Article Summary
Introduction
On traditional survey measures, Americans generally report that maintaining democratic principles and practices are important to them (e.g., “Democracy may have problems, but it is better than any other form of government. To what extent do you agree or disagree with this statement?”). Graham and Svolik argue that these measures overreport commitment to democracy. In order to assess how highly Americans value democracy, the authors test whether voters reduce support for candidates that take undemocratic positions, even when that candidate represents their party or policy preferences. They also test whether centrist voters will be stronger checks on anti-democratic candidates than more ideologically extreme voters, whether voters will be more forgiving of anti-democratic candidates from their own party, and whether larger distance between opposing candidates’ platforms leads to more tolerance of anti-democratic positions. Finding that voters are unwilling to sacrifice policy preferences or partisan alignment to punish for anti-democracy candidates would suggest that traditional survey measures dramatically overestimate Americans’ support for democracy. As voters are a primary check against anti-democratic politicians, the U.S. might be more vulnerable to democratic backsliding than revealed in previous work.
Analytical Approach
The authors first develop a formal model “of the public as a democratic check.” They then test the model through two empirical studies. In particular, to assess the weight voters place on their preference for democracy, the authors utilized situations where participants had to choose between candidates, some of whom expressed anti-democratic policy preferences. The first study is a candidate choice experiment, where researchers made up hypothetical candidates for participants to choose between, and the second is a real-world naturalistic experiment using data from a Montana special election. In the candidate choice experiment, participants made a series of 16 choices between two candidates for state legislature. Candidates were systemically varied in party affiliation, two policy stances, and one democracy-relevant policy. Candidates either endorsed one anti-democratic policy position (e.g., they “supported a proposal to reduce the number of polling stations in areas that support [the opposite party],”) or endorsed a neutral policy position (e.g. “Served on a committee that establishes the state legislature’s schedule for each session.”) In Table 1, “D-“ denotes candidates who endorsed an anti-Democratic position and “D+” denotes those who didn’t.
Table 1. Candidate choice contrasts in Graham and Svolik (2020)
| alpha | ||||
|---|---|---|---|---|
| 10 | 18 | |||
Participants had previously reported their own party affiliations and positions on the relevant policies in a prior survey. The authors modeled the distance between the participants’ own policy preferences and those of each of the candidates they saw, allowing them to weight if and when voters penalized candidates who adopted anti-democratic policy positions.
The natural experiment utilized data from a 2017 special election in Montana, where candidate Greg Gianforte physically attacked a journalist on the eve of the election, which the authors describe as a violation of democratic principles. They utilize the fact that 70% of Montanans vote by absentee ballot (i.e., before election day and before the assault) to contrast with those who voted after news of the assault was widely publicized on election day. To do this, they use a difference-in-difference model, comparing absentee vote results in 2016 to 2017, and election day vote results in 2016 to 2017, in order to see if election day voters penalized the candidate who behaved anti-democratically.
Main Findings
Americans penalize candidates who endorse anti-democratic positions, but only slightly – their vote share decreased by 11.7%. Meaning, just over 10% of participants valued democracy enough to vote against an otherwise favorable candidate because of their anti-democratic policy. Voters punish anti-democratic behavior by both parties, but they hold a double standard, punishing anti-democratic candidates in the opposite party more than within their own party. Primary elections are unlikely to serve as a check against this – they found similarly low levels of punishment for anti-democratic candidates in within-party contests (i.e., anti-democratic Republican versus a neutral Republican). In these primary-like contests, 15.9% of respondents defect from a anti-democratic candidate when both candidates have similar economic and social policies, but as the distance between the two candidates’ policies increase, participants become unwilling to defect from an anti-democratic candidate with preferred policy positions. Centrist voters are more likely to punish anti-democratic candidates than more ideologically extreme voters, so polarization exacerbates these trends.
Replicating this pattern, in the Montana election analysis they find that Gianforte was slightly punished by the loss of 3.6% of election day voters. Further, the more Republican a precinct was, the more forgiving its voters were of Gianforte’s assault. Voters in more purple precincts imposed a bigger election day penalty for Gianforte’s actions.
Implications
The most significant implication of the research is that support for democracy in the U.S. might be less robust than traditional survey methods would indicate. While Americans express strong support for democracy in principle, they are in practice much more complacent with anti-democratic candidates. This is especially true for strong partisans, and when the ideological distance between the two candidates is larger.
Questions left unanswered
While the authors demonstrate voters’ willingness to accept anti-democratic policy positions alongside other favorable policies, they make the assumption that anti-democratic policies are indeed viewed unfavorably by American voters and are orthogonal to partisan identity and other policy preferences. Future research might examine how anti-democratic policies become more or less entrenched in party platforms, and whether that affects voters’ assessments of those policies and candidates.
Methods and Analysis
Was the study and its analyses pre-registered?: No
Did the study rely on proxy variables to measure polarization?:
N/A
Were standard p-value thresholds used (p<.05 or 95% Confidence Intervals that don’t overlap zero)?: Yes
- Largest p-value presented as significant: 0.05
Were correlational results interpreted with causal language?: No
Limitations / Weaknesses
Broadly, the study question and design are insightful and well-reported. For the Montana natural experiment, only 5 out of the state’s 56 counties tally absentee and election-day ballots separately, therefore only a small fraction of the state’s data was used. It bears scrutiny of how representative those counties are of the broader state. Additionally, it is not clear the extent to which voters understood Gianforte’s assault as a violation of democracy specifically. For the candidate choice experiment, the authors systematically varied candidates’ positions on social and economic policies, along with party identification – however, in practice partisanship and social/economic policy positions are correlated with each other, and potentially also with anti-democratic policies. This detracts from the external validity of the study, as participants might have seen unrealistic combinations of candidates’ party, policies, and democratic attitudes.
Open Data & Analyses
Does the article make the replication data publicly available?: Yes
Does the article make the replication analysis scripts publicly available?: Yes
Article Citation
Graham, M. H., & Svolik, M. W. (2020). Democracy in America? Partisanship, polarization, and the robustness of support for democracy in the United States. American Political Science Review, 114(2), 392-409.
Bibtex
1
2
3
4
5
6
7
8
9
10
@article{graham2020democracy,
title={Democracy in America? Partisanship, polarization, and the robustness of support for democracy in the United States},
author={Graham, Matthew H and Svolik, Milan W},
journal={American Political Science Review},
volume={114},
number={2},
pages={392--409},
year={2020},
publisher={Cambridge University Press}
}