Social media can engage and inform citizens, reducing support for elected autocrats who control traditional media. But this threat also encourages such rulers to restrict social media access. In Uganda’s electoral autocracy, where social media content has an anti-government slant, we estimate effects of facilitating and limiting social media access on support for the long-ruling NRM party. Experimental results show that three months of individually-subsidized social media access decreased NRM support among initial supporters. In contrast, VPN users able to maintain access during an election-time social media ban became more supportive of the NRM than individuals less able to circumvent it. The difference partly reflects the government’s nationwide ban both increasing the favorability of content toward the NRM encountered by social media users and eliciting greater backlash among non-VPN users. Our findings illuminate some of the trade-offs associated with restricting social media access for incumbents seeking to maintain popular support.

What accounts for candidate gender differences in electoral success? We argue that parties systematically nominate women to run in districts where the party is less popular, making it harder for them to get elected. Our empirical strategy uses the German mixed electoral system to create counterfactual vote share differences that would occur if women and men were nominated in districts where their parties have equal popularity. Using data on all candidates for the German Bundestag across eleven elections, we document that women’s underperformance, and its variation across parties and election years, is almost entirely explained by women running in less popular districts. In contrast, we find no evidence that voter bias or candidate characteristics play a substantial role. Our argument highlights gendered party gatekeeping by local officials, which increases in district strength, as a key driver of underrepresentation.

Encouraging citizens to apply pressure on underperforming service providers has emerged as a prominent response to the failure of states to provide needed services. We outline three theoretical mechanisms through which bottom-up citizen pressure campaigns may affect service provision and investigate them via a large-scale field experiment in the Ugandan health sector. While we find modest positive impacts on health provider behavior, we find no effects on citizen pressure, utilization rates, or bottom-line health outcomes. Our findings cast doubt on the power of outside actors to generate improvements in development outcomes by mobilizing bottom-up pressure—at least under conditions similar to those in our study setting. Our results underscore the importance of baseline health conditions for the success of bottom-up, citizen-oriented pressure campaigns. Such conditions shape outcomes both across countries and within countries over time, with the latter finding holding important implications for countries undergoing rapid socioeconomic change.

Best Paper Award by the Experimental Research Section, APSA

Preanalysis plan. Vox Dev article. Supported by the Department for International Development (DfID).

This article reviews the recent literature on accountability in developing democracies through the lens of two nested principal-agent problems: the relationship between voters and elected politicians, and that between elected politicians and bureaucrats. We focus on two global trends that we view as reshaping these accountability relationships in important ways: the rise of the internet and social media on the one hand and increasing political polarization on the other. We evaluate the impacts of these developments on the sanctioning of the performance of elected officials, the selection of elected officials, and the agency problem between elected officials and bureaucrats. Rather than offering definitive conclusions, we highlight key trade trade-offs and emphasize that the overall effects are contingent on the status quo. Notably, much of the existing evidence originates from developed democracies, presenting important opportunities for future research to address gaps in understanding accountability in developing contexts.

Concerned with poor service delivery, a large literature studies accountability of politicians to voters. This article instead considers accountability relationships within governments—the ability of politicians to implement policies by holding bureaucrats responsible for their actions. In collaboration with the Ugandan government, I conducted a field experiment across 260 local governments. The objective of the reform was to empower local politicians to exercise closer oversight over the bureaucracy through training and the dissemination of financial information. Lowered oversight costs increase politicians’ monitoring effort and the quality of services, but only in areas where the political leadership is not aligned with the dominant party. In areas under ruling-party control, politicians fear uncovering mismanagement of funds. In contrast to scholars arguing that insulating bureaucrats allows them to do their jobs more effectively, these findings imply that increased political oversight can improve government responsiveness in settings with a modicum of party competition.

Preanalysis plan filed on the AEA registry (ID: AEARCTR-0000402). Supported by the International Growth Centre and Hewlett Foundation. 

Best Field Work Award by the Comparative Democratization Section, APSA

This article introduces a new explanation for why citizens may fail to vote based on government performance. We argue that when politicians have limited capacity to control bureaucrats, citizens will not know whether government performance is a good signal of the incumbent’s quality. We develop a selection model of elections in which policy is jointly determined by a politician and a bureaucrat. When politicians have incomplete power over policy, elections perform worse at separating good and bad types of incumbents. We test the theory’s predictions using survey experiments conducted with nearly 9,000 citizens and local officials in Uganda. We find that citizens and officials allocate more responsibility to politicians when they are perceived as having more power relative to bureaucrats. The allocation of responsibility has electoral consequences: When respondents believe that bureaucrats are responsible for performance, they are less likely to expect that government performance will affect incumbent vote share.

