Gender Inequality and State Security: The Effects of Women’s Social Equality on Domestic Terrorism

Abstract

Women’s roles in political violence and terrorism have largely been examined through the agent/victim dichotomy. We suggest that women’s social inequality contributes to both roles of women as victims and as perpetrators of terrorist violence, with both roles ultimately contributing to increases in supplies of domestic terrorism. Particularly, women’s social inequality contributes to terrorism in three ways: it normalizes violence in society, makes women susceptible to coercion from terrorist groups, and results in grievances in the female population that may mobilize them to violence. An in-depth case study of women in Somalia and the quantitative results both suggest that women’s political, economic, and social inequality are associated with higher levels of domestic terrorism. The results show that the impact of women’s social equality through balanced social exchanges in society subsumes the impact of vertical equality measures such as political and economic equality.

Keywords:

• Gender
• inequality
• political violence
• terrorism

Notes

1 See World Back World Development Indicators Database for fertility rate statistics by country by year.

2 Statistics published by the International Labour Organization. January 2016. https://www.ilo.org/wcmsp5/groups/public/–-africa/–-ro-addis_ababa/–-sro-addis_ababa/documents/publication/wcms_445503.pdf.

3 Jessica Davis writes that Al-Shabaab generally leaves women out of tactical roles, however, they form a significant portion of the group’s support base and are crucial to the raising finances for the group’s operations.

4 It is important to note the loss of data in the year 1993 from the GTD. This missing data will be reflected in the descriptive summary visuals but was imputed for the sake of running regression models.

5 The results of the unimputed data show that there were only 29 observations which include data across all variables needed to run the models. Though this may bring up doubt in the confidence of the analysis later on, recall, that there are actually thousands of observations in totality of the dataset. One argument that might better explain this is that since equality scores are unlikely to change rapidly, based on the structural distribution of the data, we should be able to pick up on commonalities across the panel. While it is possible that the results may still be biased due to MNAR, we have taken previously discussed steps to mitigate that as a problem. This is more true than say the last observation carried forward strategy, in which we would take the last observation for a country and carry it until we have another observation. The structure of the data makes it reasonable to assume that imputations would capture variations between countries in a manner we want, since code is a variable used in the imputation to capture those commonalities. While we could hypothetically run a regression with only the completed observations (see Appendix 3), we are nearly guaranteeing that the results are biased in favor of only the completed observations. Since there are few observations compared to the data, an imputation strategy is recommended in lieu of the default, which removes any observation with a missing variable.

6 This is especially true in the case of datasets where advanced democracies are included, However, as long as there are variables related to determinants of missing data, like income and democracy, the risks of MNAR data are mitigated. (Lall, 2016) Similarly, the use of many predictors mitigated the risks of MNAR data. (Gelman & Hill, 2007, White et al 2010). So while the data are missing in many instances, best practices were followed to avoid the risks of MNAR data, mitigating potential issues where they stand.