Repetition increases belief in climate-skeptical claims, even for climate science endorsers

Identical to results for the LMM using piloted claim type, we found a main effect for repeti- tion, F(1, 216) = 18.78, p < .001, and a main effect for subjective claim type, F(1, 240) = 94.43, p < .001, but no interaction effect, F(1, 227) = .26, p = .61. The lack of an interaction in combi- nation with a main effect of repetition suggests that people showed an ITE for both claim types. Indeed, even claims that participants self-classified as skeptic-aligning had higher esti- mated mean truth ratings when they were repeated (M = 3.14, SE = .20) than when they were not (M = 2.46, SE = .21), with a mean difference of .68, 95% CI [-.12, 1.49]. T-tests on the raw scores are not reported due to the difficult interpretation of low sample sizes. Estimated mean truth ratings across repetition and claim type are presented in S3 Fig. ITE held for strongest endorsers of climate science. It is also conceivable that our partic- ipants endorsed climate science but were not particularly committed to these beliefs. To address this possibility, we examined whether the effect sizes for skeptic-aligning claims varied depending on how strongly participants endorsed climate science. To do so, we used empiri- cally derived categories from the SASSY scale [44] in Experiment 2, which distinguishes groups of individuals based on their beliefs and behaviors regarding climate change. These groups range from the Alarmed group, who are the most concerned about climate change, to the Dismissive group, who reject that the climate problem is real. People with less extreme

views belong to one of four other groups—the Concerned (who believe in climate change being a problem, but are less involved in action than the Alarmed), Cautious (who also believe that climate change is a problem, but are less certain and involved than the Concerned), Disen- gaged (who pay little attention to climate change), or Doubtful (who believe that climate change is naturally-caused and America is already doing enough to respond to climate change). To investigate whether climate change endorsers are susceptible to the repetition of counter-attitudinal claims, we tested the effect of repeating skeptic claims for the Alarmed, Concerned, and Cautious groups, the three groups with the highest N in our sample. As shown in Fig 5, the effect sizes of repetition for skeptic-aligning claims were similar across the Alarmed, Concerned, and Cautious groups with overlapping 95% confidence intervals. Specifically, those in the Cautious segment rated repeated skeptic-aligning claims (M = 3.94, SD = 1.15) as more true than non-repeated skeptic-aligning claims (M = 3.37, SD = .77), t(26) = 2.65, p = .013, with an effect size of dz = .57, 95% CI [.17, .96]. Those in the Con- cerned segment similarly rated repeated skeptic-aligning claims (M = 3.64, SD = 1.39) as more true than non-repeated skeptic-aligning claims (M = 2.91, SD = .88), t(34) = 3.55, p = .001, with an effect size of dz = .61, 95% CI [.26, .97]. Most importantly, even those in the Alarmed segment rated repeated skeptic-aligning claims (M = 3.50, SD = 1.50) as more true than non-

repeated skeptic-aligning claims (M = 2.97, SD = 1.15), t(35) = 2.17, p = .037, with an effect size of dz = .39, 95% CI [.05, .72]. Among the most fervent climate science supporters, the Alarmed segment, this translates into a 61% chance that a randomly sampled truth rating for repeated skeptic-aligning claims was higher than a randomly sampled truth rating for non- repeated skeptic claims. As the strongest test of this question, we conducted a linear mixed effects model for only the Alarmed participants, with factors identical to the main reported model: participant (ran- dom), repetition (fixed; repeated vs. non-repeated), claim type (fixed; scientist vs. skeptic), and repetition x claim type interaction (fixed). Again, we found significant main effects of repeti- tion, F(1, 249) = 16.91, p < .001 and claim type, F(1, 249) = 29.93, p < .001, but no interaction effect, F(1, 249) = .41, p = .52 (Table 5). In sum, our analyses establish that repetition increases perceived truth of counter-attitudi- nal claims for climate science endorsers, even when they themselves categorize the claims as PLOS ONE Illusory truth effect for climate-skeptical claims PLOS ONE | https://doi.org/10.1371/journal.pone.0307294 August 7, 2024 12 / 19