ABSTRACT
Existing research identifies several factors that shape civilian targeting, including competition between armed groups. However, the role diffusion into new territory – as opposed to continued presence in long-established areas – plays is relatively underexamined. This study addresses this through a territorial analysis of eight high-activity African rebel groups, using Armed Conflict Location & Event Dataset data. It finds that groups are more likely to target civilians in new territory, and in areas with a proliferation of non-state armed actors. The study attributes this to competition and signalling incentives. This has implications for understanding civilian targeting and protection in territorially expanding conflicts.
Data availability
The dataset and log files are available in the Figshare repository.
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Notes
1. Source: Raleigh et al. (Citation2023), analysis author’s own.
2. Although a full discussion of factional dynamics within and between JAS and ISWAP is beyond the scope of this study, scholars including Warner and Lizzo (Citation2021) have called for greater attention to divisions between the groups and for research to avoid over-aggregation. In this study, the two groups are aggregated in the base model, in part owing to challenges of attributing violence to one or the other definitively (see ACLED Citation2021a), and also due to conceptual issues associated with defining ‘new’ and more established territory when associated with sequential factional splits. However, to account for the potential impact of this, sensitivity analyses re-run the models with a sub-set of (pre-split) violence, detailed further below.
3. Note that the LRA were also active prior to the dataset period, although their significant expansion in territory post-2000 means analysis of new and more established territory can still be carried out.
4. To contextualise these activity levels, the ACLED dataset records 4,895 distinct, named non-state armed groups across Africa in the same time period.
5. In a small number of observations (n = 56), armed groups were recorded as inactive in a particular subnational territory in which they had previously been active for five or more years. In a sub-set of these (n = 8), actors subsequently re-engaged in these spaces. In these cases, activity in a sub-national territory after a prolonged absence of five or more years is treated as activity in a ‘new’ territory, as an absence of this length is likely to encounter the same operational, informational and relational challenges as in territory in which they have never been active. I am grateful to participants in the VERSUS Workshop on Subnational Violence and Authority Patterns (Spring 2022) for their helpful suggestions on this point.
6. Note that controlling for an actor’s overall activity level may introduce endogeneity concerns as the total actor event count is also included in the denominator of the VAC proportion (Lis et al. Citation2021). Therefore, in robustness tests, models are re-run without this control, detailed further below.
7. ACLED Event ID BFO2058.
8. See, among others, ACLED Event ID BFO2042.
Additional information
Funding
Notes on contributors
Caitriona Dowd
Caitriona Dowd is an Assistant Professor and Ad Astra Fellow in the School of Politics and International Relations, University College Dublin. She previously worked in humanitarian response. Her research centres on violent conflict, humanitarian crises and civilian targeting.