Second-order beliefs – what political actors think others think – can shape agenda-setting and even shift public opinion. Because of the collective-action nature of mitigating human-caused climate change, such second-order political beliefs are particularly important to study. Through an innovative survey design focusing on a policy proposal to introduce meat-free days in canteens, we present the first simultaneous comparison of ordinary citizens’, locally elected political representatives’, and centrally employed public administrators’ own opinions and their ability to accurately identify the majority position of citizens. While citizens are split in their opinion on meat-free days in canteens, a clear majority of unelected elites support it, and most elected elites do not support this policy. Nonetheless, we find that all three groups tend to underestimate the level of policy support among citizens. Through rigorous analysis, we show that elected elites are significantly more likely to underestimate public support for a meat-free day compared to citizens and unelected elites. These results provide important insights into the dynamics of democratic governance and suggest that underestimation of citizens’ support for climate policies may further complicate an already challenging policy area.

When political action entails individual costs but group-contingent benefits, political participation may depend on an individual’s perceptions of others’ beliefs; yet detailed empirical attention to these second-order beliefs – beliefs about the beliefs of others – remains rare. We offer the first comprehensive examination of the distribution and content of second-order climate beliefs in the United States and China, drawing from six new opinion surveys of mass publics, political elites and intellectual elites. We demonstrate that all classes of political actors have second-order beliefs characterized by egocentric bias and global underestimation of pro-climate positions. We then demonstrate experimentally that individual support for pro-climate policies increases after respondents update their second-order beliefs. We conclude that scholars should focus more closely on second-order beliefs as a key factor shaping climate policy inaction and that scholars can use the climate case to extend their understanding of second-order beliefs more broadly.