New research from the University of Glasgow, led by political‑behaviour scholar Christopher Claassen and published in Public Opinion Quarterly, finds that an individual’s commitment to liberal-democratic institutions can shield voters from backing politicians who threaten democratic rules. The study analysed responses from 5,951 participants across three democracies facing backsliding pressures—the United States, Poland and South Korea—through a candidate-choice experiment designed to emulate a real election.
Democratic backsliding is a growing concern in many modern nations. It occurs when elected officials gradually erode democratic institutions such as free elections, independent courts and a free press. A central driver of this decline is the willingness of ordinary citizens to vote for candidates who violate democratic rules.
Past studies have shown that voters generally dislike antidemocratic behaviour in theory, yet they often support politicians who breach democratic rules if those politicians belong to their preferred party or share their policy views. This creates a tension between abstract democratic values and practical political loyalties.
Claassen explained that he sought to connect recent work on democratic backsliding with the longer-running political-culture literature, which argues that citizens’ orientations toward democracy can help sustain democratic government. “A growing body of research shows that voters frequently support politicians who threaten democratic institutions, particularly when those politicians belong to their preferred party or agree with them on important issues,” he told PsyPost. “But we know much less about whether people’s underlying attitudes toward democracy help them resist those pressures.”
The study drew on data from 5,951 participants across the United States, Poland and South Korea, who took part in a candidate-choice experiment that simulated twelve matchups. In each pairing, a prodemocratic candidate faced an antidemocratic counterpart, with the antidemocratic option proposing actions such as cutting polling stations in opposition areas, ignoring court rulings or prosecuting journalists, while the prodemocratic candidate promised to respect these institutions.
Before the simulated elections, participants were assessed on six dimensions of political culture, including preferences for democracy over autocracy, support for liberal democratic institutions and trust in political institutions. The survey also measured political tolerance, emancipative values emphasising individual autonomy and equality, and authoritarian values prioritising obedience and conformity.
Across the three countries, support for liberal democracy emerged as the most consistent predictor of willingness to vote for the prodemocratic candidate. This measure captures backing for core institutions such as free speech, equal voting rights and the rule of law. Individuals scoring highly on this measure were consistently less likely to reward candidates who threatened these democratic pillars.
“One broader implication is that democratic culture is not a single thing – its multidimensional,” Claassen noted. “I examined several kinds of orientations that political scientists have linked to democracy, and they behaved quite differently.”
“Support for liberal-democratic institutions was the most consistent predictor of pro-democratic choice, which suggests that the particular democratic commitments citizens hold may matter more than simply possessing a generally ‘democratic’ political culture,” he added.
The analysis also explored difficult scenarios known as cross-pressured choices, where a voter must choose between a prodemocratic candidate from an opposing party and an antidemocratic candidate from their own party. Even in these demanding circumstances, support for liberal democracy made a noticeable difference, helping voters resist the temptation to blindly back their own side. Specifically, a one-standard-deviation increase in a person’s support for liberal democracy was associated with a 2 to 12 percentage point increase in the probability of choosing the prodemocratic candidate across party lines, with the size of the effect varying by country and model.
“The effects are meaningful but not enormous,” the researcher said. “That can matter in a close election, but the results also show how powerful partisan loyalties remain.”
Other aspects of political culture showed more mixed results. For instance, a general preference for democracy over autocracy and emancipative values occasionally predicted prodemocratic choices, but these factors did not consistently translate across all countries and policy scenarios. Trust in political institutions, by contrast, showed almost no relationship with resisting antidemocratic candidates.
Claassen described two surprises in particular. “First, political trust had virtually no consistent relationship with choosing the pro-democratic candidate, despite a long tradition of treating trust in institutions as important for democracy. Trust may matter for democracy in other ways, but our results suggest that it does not necessarily make people more willing to defend democratic institutions at the ballot box.”
He was also struck by the weak results for authoritarian values. “There is an obvious theoretical connection between authoritarian values and authoritarian political systems – they even share the same name – but those values were not strong predictors of candidate choice in my study,” he noted. “It may be that these underlying social values are simply too distant from concrete decisions about electoral choices.”
The statistical models accounted for a range of variables, including age, gender, education, income and geographic region, and also controlled for political interest, preferred party and participants’ positions on redistribution and gender equality.
As with any study, there are caveats. The voting choices occurred in a simulated survey rather than a real election, and in actual campaigns respondents face intense messaging and social pressures that could alter how democratic violations are weighed against partisan loyalty. The authors stress that the findings are observational about political culture; because values cannot be randomly assigned, the results indicate associations rather than causation, and must be interpreted accordingly. The candidate attributes were randomized, but respondents’ democratic orientations were measured rather than manipulated.
Claassen also noted that the results show partisan loyalty remains remarkably strong: in Poland and South Korea, the predicted share voting for an opposing-party prodemocratic candidate often stayed below 50%, even among citizens with high democratic commitments. This underscores that while robust democratic values can curb backsliding, they do not automatically generate cross‑party majorities against bad actors.
“I would avoid interpreting the findings as showing that strong democratic attitudes will necessarily defeat anti-democratic politicians,” he cautioned. “The main takeaway is that what ordinary citizens think about democracy does matter. People who more strongly support liberal-democratic institutions were more likely to choose the pro-democratic candidate, including in situations where their partisan identity or ideological position pulled them in the opposite direction.”
He added that democratic resilience is not entirely outside citizens’ hands; leadership, party strategies and institutional actions also play a crucial role in whether democracy endures.
Looking ahead, Claassen aims to refine how these attitudes are studied, developing more fine-grained measures of democratic orientations and exploring why people sometimes apply their democratic beliefs to voting decisions—and sometimes do not. He emphasised interest in attitude importance and accessibility: whether democracy is something people care deeply about and whether those considerations are readily available in voters’ minds when they cast a ballot.
The study is titled Voting for Democracy? Testing Political-Cultural Constraints on Democratic Backsliding and was authored by Christopher Claassen.
