorismo y la democracia. Sin embargo, son pocos los estudios que examinan la letalidad de los atentados terroristas según el tipo de régimen. Este artículo sostiene que el terrorismo que registran las democracias será menos mortífero. La democracia aumenta la oportunidad de que los actores no estatales utilicen el terrorismo para atraer la atención hacia sus causas, lo que, según argumentamos, también mitiga la necesidad de llevar a cabo ataques mortales. Utilizando datos transnacionales sobre ataques terroristas nacionales cometidos entre 1970 y 2013, un análisis estadístico multinivel demuestra que los ataques terroristas en las democracias son menos mortales. Un análisis transversal de series cronológicas revela además que las democracias consolidadas y los regímenes autoritarios duros registran pocas muertes por terrorismo. Mientras las democracias registran grandes volúmenes de terrorismo no letal, las autocracias fuertes registran bajas cantidades de terrorismo mortal. Por lo tanto, hay pruebas sólidas de que, en un sentido importante, las democracias son más seguras frente al terrorismo.

Si une vaste littérature analyse les relations entre terrorisme et démocratie, peu de travaux se sont penchés sur la létalité des attaques terroristes selon le type de régime. Cet article émet l’hypothèse que le terrorisme qui frappe les démocraties est moins meurtrier. En effet, dans la mesure où le régime démocratique augmente les possibilités, pour les acteurs non étatiques, de recourir au terrorisme pour attirer l’attention sur les causes qu’ils défendent, il limite également la nécessité de perpétrer des attaques meurtrières. S’appuyant sur des données transnationales portant sur les attaques terroristes intérieures entre 1970 et 2013, une analyse statistique multiniveau démontre que les attaques terroristes sont moins létales dans les démocraties. Par ailleurs, une analyse transverse et temporelle révèle que les démocraties solidement établies et les régimes autoritaires ne subissent que peu de morts dues au terrorisme. Tandis que les démocraties connaissent une quantité importante de terrorisme non létal, les autocraties fortes ne souffrent que de peu de formes de terrorisme meurtrières. Par conséquent, il est clair que, dans un sens (particulièrement important), les démocraties sont un meilleur rempart contre le terrorisme.

Keywords:

• Democracy
• fatalities
• lethality
• terrorism

Acknowledgments

We want to thank Brian Lai, participants at the 2019 ISA Midwest annual meeting, and the editors and anonymous reviewers at International Interactions for their helpful feedback on earlier versions of this manuscript.

Notes

1 See Chenoweth (2013) for a review of the terrorism and democracy literature.

2 As we discuss later, the results generalize to transnational terrorism as well.

3 However, see Vreeland (2008) for an important critique.

4 Some scholars instead make civilian victimization the defining attribute of terrorism (Thomas 2014; Carter 2016). While we adhere to a different conceptualization, which is closer to the tactical definition of terrorism (Kydd and Walter 2006; Findley and Young 2012; Jones 2017), we restrict our empirical analysis to attacks against civilians in the robustness checks.

5 One exception, discussed below, is research on media freedom and terrorism, although this research almost exclusive focuses on incidence rather than lethality.

6 Shahbaz (2018) provides Freedom House data confirming that autocracies engage in these internet restrictions more frequently and extensively than democracies.

7 And, as Abrahms (2007) shows, democracies can prove to be superior counterterrorists when faced with grave threats.

8 This argument is consistent with studies, such as Chenoweth (2010), that find the generally competitive environment in democracies increases the number of terrorist attacks.

9 Accordingly, it does not matter much should we change this relationship graphically. In other words, a left-skewed relationship between democracy and attack counts is consistent with a monotonic, linear decrease in attack fatalities with increases in democracy and a strong inverted u-shaped relationship between fatalities from terrorism and democracy. We convey a linear relationship in figure 1 for simplicity and conformability with a large body of scholarship on terrorism and democracy.

10 For an exception that also uses the terrorist attack as the unit of analysis, see Conrad and Greene (2015).

11 As is well-known in the terrorism literature, the GTD is missing data for the year 1993. Although several attempts have been made to reconstruct this missing year, we believe that ensuring data consistency is a more serious concern since it is unlikely that some omitted factor related to the year 1993 could confound an otherwise robust relationship between regime type and attack severity.

12 This also follows our conceptualization of terrorism, rather than a specific target.

13 Other attacks is the base category. According to the GTD, these include hijackings, assassinations, hostage situations, infrastructure attacks, and unarmed assaults. It is reasonable to suspect that these types of attacks are systematically less deadly than armed assaults and bombings.

14 Government target includes both attacks against government and military facilities. Besides civilians broadly, attacks against educational and religious institutions are coded as civilian targets. Other targets, the omitted category, include transportation, other nonstate actor, utility, maritime, media, and airport targets, among others.

15 Recent research by Kearns (2019), however, suggests there is a curvilinear relationship between fatalities and whether an attack is claimed.

16 Empirically, there is no analytical or simulation evidence to support zero-inflated models in multilevel frameworks (Moghimbeigi et al. 2008). As the likelihood ratio test below indicates, not taking this multilevel structure into consideration represents a more serious problem. Theoretically, we do not expect that zero values are generated through distinct processes. Rather, we argue that attack fatalities follow from a similar underlying strategic logic that varies across regime type.

17 At Polity score equals 10 (the highest level of democracy) there are several white dots clustered around y equals 250. This indicates that in highly democratic countries few people die from terrorism even when they experience hundreds of terrorist attacks.

18 We eliminate the possibility that this is true for terrorist violence by controlling for even very low-intensity civil wars. We prefer this approach to using alternative Polity measures, since they are only available till 2004 and thus greatly truncate the sample.