Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
The divergence of attitudes towards their ideological extremes has become an identifying feature in the United States. Little is known about its source, how large it is, whether information can attenuate it, and its causal impact on civic behavior. We design a survey experiment that identifies differences in beliefs rather than preferences as a source of division. We randomly introduce factual information about government spending and show that it corrects beliefs. We further use this variation and estimate effects on a suite of outcomes. For individuals who learn the government spends worse than they would prefer, they become 0.35 s.d. less supportive towards the government, believe the government is less efficient by 0.42 s.d.and are less willing to compromise and trust by 0.43 s.d. We do not find any changes for those who learn the government spends more in line with their preferences. This asymmetric response is consistent with a literature showing that negative information has a greater impact on attitudes and beliefs than positive information.