Donghyun Danny Choi

In response to mounting popular and international pressure, parties in new democracies have increasingly allowed for mass participation in the candidate selection process. While these democratizing reforms should result in the diminishing influence of party elites and the ascendancy of mass preferences over the selection of party candidates, few studies have examined whether this is the case. I investigate this question in Kenya, where major incumbent and opposition parties have adopted primary elections to select their candidates for legislative office. Drawing on insights from qualitative interviews with elected officials, original data on primary elections, as well as a survey and embedded experiment of primary voters, I show that contrary to expectations, parties wield significant influence over the outcome of primaries and that primary voters can be persuaded to select the party's favored candidate over other higher quality aspirants. These findings have implications for our understanding of the relationship between internal party democracy and democratic accountability in the developing world.

When do women policymakers support liberalizing access to legal abortion? While early scholarship on substantive representation showed that women in politics often sought to expand reproductive rights, more recent research indicates that their policy preferences are also shaped by other social identities. We argue that wealth, in particular, will affect women politicians’ support for liberalizing abortion reform when their personal resources determine the extent to which they are constrained by existing laws. We corroborate this expectation through an analysis of a survey and embedded experiment conducted among more than 600 political candidates in Zambia. We show that women candidates with less wealth increase their support for liberalization once exposed to the human cost of abortion restrictions. We show that this effect is conditioned by wealth rather than marital status or education. Our findings underscore how economic inequalities can influence the substantive representation offered by women politicians.

Overcoming America's deep partisan polarization poses a unique challenge: Americans must be able to disagree on policy while nonetheless agreeing on more fundamental democratic principles. We study one model of depolarization--reciprocal group reflection--inspired by marital counseling and implemented by a non-governmental organization,''Braver Angels.'' We randomly assigned undergraduate students at four universities either to participate in a Braver Angels workshop or simply to complete three rounds of surveys. The workshops significantly reduced polarization according to explicit and implicit measures. They also increased participants' willingness to donate to programs aimed at depolarizing political conversations. These effects are consistent across partisan groups, though some dissipate over time. Using qualitative data collected during the workshops, we inductively generate a new theory of depolarization that combines both informational and emotional components such that citizens, moved to empathize with an outgroup, become more likely to internalize new information about outgroup members.

We examine how the expansion of mobile internet infrastructure affects national identity in sub-Saharan Africa. In diverse societies where elections are contested along ethno-communal lines, we argue that access to mobile internet undermines national identity because it facilitates voter exposure to the polarizing tendencies of internet-based social media and communication platforms. Applying a difference-in-differences design on mobile coverage maps and geocoded survey data of more than 50,000 African citizens, we show that access to mobile internet reduces identification with the nation by up to 5-7 percentage points. To establish support for our electoral mechanism, we exploit as-if random variation in the timing of individuals' survey interviews relative to presidential elections, during which we argue divisive and polarizing forces are at their peak. Our analysis shows that electoral proximity intensifies the negative effect of mobile internet. These findings highlight how technological innovations can inhibit the process of state-building in diverse societies.

Are cross-cleavage campaigns effective in polarized societies? While social demographics and electoral rules in many countries compel candidates to pursue votes outside their own identity groups, the efficacy of such campaigns remains unclear in polarized contexts. We argue that cross-cleavage electoral outreach through in-person campaign rallies can inadvertently trigger inter-group differentiation and competition, resulting in the heightened salience of identity and depressed voter support for outgroup candidates. We assess these claims by exploiting the timing of an unscheduled campaign rally held by an outgroup presidential candidate in another ethnic group's stronghold during Kenya's 2017 election. Comparing survey respondents before and after the rally, we find that the outgroup candidate's post-rally favorability significantly decreased among ingroup voters, while the proportion of voters identifying in ethnic terms simultaneously increased. These findings contribute to a more nuanced understanding of the challenges faced in democratic elections in socially divided societies.

How does democratic backsliding affect the behavior of incumbent party and opposition legislators in parliament? We investigate this question in the context of Zambia, which experienced a significant spate of oppression against the opposition party during the Presidency of Edgar Lungu. Using computational text analysis on a corpus of 1,292 parliamentary sessions from 2001—2021, we systematically analyze how legislative debates between incumbent and opposition players evolved during a period of democratic turmoil compared to periods of relative democratic stability. We specifically examine changes in the composition of legislators who speak up during these debates, the topics raised, as well as the extent of incivility across party lines. Our findings reveal how democratic backsliding manifests in the legislative arena, which has implications for freedom of speech and democratic discourse.

Although LGBT+ individuals in some countries around the world are enjoying the gradual expansion of civil liberties and fundamental protections, in many other countries they are experiencing increasing government repression and social sanctions. In this context, it remains unclear to what extent popular attitudes are conditioned by the decisions of state institutions like courts or the social campaigns of civil society organizations -- in either conservative or liberal directions. We assess such dynamics through a three-wave panel survey conducted among a nationally representative sample of Kenyans throughout 2019. Respondents were asked about their attitudes toward LGBT+ individuals, including criminal penalties and social sanctions, before and after Kenya's High Court issued a decision upholding the country's colonial-era anti-sodomy law. Respondents were also randomly exposed to informational and empathy treatments modeled after those employed by LGBT+ rights organizations in Kenya. Our analysis shows that while the information and empathy treatments about LGBT+ individuals did have some sizable short-term effects consistent with prejudice reduction, the effects did not persist beyond the single survey round. Furthermore, results from the post-court decision survey round show that even when respondents were nudged to recall the experience of the information and perspective-taking treatments, they were no less likely to exhibit prejudicial attitudes towards LGBT+ individuals. Taken together, our findings provide evidence for how social campaigns and state institutions may constrain the effectiveness of prejudice-reducing interventions.

Although women have entered government in African countries at an unprecedented rate over the past three decades, it remains unknown to what extent they have acquired the influence necessary to shape policymaking. Are women able to exercise personal influence to the same degree or in the same ways as their male counterparts? We argue that women tend to be less influential than men due to the structure of their personal networks with other politicians. Prior scholarship on African politics has demonstrated that political outcomes depend on the personal ties that connect politicians to one other. Based on a novel network survey among Zambian candidates, we demonstrate that women tend to be peripherally situated within networks. We find that women are systematically less likely to be connected to others in social or work networks among politicians. We also demonstrate that, while having fewer connections than men, women have connections with more important people in both social and work networks.

Political decentralization in African countries has the potential to transform the relationship between citizens and their representatives. But scholars have yet to fully understand whether or how voters adapt to the accountability demands imposed by having to elect representatives at multiple levels of government. Do voter assessments of local politicians depend on coethnicity, as prior studies have shown with national politicians? Are voters willing to update their assessments of incumbent performance based on new information? To answer such questions, we examine how voter evaluations of local incumbents are conditioned by information and partisanship. An experimental survey design presented respondents in two Kenyan counties, Uasin Gishu and Trans Nzoia, with randomized positive versus negative information about the performance of incumbent county governors in the run-up to the 2017 elections. The results show that some voters are willing to update their vote preference based on positive information about incumbent performance. However, these effects are found almost exclusively in Trans Nzoia, where exposure to positive information significantly increases the likelihood of voting for the incumbent even among non-coethnics, non-copartisans, and respondents with prior negative evaluations of the incumbent. In Uasin Gishu, by contrast, we find no such effects. We postulate that the differences between the two counties may be driven by distinct patterns in the distribution of political support for the two incumbents.