You're Definitely Wrong, Maybe: Correction Style Has Minimal Effect on Corrections of Misinformation Online

Key findings

• Minimal effect of correction style: Testing 2,228 participants, the study found no significant differences in whether people reply to, accept, or resist corrections based on whether the correction was direct or hedged, or simple vs. detailed.
• Analytic thinking predicts acceptance: Participants with higher cognitive reflection and actively open-minded thinking were significantly more likely to accept corrective information, independent of correction framing.
• Implications: User-generated corrections on social media may be similarly effective across a range of reasonable correction styles, but individual differences in thinking patterns matter more than message format.

Method

The study recruited participants from Lucid (N = 2,228) who indicated they would share a false news story in an initial survey. Researchers then presented corrective responses that varied along two dimensions:

• Correction strength: Direct (e.g., "This is false") vs. hedged (e.g., "This might not be accurate")
• Explanatory depth: Simple explanation of why the claim is false vs. detailed, multi-part explanation

Outcomes measured: likelihood of replying to the correction and acceptance/resistance of the corrective information.

Results

• No statistically significant effects of correction strength or depth on engagement or acceptance
• Analytic thinking (measured via cognitive reflection test and actively open-minded thinking scale) strongly predicts acceptance across all correction conditions
• Suggests that individual cognitive predispositions matter more than strategic framing choices in determining whether users accept corrections

Notes

This work is consistent with the Rand lab's broader finding that individual differences in reasoning ability and epistemic values (e.g., accuracy salience) matter more than message framing for misinformation interventions. The minimal effect of politeness and explanation depth is somewhat reassuring for crowd-sourced correction systems, which often receive a diversity of correction styles—the result suggests that even imperfectly phrased corrections may be effective if the recipient is analytically inclined to consider them.