Publications

How and why do incumbents use local elections as tools for subverting democracy and establishing party dominance? Integrating literatures on political violence, and decentralization, I argue that incumbents rely on violence to bar opposition candidates from contesting in local elections. Drawing on the strength of local networks, incumbents prioritize their own strongholds for such coercion, which leads candidates to withdraw and seats go uncontested. I explore my argument with data on local elections from West Bengal. A state in India that has held elections at the local level since the 1970s, but where competition is spatially uneven, resulting in a significant number of uncontested seats. I test my theoretical expectations with disaggregated data on competitiveness and uncontested seats for more than 3,000 local electoral units. My results show that incumbent-controlled units have a greater probability of going uncontested. I complement these findings with qualitative interviews from political elites and non-elites, which show that violence is the mechanism through which incumbents establish local dominance. These findings have important implications for work on decentralization and subnational authoritarianism.

Violence against voters is used strategically during elections in many countries. While our understanding of overt and lethal forms of electoral violence has advanced greatly, we know much less about the micro-foundations of threats and intimidation. We present new theory and evidence on the repertoire of electoral intimidation, suggesting that threats can be used to deter rival party supporters from voting but also to mobilize citizens to vote for a particular party. We expect these strategies to unfold in the same electoral context, but differ in targeting and incidence; while threats to demobilize are concentrated in closely contested areas and occur more frequently overall, threats to mobilize target fewer voters and are centered in parties’ own strongholds. Recognizing the difficulty of surveying citizens about sensitive experiences, our empirical analysis relies on a list experiment embedded in an original survey conducted after the 2019 elections in the Indian state of West Bengal. The experiment allows for eliciting truthful responses about the prevalence of voter intimidation in a tense electoral setting, where respondents might have incentives not to reveal such experiences. Our findings support our expectations, showing that threats to mobilize citizens for a candidate or party are less prevalent than those aimed to demobilize them from voting altogether. We also find that citizens in party strongholds are more often targeted to vote for a particular party, whereas those in contested areas more often report violence intended to keep them from voting. These results show that areas without overt violence may nevertheless be highly coercive, underlining the importance of studying the full repertoire of electoral violence.        

This paper examines the use of misinformation for polarizing voters on a radical new issue. While political parties routinely employ misinformation, its strategic purpose and effectiveness are not well understood. Drawing on work on issue entrepreneurship and polarization, we propose that opposition parties are most likely to rely on misinformation as a strategy to introduce radical new policy positions. Disadvantaged by the status quo of issue competition, opposition parties hope radical issues will polarize the electorate and draw in new voters. Misinformation is an attractive strategy for doing so because it is available to outsiders, allows for direct communication with voters, and circumvents traditional media and gatekeepers. Our empirical analysis draws on a pre-registered vignette experiment embedded in a representative survey conducted after West Bengal’s 2021 elections. In this election, the opposition Bharatiya Janata Party (BJP) sought to polarize voters on the state government’s role on religion. While a majority of voters preferred a secular government, the opposition spread false news alleging government favoritism towards religious out-groups, aiming to polarize voters into secular and majoritarian camps. Our experimental design mimics this strategy, randomly exposing subsets of voters to claims about out-group favoritism, in-group favoritism, correction treatments, and a control condition. We find that misinformation failed, for several reasons. First, the new issue reinforced partisan cleavages instead of cutting across them, persuading only BJP voters. Second, among BJP supporters, misinformation convinced only a subset of voters, effectively splitting the party into two camps. Finally, while some BJP voters were initially swayed by misinformation, they reverted back to secularism when exposed to a correction. Our study illustrates the incentives behind misinformation and the conditions under which it is likely to work, clarifying the supply-side dynamics of political misinformation.          

Existing studies show that voters not only dislike violence but also sanction parties and candidates for using it. But how do we reconcile these findings with the fairly frequent incidence of political violence in democracies? Why do voters in contexts with free and fair elections, mature parties, and high electoral competition vote for violent politicians? Departing from earlier work, we suggest that core and swing voters become more rather than less supportive of parties when exposed to information about violence, leading sanctioning to fail. During campaigns, parties instrumentalize violence for electoral benefit, constructing narratives around violence that portray it as a response to a threat or injustice. These grievance frames lead voters to approve of violence against out-groups. Moreover, when presented with a rationale for violence, voters may not only support violent actions by their party, but could even sanction peaceful responses. Empirically, the paper relies on pre-registered vignette experiments embedded in a representative survey of registered voters. We field our survey during the 2022 elections in Uttar Pradesh, a state of more than 200 million in India. These elections fit our scope conditions of free and fair elections while also experiencing persistent low-intensity violence against religious and ethnic out-groups. Our findings support expectations; when provided with a rationale for violence, voters are more likely to approve of it, and this finding holds for all voters. We also find support for sanctioning of peace among co-partisans, although this effect is limited to voters belonging to the majority group; i.e. Hindus. Specifically, BJP voters, upper caste Hindus, and non-Yadav OBCs become more supportive of their party when exposed to violent involvement of co-partisans, while they become less supportive if their party promotes peace.          

Incumbents often sponsor violence in elections but do not typically engage in it themselves, instead delegating violence to security forces, armed groups, party wings, or regular citizens. This variation is poorly explained by existing theory, which has privileged incumbents’ incentives to plausibly deny their involvement. What explains variation in incumbents’ choice of violent perpetrator? This paper develops a novel theory centered around unevenness in incumbents’ access to rents. Pitched at the subnational level, I argue that variation in rents helps explain why some incumbents recruit violence specialists while others rely on co-ethnics or party supporters. Incumbents with access to large rent can afford to hire costly yet effective violence specialists. In contrast, those with limited rent recruit cheaper but less capable alternatives such as co-ethnics and party supporters. I probe the plausibility of the argument with over 100 interviews conducted with politicians, journalists, voters, civil society members, and party supporters in three Nigerian states: Lagos, Rivers, and Nasarawa. I  triangulate interview findings with newspapers and observer reports. Findings show that in Lagos and Rivers, incumbents hire violence specialists such as transport workers and cult groups because have access to large rent, while those in the low rent state of Nasarawa must rely on party supporters to perpetuate violence. The study contributes to the literatures on decentralization, joint production of political violence, and the resource curse.

It is well known that political violence hardens existing identities and deepens political cleavages. Departing from conventional wisdom, we argue that violence can be used to transform rather than only maintain existing lines of political conflict. We propose that political parties can sponsor violence around a politically dormant social cleavage to establish and strengthen a relationship with those sharing the social identity. Citizens exposed to violence begin to identify more closely with the secondary identity, become more hostile towards the out-group, and increase their support for the political sponsors of violence. These shifts are the result of narratives around violence shared by politicians, who portray violence as necessary to protect co-ethnics. We explore the empirical implications of our argument in India, relying on the subnational expansion of India’s largest party as a test case. We establish individual-level effects on identity and polarization through original surveys with citizens in the proximity of religious rallies, some of whom were exposed to violence while others were not. To demonstrate changes in voting behavior, we rely on polling station data to compare vote shares in localities that witnessed violent rallies before elections to those experiencing them after. Finally, we use process tracing of campaign speeches to establish political parties’ use of violence as a strategy of cleavage transformation.