Science beliefs, political ideology, and cognitive sophistication

Contributions

• Compares four major theoretical accounts of science belief formation using preregistered studies across two large representative US samples (N=691 and N=1,018).
• Examines beliefs on a broad range of science topics (17 items covering global warming, evolution, vaccines, GMOs, etc.) rather than the few high-salience issues that dominate the literature, providing a more balanced empirical basis for generalizing about science beliefs.
• Demonstrates that knowledge deficit and analytic thinking accounts explain more variance in pro-science beliefs than political ideology or identity-protective cognition, with heterogeneity across topics.
• Tests and rejects the claim that cognitive sophistication exacerbates political polarization on climate change; instead finds that sophistication is associated with decreased polarization under partisan motivation conditions.

Method

Study 1 design: Two nationally representative samples (Lucid N=691, Dynata N=1,018) completed an online survey measuring:

• Pro-science beliefs: 17 statements on controversial scientific topics rated on a 1–6 scale (strongly disagree to strongly agree); responses coded as pro-science and separated into items perceived as liberal-leaning (global warming, evolution, big bang, sex ambiguity) vs. conservative-leaning (remaining items).
• Science knowledge: 12 basic science knowledge questions (e.g., "Antibiotics kill viruses as well as bacteria"; "The center of the earth is very hot") with correct answers summed.
• Science methodology knowledge: 9 multiple-choice questions on scientific method (e.g., "What is the difference between a theory and a hypothesis?").
• Cognitive sophistication: Three measures capturing different aspects:
• Cognitive Reflection Test (CRT; 6 items): Problems with incorrect intuitive answers, measuring disposition for analytic thinking.
• Actively Open-Minded Thinking about Evidence scale (AOT-E; 8 items): Willingness to change beliefs based on evidence.
• Bullshit Receptivity Scale (BSR; 5 items): Susceptibility to pseudo-profound nonsense statements.
• Controls: Political ideology/partisanship, religiosity.

Study 2 experiments: Two experiments testing climate change specifically, manipulating partisan motivation and measuring whether sophisticated reasoners become more or less polarized. Participants evaluated arguments in partisan framing conditions and measured cognitive sophistication (CRT).

Results

Study 1 correlations:

• Science beliefs correlated strongly with science knowledge (r = .36–.37), science methodology knowledge (r = .27–.31), and analytic thinking measures (CRT r = .36; AOT-E r = .33; all large effect sizes by psychological standards).
• Political ideology/partisanship showed weak correlations with overall science belief (r = −.14–−.15), though stronger for conservative pro-science beliefs (r = .18–.22).
• CRT and BSR correlated specifically with conservative (not liberal) pro-science beliefs, indicating topic-level heterogeneity.

Political partisanship distribution: Surprisingly, the distribution of agreement ratings on most science belief items was nearly identical between Republicans and Democrats, despite explicit sampling to capture topics perceived as politically divisive. Most items showed broad consensus (high contentious variability but similar mean agreement).

Study 2 (climate experiments): Contrary to the identity-protective cognition hypothesis, increased cognitive sophistication was associated with decreased polarization on climate change under partisan motivation. More sophisticated reasoners did not defensively interpret arguments in politically motivated ways; instead, analytical thinking predicted more accurate, less polarized judgments.

Notes

Strength: Broadens the empirical base by testing across 17 diverse scientific topics rather than focusing on 2–3 high-salience politically charged issues (climate, evolution, vaccines). This methodological choice reveals that much of the "science and politics" literature overgeneralizes from outliers. The preregistered design and two independent samples enhance confidence.

Key limitation: The paper tests correlational relationships; causal mechanisms remain underspecified. Does science knowledge cause pro-science beliefs, or do pro-science beliefs lead people to acquire knowledge? AOT-E (openness to evidence) is measured post-hoc, not experimentally manipulated, limiting causal claims about whether openness prevents motivated reasoning.

Policy implication: Results suggest educators and policymakers should prioritize increasing basic science literacy and critical thinking habits rather than focus on combating "partisan misinformation" or appealing to group identity. The consistent finding that knowledge explains more variance than ideology implies that knowledge-deficit interventions (teaching science) may be more effective and less politically fraught than ideology-targeting approaches.

Broader context: This work is part of a sustained program by Pennycook's group challenging the prominence of motivated reasoning in explaining misinformation and belief; see also Pennycook & Rand's work on lazy vs. biased theories and the effectiveness of simple accuracy nudges.