A study explores how social networks influence interpretations of climate data. Interpretation of climate data is vulnerable to political and psychological biases. To test whether exposure to opposing political views reduces partisan bias in climate data interpretation, Damon Centola and colleagues recruited 2,400 participants from Amazon's Mechanical Turk. Participants were randomly assigned to 40-person bipartisan social networks and asked to interpret a graph of Arctic sea ice published by NASA. Participants were allowed to revise their responses while being shown the average response of their network neighbors. Initially, participants exhibited strong partisan bias; liberals' interpretations were more likely to be accurate than those of conservatives. In networks where participants were exposed to each other's opinions while seeing Democratic and Republican party logos at the bottom of the screen, no significant improvements by either liberals or conservatives were observed, and partisan bias remained strong. However, in networks where participants were exposed to each other's opinions in the absence of political imagery, both conservatives and liberals became more accurate, and partisan bias in their responses was eliminated. The results suggest that social learning from exposure to opposing political views can improve accuracy and remove partisan bias, but displaying political symbols during cross-party communication can prevent such learning, according to the authors.
Article #17-22664: "Social learning and partisan bias in the interpretation of climate trends," by Douglas Guilbeault, Joshua Becker, and Damon Centola.
MEDIA CONTACT: Damon Centola, Annenberg School for Communication, University of Pennsylvania, Philadelphia, PA; tel: 347-306-1277, 215-898-7954; e-mail: <firstname.lastname@example.org>