Limits of the Ease-of-Retrieval Effect in Real and Fake News Credibility Judgments: Two Preregistered Experiments.

Ease-of-retrieval theories predict that information seems more credible when supporting reasons come to mind easily. However, it is unclear whether this holds for realistically ambiguous news headlines. We conducted two preregistered experiments (N = 128; N = 135) in which participants evaluated six pilot-tested real and fake headlines selected to minimize baseline credibility differences between veracity categories. In Experiment 1, participants generated either two or six reasons supporting or opposing each headline's truth. In Experiment 2, the two-versus-six manipulation was crossed with a 20 s time limit. Headline veracity (real vs. fake) varied within participants, who then rated perceived task difficulty, credibility, and familiarity. Across experiments, generating six (vs. two) reasons increased perceived difficulty and reduced perceived deliberation time, indicating that the manipulation affected subjective fluency. However, linear mixed-effects analyses showed no reliable effect of the number of reasons on credibility. Credibility instead shifted with the direction of reasons required (supporting vs. opposing; ΔM ≈ 1.3) and increased with headline familiarity (r ≈ 0.40). These findings suggest that for ambiguous real-world headlines, classic ease-of-retrieval manipulations may alter perceived effort without translating into credibility judgments. Future work should test stronger fluency interventions and account for familiarity and motivational factors.