Both trust in, and polarization of trust in, relevant sciences have increased through the COVID-19 pandemic

“

TL;DR

A UK survey (N=2035) conducted in June 2021 found a net increase in self-reported trust in scientists during the COVID-19 pandemic, with 33% reporting increased trust versus only 7% reporting decreased trust. The increase was specific to relevant science groups: a negative control (geologists) showed no comparable shift, and naming Pfizer (a vaccine producer) prompted greater reported trust in pharmaceutical companies than naming GSK (Pfizer mean trust score 0.2304 vs GSK 0.0611, P=6.487×10⁻⁷). Critically, trust polarized—those who distrusted science pre-pandemic became more distrustful (suggesting a backfire effect), and decreased trust strongly predicted COVID-19 vaccine refusal, with public health consequences.

Full summary

4,953 CHARS

**Background** Trust in science, rather than mere scientific knowledge, is now thought foundational to acceptance of well-evidenced science. The COVID-19 pandemic provided an unprecedented natural experiment on trust modulation through extensive public exposure to scientists and pharmaceutical companies. This study addressed three questions: (1) whether trust in scientists has on average changed through the pandemic, (2) whether any change is real and attributable to pandemic-relevant activity, and (3) what predicts differences in trust change, including whether pre-pandemic trust itself predicts change (i.e., polarization).

**Methods** The Genetics Society funded a survey delivered by Kantar Public in June 2021 to over 2000 UK adults (recruited through Public Voice panel, survey 11), with weight efficiency of 72% and relative statistical efficiency of 177%. Respondents answered via online or telephone means. The survey asked whether self-reported trust in "scientists," "geneticists," and "geologists" (negative control) had changed since the start of the pandemic, scored from −2 (much less) to +2 (much more). An embedded experiment randomly assigned respondents to rate change in trust in pharmaceutical companies with either "Pfizer" or "GlaxoSmithKline" (GSK) as exemplars—Pfizer having released an mRNA vaccine, GSK not. Pre-pandemic trust was assessed via recall on a −2 to +2 Likert scale ("Those in charge of new developments in genetic science cannot be trusted to act in society's interests"). Covariates included age, sex, educational attainment, religiosity, political attitude (10-item scale, −1 left to +1 right wing), and prior COVID-19 infection. Behavioral outcome was willingness to accept a COVID-19 vaccine. Analysis used non-parametric tests (Kruskal-Wallis, Mann-Whitney U, Spearman rank, Dunn post-hoc), Spearman partial correlations for covariate control, and a randomization test (1,000,000 iterations) comparing observed variance in post-pandemic trust to null distributions.

**Key Results** Of 2035 respondents, 60% (N=1215) reported no change in trust in scientists, while 33% (N=673) reported increased trust and 7% (N=147) reported decreased trust. Among those reporting any change (N=820), 82% reported an increase (binomial test, P<2.2×10⁻¹⁶). A Kruskal-Wallis test confirmed significant differences across scientists, geneticists, and geologists (H(2)=240.9, P<2.2×10⁻¹⁶), with post-hoc Dunn tests significant for all pairwise comparisons. Geologists served as a negative control showing minimal change (90% no change).

For pharmaceutical companies, binomial tests showed significantly more respondents reporting increased versus decreased trust in both Pfizer (P<2.2×10⁻¹⁶) and GSK groups (P=8.146×10⁻⁶). A Mann-Whitney U test revealed a significantly more positive trust change with Pfizer (mean 0.2304) than GSK (mean 0.0611), P=6.487×10⁻⁷, with the discrepancy driven by greater increases in trust in the Pfizer group ("Trust them a little more": χ²(1)=11.01, P=9.08×10⁻⁴; "Trust them much more": χ²(1)=16.28, P=5.48×10⁻⁵).

POLARIZATION WAS EVIDENT

observed post-pandemic trust variance (1.397) exceeded all 1,000,000 null simulations (P<1×10⁻⁶). Pre-pandemic trust positively correlated with change in trust (Spearman rs=0.31, P<0.0001, N=2039), remaining robust after controlling for age, sex, education, political attitude, and religiosity (partial rs=0.31, P<0.0001). Only age additionally weakly predicted change in trust (rs=0.06, P=0.01).

Of 2015 respondents, 1930 were willing to accept a COVID-19 vaccine and 85 were not. Mann-Whitney U test showed mean trust change of 0.0601 among those willing versus −0.2706 among those refusing (P=2.70×10⁻⁸). The correlation between vaccine willingness and trust change remained significant after controlling for covariates (partial rs=0.11, P=6.88×10⁻⁷, N=1987). Age also predicted vaccination (partial rs=0.13, P=5.16×10⁻⁹), with sex borderline (partial rs=0.04, P=0.05).

**Clinical Implications** Trust in science is modifiable to substantial degrees, with pandemic-related scientific activity plausibly driving increased trust in relevant groups. However, the simultaneous polarization—wherein those who distrusted science pre-pandemic became more distrustful—signals a potential backfire effect from uniform trust-building approaches. The strong link between decreased trust and vaccine refusal indicates that trust dynamics carry direct public health consequences. Interventions should be tailored rather than one-size-fits-all, and qualitative work is needed to understand what might engage those whose distrust has hardened. Caveats include reliance on self-reported (rather than longitudinally measured) change in trust, which could inflate apparent polarization through recall bias, and UK-specific findings that may not generalize globally (e.g., religiosity and educational attainment predictors differed from international studies).