Do Facts Speak for Themselves? Causes and Consequences of Partisan Bias in Factual Beliefs
Citation: Khanna, K. (2019). Do Facts Speak for Themselves? Causes and Consequences of Partisan Bias in Factual Beliefs (Doctoral dissertation, Princeton University).
Abstract:
Observers of contemporary U.S. politics lament the seeming inability of Democrats and Republicans to agree on objective truth, such as the unemployment rate, federal deficit, and other politically relevant matters of fact. These apparent divisions are troubling from a normative perspective. They cast doubt on citizens’ ability to form informed opinions, engage in productive debate with each other, and reward and punish elected officials.
This dissertation explores causes and consequences of these partisan gaps in the public’s factual beliefs. Using a combination of survey meta-analysis and original data collection, I measure the extent of partisan bias in beliefs and experimentally probe its proposed mechanisms. After building a database of national polling data on a variety of economic indicators from 1980 to the present, I find that partisan bias in perceptions of the economy is less severe than past work has suggested. However, the database reveals considerable heterogeneity in bias across various survey items and real-word conditions.
In a series of experiments, I test a prominent and troubling theory of partisan bias: selective learning of factual information. I expose partisans to actual facts with either positive (congenial) or negative (uncongenial) implications for their party. I find little evidence of selective learning across studies. Instead, partisans learn facts evenhandedly. While many forget factual information after several days, they are equally likely to recall congenial and uncongenial information, mitigating concerns about bias. I also find that partisans assimilate both congenial facts and uncongenial facts into relevant attitudes, such as incumbent evaluations and policy preferences. These attitudinal shifts persist several days after initial exposure to factual information.
Rather than revealing a deep-seated bias in how people learn, partisan gaps in surveys of factual beliefs are more likely to arise from selective reporting on the part of respondents with motivation to express partisan loyalty. Evidence of selective reporting comes from a series of survey experiments that nudge respondents to accurately report their beliefs in response to factual questions without providing respondents with any additional information. However, incentivizing respondents to counter their bias in reporting factual information can have adverse effects on their subjective attitudes.
Read full dissertation here.