Go to the U of M home page


Wednesday, March 2, 2016

"What Before Why: Taking Descriptive Inference Seriously in Quantitative Conflict Studies"

Databases have growing importance in human rights studies.  Researchers rely on databases that count human rights violations and violence, such as homicides, sexual assaults and other crimes.  Yet just how reliable are these counts? 

On February 15, the Minnesota Political Methodology Colloquium and the Human Rights Program co-hosted Amelia Hoover Green, Assistant Professor in the Department of History and Politics at Drexel University, where she discussed her work regarding quantitative conflict studies.  After observing a trend of inconsistent findings between various data sets studying the same phenomena, Green began a project to critically assess these inconsistencies.  

Hoover examined research in the Colombian conflict, specifically eight different data sets recording conflict related homicides. The discrepancies between the data sets stemmed largely from debates over definitions and technological imbalances between research groups. The researchers she observed also failed to investigate who was not reporting to them and why, yet would still claim a comprehensive data set. After looking at another group of data sets regarding sexual violence in Colombia, such problems proved to be even more severe. With a serious mistrust between victims and state institutions, the threat of danger prevented victims from reporting; victims also faced logistical issues. Consequently, reporting rates on sexual violence remained incredibly low, despite researchers claiming complete knowledge of the rates of these crimes.

After looking at these data sets, Hoover carried out statistical analysis to assess homicide data from five Colombian sources and discovered that undercounting and underreporting were hindering effective data collection. What can be done about such problematic data sets? Hoover offered some potential solutions including quantitative approaches such as multiple systems estimation, simulations but also the use of survey data and ethnography.  Regarding the ethnographic data, Hoover stressed the importance of knowing one’s case, including the effects of cultural and political structures in place to account for low reporting rates.  Hoover also highlighted the importance of practicing humility when approaching this kind of data collection and acknowledging the barriers to accessing credible information and reporting rates.  With time, researchers can observe which variables in the quantitative design will still hold as they add in the missing pieces of the data. Researchers must also understand the political power dynamic of reporting, especially when using data retrieved from government agencies.  

Hoover’s conclusions produced a series of implications for human rights advocacy in these contexts. Rather than perpetuating statistical data collection error, human rights organizations should focus on documenting violations on a case by case basis to the best of their ability.  Disseminating this information can serve to motivate public discussion or call attention to a particular set of problems. Moreover, NGOs can be critical of quantitative methods while also contributing to this sort of research, due to their expertise and local knowledge of complexities on the ground. Hoover embodies this critical approach as a field consultant for the Human Rights Data Analysis Group, providing technical support to data collection of human rights violations, conflict, and testimonies in war crimes trials. 

-Written by Marie-Christine Ghreichi