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

Across both experiments, there was a significant main effect of claim type, Experiment 1: F(1, 701) = 59.36, p < .001; Experiment 2: F(1, 641) = 74.58, p < .001. Comparison of raw M and SDs showed consistent findings across both experiments. In Experiment 1, scientist-aligned claims (M = 3.95, SD = .79) were rated as more true than skeptic-aligned claims (M = 3.13, SD = 1.01), t(46) = 5.94, p < .001, dz = .89, 95% CI [.55, 1.25]. This pattern replicated in Exper- iment 2—climate scientist-aligned claims (M = 4.13, SD = .76) were again rated as more true than skeptic-aligned claims (M = 3.48, SD = 1.04), t(109) = 5.54, p < .001, dz = .71, 95% CI [.44, .98]. Replicating the familiar ITE, people rated repeated claims as truer than non-repeated claims. Across both experiments, we found a significant main effect of repetition, Experiment 1: F(1, 701) = 32.34, p < .001; Experiment 2: F(1, 641) = 42.97, p < .001. Comparison of raw M and SDs showed consistent findings across both experiments. In Experiment 1, repeated claims (M = 4.08) were rated as more true than non-repeated claims (M = 3.30), dz = .87, 95% CI [.48, 1.27]. In Experiment 2, repeated claims (M = 4.50) were again rated as more true than non-repeated claims (M = 3.59), dz = 1.19, 95% CI [.90, 1.50]., A significant interaction between repetition and claim type would indicate that the effect of repetition on truth assessments is moderated by whether the claim is counter-attitudinal or not. We did not find this effect in Experiment 1, F(1, 701) = 1.38, p = .24, or Experiment 2, F(1,

641) = .27, p = .60. Estimated mean truth ratings across both repetition and claim type are pre- sented in Fig 1 (Experiment 1) and Fig 2 (Experiment 2). As an exploratory analysis of the size of ITE, we found a medium to large effect of repetition for scientist-aligning as well as skeptic- aligning claims in both experiments (Table 3). Taken together, these findings show that repeti- tion increased perceived truth to a similar extent for each claim type, indicating that repetition is similarly influential for pro- and counter-attitudinal claims. In S2 File we also report pre-registered mixed models which include the factor of belief (measured through beliefs and attitudes scales in both experiments, and SASSY in Experiment 2) as a covariate. These analyses allowed us to inspect truth assessment across claim type and repetition while taking into account people’s variation across general endorsement of climate change science. These pre-registered mixed models report the same significant pattern of results with repetition, and no interaction between repetition x claim type, consistent with what we reported above. Visual representation of the ITE across claim types. To visualize the magnitude of the repetition effect for each claim type, we calculated the difference in truth ratings between Table 1. Estimates of fixed effects in Experiment 1 with endorsers only. Variable B SE df t p 95% CI Intercept 2.88 .15 125.65 19.74 < .001 [2.59, 3.17] Repetition [repeated] .48 .15 701.03 3.19 .001 [.19, .78]

Claim Type [scientist-aligning] .70 .15 701.09 4.62 < .001 [.40, 1.00] Repetition [repeated] * Claim Type [scientist-aligning] .25 .21 701.06 1.17 .24 [-.17, .67] https://doi.org/10.1371/journal.pone.0307294.t001 Table 2. Estimates of fixed effects in Experiment 2 with endorsers only. Variable B SE df t p 95% CI Intercept 3.02 .12 446.78 26.08 < .001 [2.79, 3.24] Repetition [repeated] .63 .15 641.00 4.27 < .001 [.34, .92] Claim Type [scientist-aligning] .85 .15 641.00 5.74 < .001 [.56, 1.14] Repetition [repeated] * Claim Type [scientist-aligning] .11 .21 641.00 .520 .60 [-.30, .52] https://doi.org/10.1371/journal.pone.0307294.t002 PLOS ONE Illusory truth effect for climate-skeptical claims PLOS ONE | https://doi.org/10.1371/journal.pone.0307294 August 7, 2024 7 / 19