Online appendices. Replication files. Preanalysis plan filed on the AEA registry (ID: AEARCTR-0000767).

What role does information play in shaping mass support in dominant party settings? We conduct a field experiment during the 2016 Ugandan parliamentary elections that provides voters with information about candidates from all competing political parties. Specifically, we produce and screen video-recorded candidate interviews in randomly selected villages just before the election. Voters have lower baseline knowledge about opposition candidates compared to ruling party candidates. We find that the video screening reduced this knowledge gap and caused voters to update more positively about the opposition. Further, those who watched the videos were less likely to vote for ruling party candidates, and those initially leaning toward ruling party candidates were more likely to vote for the opposition. These findings suggest that information asymmetries play a role in sustaining mass support for ruling parties in dominant party settings and that reducing them may strengthen electoral competition.

Ungated version here.

Voters may be unable to hold politicians to account if they lack basic information about their representatives’ performance. Civil society groups and international donors therefore advocate using voter information campaigns to improve democratic accountability. Yet, are these campaigns effective? Limited replication, measurement heterogeneity, and publication biases may undermine the reliability of published research. We implemented a new approach to cumulative learning, coordinating the design of seven randomized controlled trials to be fielded in six countries by independent research teams. Uncommon for multisite trials in the social sciences, we jointly preregistered a meta-analysis of results in advance of seeing the data. We find no evidence overall that typical, nonpartisan voter information campaigns shape voter behavior, although exploratory and subgroup analyses suggest conditions under which informational campaigns could be more effective. Such null estimated effects are too seldom published, yet they can be critical for scientific progress and cumulative, policy-relevant learning.

Informational interventions designed to promote electoral accountability often fail to change voters’ behavior. One explanation is motivated reasoning: voters resist unwelcome information about politicians. We test this explanation directly by reanalyzing the Metaketa I experiments, asking whether voters selectively incorporate and remember the information provided. At first glance, the answer appears to be yes: participants given information about candidates’ policy positions recall it more accurately when it aligns with their own preferences. But the same pattern appears among participants who received no information: voters project their own positions onto candidates they support, and differential guessing—not motivated memory — drives the result. Across four experiments providing incumbent performance information, negative information is more memorable, and this holds whether or not voters support the incumbent. Motivated reasoning of this form is thus unlikely to explain the weak effects of informational campaigns, and pessimism about voters’ capacity to learn unwelcome facts may be overstated. We provide practical guidance for separating motivated reasoning from projection and guessing.

Pricing policy for any experience good faces a key tradeoff. On one hand, a price reduction increases immediate demand and hence more people learn about the product. On the other hand, lower prices may serve as price anchors and, through a comparison effect, decrease subsequent demand. This tension is particularly important for the distribution of health products in low-income countries, where free or heavily subsidized distribution is a common but controversial practice. Based on a model combining the learning aspect of experience goods with reference-dependent preferences, we set up a field experiment in Northern Uganda in which three health products differing in their scope for learning were initially offered either for free or for sale at market prices. In line with prior studies, when the product has the potential for positive learning, we do not find an effect of free distribution on future demand. However, for products without scope for positive learning, we find evidence of price anchors: future demand is lower after a free distribution than after a distribution at market prices.

We evaluate the impact of a health information intervention implemented through mobile phones, using a clustered randomized control trial augmented by qualitative interviews. The intervention aimed to improve sexual health knowledge and shift individuals towards safer sexual behavior by providing reliable information about sexual health. The novel technology designed by Google and Grameen Technology Center provided automated searches of an advice database on topics requested by users via SMS. It was offered by MTN Uganda at no cost to users. Quantitative survey results allow us to reject the hypothesis that improving access to information would increase knowledge and shift behavior to less risky sexual activities. In fact, we find that the service led to an increase in promiscuity, and no shift in the perception of norms. Qualitative focus group discussions support the findings of the quantitative survey results. We conclude by discussing a potential mechanism explaining the counterintuitive findings.

Coverage in Businessweek,  IRIN,  and the Yale Alumni Magazine.