New research: People can learn from fact-checking, even when they don’t agree with the politics

In one of the most fact-free elections in history that features two of the most mistrusted candidates in history, there may be a bit of hope. A new study indicates that people can learn what’s true and what’s false after reading fact checks of political claims.

And, perhaps more surprisingly, they even learn from fact-checks that run counter to their political preferences.

The working paper, published this week by veteran fact-checking researchers Brendan Nyhan of Dartmouth College and Jason Reifler of the University of Exeter, studied the effectiveness of fact-checking during U.S. political campaigns in fall 2014.  The research is sponsored by the American Press Institute.

Participants in the study were shown several articles that fact-checked a claim by a candidate or political figure; or, they were shown “placebo,” non-political articles if they were in the control group. 

After a specific time period ranging from days up to nearly a month, participants were asked questions about those fact checks. For example, participants who had read a fact check of Florida Gov. Rick Scott’s claim that electricity rates declined while he was in office (rated false) were asked to evaluate the veracity of the statement: “Electricity rates have declined while Rick Scott has been governor of Florida.”

For comparison, a separate control group was given “placebo” press releases to read.

Overall, the researchers found, the “fact-checking treatment” increased the proportion of accurate answers by about 14 percentage points. This is significant, the authors say, because the mean proportion of accurate answers in the control group was a mere 33 percent.

The study also found that:

• Republicans “feel less positively” about the journalistic practice of fact-checking than Democrats do, supporting findings of earlier research. However, the authors note, “…these partisan differences seem to be driven more by Democratic fondness for fact-checking rather than Republican antipathy [for fact-checking].”
• Exposure to fact-checking had a positive effect on accuracy even when fact-checks were not “belief-consistent” (i.e., in line with the respondent’s party politics).
• Exposure to fact-checking seems to have “little or no measurable effect” on people’s trust in politics overall.
• Fact-checking exposure increases the accuracy of people’s beliefs even more among those who already have a higher level of political knowledge.

The authors acknowledge that some of their findings are at odds with the results of previous studies, including one of their own. They theorize that differences in methodology might be an explanation, including their efforts to encourage respondents to read fact checks carefully by asking them basic questions about the content.

They also suspect that conducting their research during a lower-interest midterm election as opposed to a presidential campaign meant people were less passionate about topics and “less likely to engage in motivated reasoning.”

The authors suggest additional research, including a study that uses traditional “he said-she said” political coverage as the placebo content, rather than the press releases used in this study.

This research was commissioned by the American Press Institute through a grant from the Democracy Fund. You can read more API research and other studies on fact-checking here.