Keywords: democracy, public opinion, conjoint, experiment

In the 2022 U.S. midterm elections and the 2024 presidential race, President Biden and the Democratic Party framed the primary electoral choice as between democracy and anti-democratic, illiberal governance. Despite this "Democracy on the Ballot" narrative, a paradox emerged in the 2024 U.S. presidential election: Donald Trump won the majority of voters who believed that "Democracy was Threatened" by three percentage points, according to CNN's exit poll. This disconnect between political rhetoric and voter behavior raises a critical question: What does democracy mean to the American electorate? While democracy remains a widely embraced ideal, its interpretation varies significantly, making it a potent but ambiguous theme in political discourse and public opinion.

In an era of democratic backsliding, understanding how citizens conceptualize democracy is crucial. Traditional surveys often ask about satisfaction with democracy and perceptions of democratic status without defining democracy, raising concerns about data validity. Democracy is a multifaceted concept, and subjective measures of democratic attitudes often produce inconsistent results. While objective indicators like electoral competitiveness and media freedom help assess democracy, these macro-level measures fail to capture how individuals perceive and value democratic principles. Understanding democracy's role requires combining objective metrics with subjective, citizen-level measures, as individuals' conceptualizations influence their willingness to promote or deviate from democratic norms.

This study examines two central questions: What do Americans believe constitutes ideal democracy, and do their self-reported beliefs align with their revealed preferences when forced to make choices? To address the first question, we used a World Values Survey item asking participants to rank what they believe is most essential to democracy. For the second question, we conducted a conjoint experiment based on Hainmueller et al. (2014). Participants evaluated randomized country profiles with attributes representing procedural democracy, distributive democracy, and liberal democracy: economic well-being, political equity, freedom of speech, and rule of law. Each attribute had three levels: normative, illiberal minority, and illiberal majority. Over five rounds, participants chose the country that best reflected their ideal version of democracy.

The sample consisted of 623 participants recruited from an online opt-in panel, reflecting the American population's demographic composition. We analyzed revealed preferences using frequentist methods and Bayesian Additive Regression Trees (BART), which captured heterogeneous relationships between attributes, outcomes, and participant demographics.

Results indicate that self-reported survey attitudes toward democracy often diverge from revealed preferences. Preferences for political equality, rule of law, and freedom of speech are moderated by economic well-being. While Americans broadly value democratic norms, their support is malleable depending on economic circumstances. Furthermore, education significantly influences democratic preferences: individuals with higher education are more willing to accept being in the minority and prioritize liberal and procedural democracy. Those with lower education levels prefer distributive democracy over liberal democracy and resist being in the minority on democratic attributes. These findings highlight the complexity of Americans' democratic values and their implications for understanding public opinion and electoral behavior.