Does Wording Matter? Examining the Effect of Phrasing on Memory for Negated Political Fact Checks

General Audience Summary In recent years, there have been a growing number of new fact-checking projects, with each broadly seeking to verify and communicate the truth of questionable claims circulating among the public. These efforts often include brief social media posts that introduce a claim and then affirm or negate it (e.g., "A video purports to show Joe Biden admitting to bribery. He didn't.") A critical challenge is ensuring that people who see these fact checks remember them over time. We hypothesized that one way to improve memory for these fact checks would be to rephrase them such that the affirmation or negation comes before the claim is fully stated (e.g., "No, a video does not show Joe Biden admitting to bribery."), rather than after the claim. Across three experiments, we had participants (all based in the United States) read a series of fact checks from PolitiFact about true and false claims regarding contemporary American politics, and then rate the truth of a series of claims. Overall, these fact checks were effective. Participants were more accurate at rating the truth of claims immediately after reading related fact checks, and they were able to remember whether each claim was true or false over the course of 1 week. After 3 weeks, however, participants did begin to forget whether claims were true or false. Most importantly though, participants began to forget both types of fact checks at similar rates, contrary to our predictions. Overall, this research contributes to a growing literature suggesting that fact checks can improve the accuracy of people's factual beliefs and suggests that there are multiple, equally effective ways of communicating fact checks in the context of social media. After encountering negated messages, people may remember the core claim while forgetting the negative evaluation. These memory errors are of particular concern for fact checks on social media, which often use brief affirmations or negations to help the public learn the truth behind questionable claims. Across three experiments, we examined whether these memory errors could be minimized by placing evaluations before the entire claim is stated (e.g., "No, X did not do Y, as A claims"), rather than after (e.g., "A claims X did Y. No, this is false"). Participants remembered whether fact-checked political claims were affirmed or negated immediately (Experiment 1) and 1 week later (Experiment 2). While participants began to forget these fact checks after 3 weeks, this forgetting was similar for before- and after-claim evaluations, contrary to our predictions (Experiment 3). These results suggest that there are multiple, equally memorable formats for communicating affirmations and negations.