![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
ABSTRACT
Fraudulent elections can reduce citizen trust in elections and other political institutions. But what about the impact of contentious elections that resolve successfully, leading to democratizing change? Do national movements toward democracy trump individual experiences with electoral manipulation? Using public opinion survey data collected before and after the 2004 Orange Revolution in Ukraine, we evaluate changes in voter confidence in electoral practices, political institutions, and democracy. Although national trends show increased voter confidence overall, subnational variation suggests pervasive partisan differences in opinions about election quality and institutional confidence. Remarkably, we find that direct exposure to fraud matters far less than anticipated; voters who were personally exposed to fraud felt no more or less confident than their co-partisans. We show that partisanship and the national electoral context may interact in ways that complicate the effects of democratizing elections, suggesting important avenues for future research.
Acknowledgements
We would like to thank Rakesh Sharma and Lauren Serpe of the Applied Research Center at IFES for providing access to the data used in this paper. We also thank participants at the August 2013 EIP “Emerging Challenges of Electoral Integrity” Workshop and the July 2016 EIP/IDEA “Contentious Elections, Conflict, and Regime Transitions” workshop for comments on an earlier versions of the paper, as well as the anonymous reviewers for their insightful feedback. Any errors in analysis or interpretation are solely the responsibility of the authors.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes on contributors
Elizabeth Iams Wellman is a Ph.D. candidate in Political Science at Yale University. She studies comparative democratization, migration and citizenship, and transnational political participation.
Susan D. Hyde is Professor of Political Science and the Avice M. Saint Chair in Public Policy at the University of California, Berkeley. She is an expert on international election observation, election fraud, and democracy promotion.
Thad E. Hall is a senior political scientist at Fors Marsh Group. He is the author of numerous books and articles on election administration and voting, as well as public policy-making in the United States.
Notes
1 International Foundation for Electoral Systems (IFES) did not conduct a survey between the second and third rounds; citizens were asked about both rounds retrospectively on the February 2005 survey.
2 We use oblast as a proxy for partisan support for numerous theoretical and empirical reasons outlined below.
3 We thank a reviewer for their clarifying comments on this central implication.
4 Author's calculation using version 4 of the NELDA data (Hyde and Marinov Citation2012). Out of all elections since 1990, 634 out of 1329 experienced reports of fraud (NELDA 11 or 28), violence (NELDA14 or NELDA 31), or post-election protest (NELDA29).
5 Note that we do not measure changes in the size of the winner–loser gap, but build on insights from this literature.
6 These reports are available in the supplementary online appendix.
7 Note that an important limitation of the data is that TNS did not conduct a survey between the second and third rounds of the election. Rather, assessments of citizen attitudes about the quality of both rounds were asked retrospectively on the February 2005 survey. It is important to note that differences in voter recall about their own second-round attitudes could correlate with partisanship.
8 For detailed information on the survey methodology, please see TNS Ukraine (Citation2004, 5–8) and TNS Ukraine (Citation2005, 5–8) in the supplementary online appendix.
9 For example, see Beissinger (Citation2013).
10 For example, a working paper using IFES survey data from Nigeria (Shah and Kerr Citation2017) takes a similar methodological approach to ours, analyzing how different types of electoral victimization are linked to assessments of electoral quality rather than relying on a single or index variable.
11 Self-reports of experiencing fraud can be biased and may reflect a misunderstanding of what is considered fraud in a legal sense.
12 Results comparing voters exposed to fraud to the overall survey population are in the SA. Voters exposed to fraud are not statistically different in their perceptions than the survey population with the exception of the questions noted.
13 According to the official third-round results, of the 20 oblasts included in the survey, the vote margin between the winning and losing candidate ranged from 29% to 93%, with a majority of oblasts exhibiting a margin >60% (“Ukraine. Presidential Election. 2004 | Electoral Geography 2.0” Citation2014). A drawback of using the oblast proxy is that it also includes survey respondents who voted for the candidate that did not win in their oblast (N = 208) as well as those who did not vote at all (N = 198). We find political minority voters shared similar confidence levels to their co-partisans. The confidence levels of non-voters were not statistically different from either the overall population or their oblasts. Figures are located in the SA.
14 While the February 2005 survey asked respondents whom they voted for during each round of the election, the survey taken in October 2004 did not include a question on likely vote choice.
15 We present figures showing clustered standard errors in the SA. This increases the variance of our estimates, though the substantive interpretation remains largely the same. Recent discussions of robust inference note that while fixed effects do not control for within-cluster correlation of errors, there are certain cases where only including cluster-specific fixed effects may suffice, particularly where a common shock may be driving within-cluster error correlation (Cameron and Miller Citation2015).
16 The empirical analysis could estimate each response on the Likert scales, or employ alternative models (e.g. ordered logit).
17 The survey conducted in February 2005 did not include questions about confidence in the Supreme Court.
18 As the February 2005 survey asked questions about attitudes following Round 2 and Round 3 at the same time, opinions about the short-lived Yanukovych victory may have been tempered in retrospect.