Authors: Daniel Ebanks, R. Michael Alvarez, Ramit Debnath
The scientific community produces high-quality research that informs our understanding of a vast array of natural, physical, and social phenomena. This broad research community frequently produces work that has serious implications for society at large. Where those findings intersect with policy, the public is more likely to scrutinize the experts’ findings. Scientists generally welcome healthy skepticism of their work–the system of academic peer review for publication relies on carefully considered critiques of the work. Often, good science arises from overturning long-established findings. Of concern to scientists is if the public perceives research experts as ideological actors, then the public is likely to polarize on the experts’ findings. This polarization can sow distrust, even in the best scientific research. If the public polarizes politically on scientific findings, scientists may be unable to reach segments of the population who are aligned with particular political parties, even if their work is otherwise persuasive.
In this project, we run a survey conjoint experiment to tease out three questions where political polarization mediates trust in scientists as a causal mechanism. This work is based on a broader effort in social science that employs conjoint experiments to tease out the publics’ multidimensional preferences. First, is the public likely to trust scientists with respect to scientific findings that have no obvious political valence? Second, even if the public tends to trust scientists on these apolititical matters, are conservatives more likely to distrust scientists because they perceive them as politically liberal? Third, are liberals more likely to trust scientists because they perceive them as co-partisans? In a nationally representative sample, we test key hypotheses related to the aforementioned causal questions driving trust and distrust in research scientists. To tease out the causal mechanism driving trust or distrust in scientists, we employ a discrete-choice 2x3 factorial conjoint described in further detail below.