Research

Dissertation

Why Women Vote: When and How Clientelism Closes the Gender Turnout Gap (Book Project)

When do women turn out at equal rates to men? My dissertation re-investigates this question in the context of developing countries. Existing theories of women’s political participation are largely resource-based, yet in many developing countries, women turn out at par with men in the face of low levels of economic development and female labor force participation, and despite gendered differences in individual-level resource endowments. Based on an in-depth investigation of India, I argue that there is a second path to women’s equal political participation that does not rely on individual-level resources, but instead depends on clientelism and household support for female turnout. Where households are supportive, they can bridge the resource gap for women. Household support, in turn, depends on high levels of clientelist returns to a vote. I provide several pieces of empirical evidence from India consistent with this theory, based on two original surveys and a novel panel dataset on the clientelist party mobilization. I show that female turnout is higher in a poorer and more clientelist state than in a better developed but less clientelist state, and that household support for female turnout – but not other forms of political participation – is high under clientelism. I also demonstrate that increases in levels of clientelist mobilization – measured as a rise in the number of ethnic groups targeted by clientelist parties – leads to smaller gender turnout gaps at the constituency level across several states. My research has important implication for our understanding of the relationship between development and female political participation, as well as the consequences of clientelism.

Download Introduction

Working Papers

How Clientelist Party Mobilization Closes the Gender Turnout Gap: Theory and Evidence from India (JMP)

I argue that there is a second path to women’s equal electoral participation that does not rely on individual-level resources, but instead depends on clientelist mobilization in combination with household preferences over female turnout. Where households are supportive of women’s participation, they can bridge the resource gap to enable female turnout even in the most unlikely of contexts, namely where women face a lack of resources and high costs to participation in public life. Households will only be supportive when returns to a vote are high enough to turn votes into assets and therefore compensate households for resources spent or social costs incurred for women’s participation in public life. I test this theory using a novel panel dataset on the incorporation of ethnic groups into clientelist parties from 1977 to 2007 across two major states in India. Incorporated ethnic groups expect access to state resources after elections, meaning returns to a vote are higher for incorporated groups. Households from incorporated groups, therefore, should support female turnout; where a large enough share of the population belongs to incorporated groups, the gender turnout gap should close. I show that, indeed, an increase in the number of incorporated ethnic groups indeed leads to smaller gender turnout gaps at the constituency level. This result is robust to a number of different specifications, and not easily accounted for by alternative explanations, such as economic development, migration, or quotas.

Download Job Market Paper

How Faulty Voter Lists Disenfranchise Women in India

Voter roll quality is a major factor in determining both turnout and democratic legitimacy in a country. While under-enrollment and the extent of “deadwood” on voter lists have come under considerable scrutiny in the United States, the accuracy of electoral rolls in developing countries has not attracted the same scholarly attention so far. That is despite the fact that faulty lists hinder the electoral process, disenfranchise eligible individuals and distort official turnout statistics. Using original data from two full village censuses and a large-scale household survey, I examine the quality of the electoral roll in India’s largest state, Uttar Pradesh. I show that voter lists are considerably inflated in Uttar Pradesh; that electoral rolls include more obsolete records of females than males; and that actual participation rates far exceed official accounts. The reason for the inflation of voter lists, I argue, lies in the incentive structure for low-level officers in charge of maintaining the decentralized rolls.

(Manuscript available upon request.)

Voting as a Family? Investigating Differences in Political Preferences Within Household Units in Delhi (with Rahul Verma, CPR, Delhi)

How important is the household in the study of political behavior, especially in the context of developing countries? A vast literature on policy preferences and vote choice, often based on the US case, suggests that these decisions are taken at the individual level and affected by a number of personal factors, such as income, education or gender. At the same time, research in economics and sociology stresses the importance of the household as a decision-making unit in its own right, sometimes over-riding individual members’ preferences, depending on their relative agency within the family. Studying the extent of intra-household (dis)agreement on political preferences and behaviors empirically runs into important data limitations, as most election surveys sample only one respondent per household. We implemented a survey around the 2020 Delhi Assembly Elections that interviewed either two or three members of voting age per household. We find considerable disagreement within households across a number of preferences. All households in our sample are ethnically homogeneous; we find that households where all members identify strongly with their community on average are significantly less likely to exhibit disagreement. Intra-household differences in individual-level resource endowments with education or political knowledge are not driving disagreement within families.

(Manuscript available upon request.)

Ethnic Identity Salience and Attitudes Toward Women

Are political attitudes affected by which of their multiple ethnic identities is most salient for a person? If each ethnic identity comes with certain social norms that are practiced and enforced by an ethnic group, then making one identity salient over another should restrict the choice space of acceptable behavior and attitudes accordingly. I test this theory using particularly meaningful attitudes: men’s attitudes toward women’s political and economic participation, which have been shown to affect women’s labor force participation and partaking in the political process around the world. Using an original online survey experiment in India, I show that it indeed matters which ethnic identity is most salient for a man. Males who were primed on their religious identity showed less support for women’s political and economic participation than those in the control group. The results were less clear-cut for the treatment that primed men on their caste identity: while males self-identifying as Brahmins, members of other forward castes or Dalits increased their support for women’s participation when treated, men who belong to the Other Backward Classes (OBC) lowered their support when treated.

(Manuscript available upon request.)

Work in Progress

The Challenge of Measuring the Gender Turnout Gap

Studies of women’s electoral participation face a major data constraint: very few countries collect and publish gender-disaggregated turnout data. Accordingly, previous scholarship mostly relied on survey data to gauge women’s electoral participation relative to men’s, despite the fact that survey respondents notoriously over-report having voted and we know nothing about whether this behavior is gendered. I explore this potential source of error by comparing a new dataset of administrative turnout data disaggregated by gender and survey-based estimates of turnout among men and women for about two dozen countries worldwide. I find that men overreport their turnout at even higher rates than women, obfuscating the gender turnout gap, that is, the difference between male and female turnout rates. This finding has important implications for scholars studying electoral participation, as survey results can either hide or exaggerate actual gender differences in turnout.