Measuring insecurity-related experiences and preferences in a fragile state : a list experiment in Mali

Measuring behaviors and preferences in times of conflict is of great interest for understanding conflict dynamics and designing conflict-resolution interventions. Yet, data users often cast doubts on the reliability of sensitive self-reported measures, especially in fragile contexts. We study sensitive experiences and preferences related to insecurity in a fragile State - Mali - by explicitly addressing potential response biases using a List Experiment (LE) method. We survey 1,500 individuals across the entire country and randomly assign respondents to answer sensitive questions through the LE or direct questions (DQ) techniques to measure response biases. We focus on three experience-related items (physical assault victimization, firearms' possession, willingness to engage in violence) and two preference-related items (support for the military regime and trust in foreign armed forces in Mali). Results show significant biases affecting responses about preference-related items. Our analysis confirms that popular support for the military regime and mistrust in the foreign armed forces are large, but suffer from a substantial overestimation. Misreporting is not uniformly distributed across the population, but varies depending on gender, education and conflict exposure. Further results suggest that such heterogeneity in response bias can yield fake significant correlations between individual characteristics and sensitive items' prevalence rates depending on the survey technique used.