Are Politicians More Responsive Towards Men's or Women's Service Delivery Requests? A Survey Experiment with Ugandan Politicians. Forthcoming. Journal of Experimental Political Science. (with Kristin Michelitch).

This study examines whether politicians exhibit gender bias in responsiveness to constituents’ requests for public service delivery improvements in Uganda. We leverage an in-person survey experiment conducted with 333 subnational politicians, of which one-third are elected to women’s reserved seats. Politicians hear two constituents request improvements in staff absenteeism in their local school and health clinic and must decide how to allocate a fixed (hypothetical) budget between the two improvements. The voices of the citizens are randomly assigned to be (1) male-school, female-health or (2) female-school, male-health. We find no evidence of gender bias toward men versus women, or toward same-gender constituents. This study expands on the mixed results of prior studies examining gender bias in politician responsiveness (typically over email) by adding a critical new case: a low-income context with women’s reserved seats.

Shortcomings in Substantive Representation for Women: Accountability Gaps and Policy Preference Incongruence. Working paper. (with Kristin Michelitch). Supplemental Information

We use plenary session meeting minutes to investigate whether and why politicians might better substantively represent same-gender citizens across 49 subnational Ugandan legislatures (where one-third of seats are reserved for women). Male politicians' legislative actions better mirror male citizens' top policy priorities (roads/transport), yet significant gaps exist between reserved-seat female politicians' (RS-females') legislative actions and female citizens' top policy priorities (water). Why? We find evidence that male citizens exert overall more accountability pressures, and male politicians receive more accountability pressures, especially from male citizens. Additionally, male politicians have more congruence with same-gender citizens in their personal policy priorities. Finally, although women's welfare has low salience for both citizens and politicians, RS-females are significantly more active in this domain.

Traditional Leaders

To Whom Are Traditional Leaders Accountable After the Rise of Competitive Challenges? Evidence from Central Malawi. Forthcoming. The Journal of the Middle East and Africa.

Despite a burgeoning interest in studying the effects that traditional leaders have on service delivery and political accountability, scholars have paid little attention to political competition and its role in accountability for the unelected leaders. This article investigates these questions leveraging original surveys collected from 684 village-level traditional leaders (658 for the second round), 680 members of their ruling family, and 669 secretaries of the leaders in Central Malawi. This study first establishes that leadership contestation appeared in 14 percent of the leadership positions, which challenges the conventional belief that such disputes occur exceptionally rarely for such hereditary positions. The in-depth analyses in this article demonstrate that traditional leaders’ legitimacy stems from the support from their ruling family, and that the family wields significant power in the installation and removal of the leader as selectors. This work further finds that traditional leaders experienced competitive challenges exhibit a higher propensity to share the resources at their disposal with their ruling family than other leaders did not encounter the challenges. The findings align with the theoretical expectation that leaders who undergo competitive challenges adopt a heightened sense of the importance of the ruling family and are incentivized to buy their loyalty because securing this support or the lack thereof significantly affects the leaders’ fate and authority. Finally, this article expands on the recent studies concerning competition in traditional leadership by shifting the focus to the ruling family – the kingmakers behind the scenes – and discusses the ramifications of this phenomenon on political accountability.


Disaster and Political Trust: A Natural Experiment from the 2017 Mexico City Earthquake. Working paper. (with Margaret Frost, Carlos Scartascini, Paula Zamora Riano, and Elizabeth Zechmeister).

Political trust undergirds democratic legitimacy, representative government, and the provision of effective public policy. We theorize over how natural disasters shape general political trust, to the extent they undermine confidence in the beneficence and competence of public servants. Further, we consider whether disaster relief can countervail these effects. We test the relationship among disaster, trust, and aid using novel data gathered immediately before and after a 7.1-magnitude earthquake struck Mexico City in September 2017. Our results demonstrate that the earthquake negatively affected several measures of political trust. At the same time, access to disaster relief offset this decrease in trust. The results add to scholarship on the politics of disaster and are important for policy, shedding light on the potential for disaster assistance to restore generalized political trust following a disaster.

Survey Break-offs and Questionnaire Design in Politically Sensitive Contexts. Working paper. (with Elizabeth Zechmeister).

Survey break-offs have implications for efficiency and data quality, but only a small body of research examines their causes and consequences. We present results from a quasi-experiment focused on questionnaire design. While most break-off research focuses on self-administered web surveys, our data are from a national random-digit dial (RDD) computer-assisted phone survey of public opinion in Nicaragua. The country’s politically sensitive and polarized environment created conditions for high break-off rates: 38% overall. The study’s intervention is the placement of personal questions – gender and location (municipality) – at either the beginning (condition 1) or end of the survey (condition 2). Under a responsive design approach, the location of the module shifted mid-fieldwork, an intervention that can be treated as if random under RDD. We find a stark difference in break-off rates: 57% in condition 1 and 29% in condition 2. We show that this latter rate is equivalent to that found in similar phone surveys in less sensitive Latin American contexts. We further show that in condition 1 personal information questions are among the strongest predictors of break-offs, yet these questions are not key predictors in condition 2. In addition to cost implications, differential break-off rates have implications for data – if those inclined to break-off when personal questions are asked at the start are fundamentally different from those who remain. In fact, we find more political and social trust in condition 1: those who remained in the survey after being asked personal questions upfront are more trusting. However, we do not find differences of opinion over a range of other topics included in the survey. In brief, in a politically sensitive context, questionnaires that ask personal information upfront can elicit high break-off rates; further, while not affecting opinion on all subjects, they may inflate inferences regarding levels of trust in a population.