The main goal of my current research is to build valid theories explaining how various aspects of ethnicity shape political, conflict and development processes. In particular, my work on Africa addresses why ethnic diversity is linked to the factors of inefficient resource distribution and underdevelopment, namely (1) clientelism and distributive politics, (2) ethnic voting, and (3) ethnic conflict.

Publication Under Review
  • "Facing Risks Together: Preference for Co-Ethnics and Neighbors in Informal Savings Groups among the Urban Poor in Kenya." Social Science Quarterly
  • "Public Opinion on U.S. FDI Outflows to Mexico and China: A Survey Experiment on the Effects of Information.” (with Youngchae Lee and Scott Huffmon) Political Studies Review

​Dissertation:  Ethnicity, Distributive Politics, and Voting in Uganda
Committee: G. Bingham Powell, Jr. (chair), Bonnie Meguid, Matt Blackwell, Robin Harding
Summary: My dissertation demonstrates that voters' ethnic preferences for different ethnic groups determine the likelihood of voters’ defection based on clientelistic promises, the efficacy of clientelistic promises made by politicians, and the politicians’ electoral incentives to target non-coethnic voters with public goods distribution. The first paper shows that while the preference toward coethnic candidates tends to be persistent, the preference for a non-coethnic candidate is often a reflection of other voting predictors affected by voters' ethnicity such as partisan affiliation and evaluation of the incumbent's performance. In the second paper I conduct an original survey experiment in Uganda to show that voters' ethnic-preferences determine the likelihood of voter defection once a clientelistic appeal is made by a politician. In the third paper, I argue that only under rare conditions do politicians have incentive to target their own strongholds with goods distribution and that they use the ethnic composition of constituencies to assess their strongholds, opposition strongholds, and swing areas. In all four types of public goods considered, the incumbent's ethnic stronghold was not found to be favored in resource distribution. 

Working Papers
Ethnic Inclusion as a Campaign Strategy in Africa: A Survey Experiment in Kenya
The study of ethnic mobilization helps explain why politicians mobilize support from the ethnic groups to which they belong. However, politicians in multiethnic democracies use ethnic appeals more broadly than simply mobilizing their own ethnic groups. In particular, an increasing number of African politicians make ethnically inclusive appeals to gain support from a broader range of ethnic groups. Using an original survey experiment carried out in Nairobi, Kenya, this study examines the effect of politicians’ ethnic inclusiveness on voters’ support, and shows that the respondents in Nairobi rewarded ethnically inclusive politicians. However, the effectiveness of ethnic inclusiveness was found only in ethnically diverse constituencies, and not in ethnically homogeneous constituencies. I argue that voters in diverse constituencies find broad and inclusive appeals to be more viable, compared to ethnic mobilization that targets a single ethnic group. This leads voters to reward inclusive candidates for having a more viable electoral strategy. The findings explain why ethnically inclusive appeals are used by politicians in Africa by demonstrating that ethnic diversity in constituencies strengthens the efficacy of ethnically inclusive appeals.

Power-sharing and Exclusion: How a Leader's Opportunity to Exploit a Commitment Problem Leads to War (Coauthored with Shawn Ramirez)
Why would a leader politically exclude a group, when exclusion makes war more likely? We use a formal model to establish the relationship between power-sharing, exclusion, and war. The model generates the theory that: since war turns power-sharing into an all-or-nothing game, and a high capacity group cannot commit to peace if power is reduced, a leader will incite a war, by excluding a politically-powerful group, as a gamble for full control. We specify conditions for exclusion, political downgrades, and war. We show that the likelihoods of war and exclusion for politically relevant ethnic groups from 1946 to 2010 are accurately predicted. Our use of generalized additive models flexibly demonstrates the anticipated non-monotonic effects of capacity and prior power in two dimensions. This research implies that power-sharing must overcome the commitment problem, giving incentives to high capacity groups to maintain peaceful power-sharing even with diminished power, to promote democratic stability.

Is State Incapacity Necessary for Civil Conflict? (Coauthored with Rob Carroll)
Some argue that civil wars are more likely to begin when governments are too weak to deter them. Their state capacity argument stands in opposition to economic theories of civil war. While both theories have been influential, no attempt to adjudicate between them has employed appropriate empirical models aimed at assessing the relative effects of capacity and economics. We remedy this shortcoming by employing a split population model that explicitly links state capacity and the economy; this model takes the state capacity argument at its word and is more logically consistent than their traditional models. Substantively, we find that state capacity, when appropriately modeled, does not predict onset, but other factors like agricultural salience, mountainous terrain, and ethnic fractionalization do.

Voting for Non-Coethnics: A Survey Experiment in Uganda
Despite a standard assumption in the ethnic voting literature that voters would not find noncoethnic candidates’ clientelistic promises credible, why do presidential candidates in Africa frequently target non-coethnic voters with clientelistic goods and successfully gain their votes? Based on an original survey experiment conducted in Uganda, this paper provides an explanation that ethnicity conditions the effectiveness of clientelistic promises through their effects on the share of swing voters. Targeting non-coethnic voters, especially those without a coethnic candidate, is effective in gaining votes because such voters are less likely to have a strong ethnic preference between candidates and thus are more likely to be affected by clientelistic appeals. The findings also suggest that neither ethnic geography nor the type of clientelistic appeal, whether ethnic or geographic, is deterministic in affecting voters’ preferences.

Distributive Politics, Local Service Delivery, and Voting in Uganda
This paper examines the conditions under which a vote-seeking incumbent provides a targeted benefit for his own strongholds, swing areas, and opposition strongholds, where the incumbent and opposition strongholds are measured by (a) the previous electoral outcomes, and (b) the ethnic composition of electoral areas. In this paper, I argue that the incumbent seldom faces incentives to target his own stronghold in Uganda. The empirical examination of how electoral calculations influence the allocation of local service delivery, including electricity supply, local public roads, health and medical facilities, and local primary universal education confirms this prediction: in all four cases of local service delivery in Uganda, the incumbent’s stronghold has not benefited from the allocation of service delivery, whether the incumbent’s stronghold is measured based on the previous electoral outcome or the ethnic composition of the electoral area. In particular, the incumbent’s ethnic stronghold, where the electoral area is predominantly the Banyankole, was less likely to experience improved service delivery of electricity supply and local public roads compared to the previous election cycle.