My research investigates the causes of ethnic conflict, political violence and individual participation in high-risk collective action, with an emphasis on field methods, survey design and implementation, and a regional focus on Sub-Saharan Africa. Explaining participation in violence often requires the researcher to collect data on highly sensitive topics in data-poor environments. My research agenda suggests that these need not be insurmountable obstacles, however, and that we should not shy away from asking difficult questions about participation in violence.

Information about three current research projects can be found here. This statement of future research outlines two projects in the planning stages that explore political violence from different angles. The first investigates theoretical and empirical links between the potential causes of riot onset and participation. The second project examines participation in the context of an ongoing civil war in the Niger Delta.

Dissertation

My dissertation asks two questions about the causes of ethnic violence in diverse societies. First, who are the people who take to the streets and commit acts of violence during the chaotic chains of events we know as ethnic riots? Second, why does this set of people ultimately decide to riot? Most contemporary studies of ethnic conflict overlook these questions and focus instead on the conditions and incentives driving political elites to attempt to instigate violence. This literature struggles to explain why ordinary people choose to accept the risks and costs involved in carrying out violence on a local scale.

The empirical strategy used in this project draws on qualitative and quantitative evidence to help isolate the reasons why people participated in two large-scale violent Christian-Muslim riots that took place in northern Nigeria since 2000. Working with a team of outstanding Nigerian research assistants over the course of six months in the field, I conducted about 40 in-depth interviews with riot organizers and participants and implemented an original survey of over 800 individuals who chose to (or chose not to) participate in riots in the cities of Kaduna and Jos. The survey contains direct measures of past participation in violent events, makes use of new methods to protect respondent anonymity, and relies on a novel sampling strategy in order to locate rioters and elicit honest responses from them.

This micro-level data is used to test an original argument about the joint influence of poverty and individual centrality in local social networks to explain why some individuals take to the streets during times of crisis, and others stay on the sidelines. My dissertation argues that, given poverty, important opportunities for violence arise from day-to-day interactions during participation in neighborhood-level social networks.

My dissertation outline is available here.

"Who Riots? Explaining Individual Participation in Ethnic Violence", is a writing sample from my dissertation.

"External Validity: Violent Demonstrations Across Nigerian Local Government Areas", is Chapter 6 of the dissertation.

Papers

"What the Numbers Say: A Digit-Based Test for Election Fraud Using New Data from Nigeria" (with Bernd Beber), under review. We make use of psychological biases to detect election fraud by analyzing digit patterns on return sheets: People favor small numbers and avoid repetition when manipulating numbers. The project was linked on Freakonomics, by way of Chris Blattman's blog. For our presentation at APSA in August 2008, we added an analysis of data from Chicago in the 1920s. Thanks to Kevin Corder for sharing this data with us.

"A Snowball's Chance in Nigeria: Finding Rioters using Respondent-Driven Sampling" (with Rachel Schutt). We tackle the problem of how to correctly analyze a snowball sample with two general characteristics: (1) the number of randomly sampled seeds is high and each referral chain is short, and (2) respondents cannot be recruited only from the target population. The solution we develop is to collect a random sample alongside the snowball sample, and use the random sample to construct a set of weights to correct for possible biases in the snowball sample.

"Individual Participation in Violent Demonstrations in Nigeria". Using nationwide household data from the Nigerian Living Standards Survey (NLSS), survey data from Afrobarometer and unique data on characteristics of Nigeria's 773 Local Government Areas (LGAs), I find that people who are more active in community-level networks and those who live in areas with no clear ethnic majority are more willing to participate in violent demonstrations than others.