Can Fact-checking Prevent Politicians from Lying ?

Fact-checking has become popular in recent elections as a means to ensure that politicians make truthful statements. Due to a lack of empirical analyses and threats to inference, however, the question of whether campaign strategists respond to fact-checking remains unresolved. Using an interrupted time series design, this paper shows that presidential candidates from the 2012 and 2016 US Presidential Election were less likely to repeat false claims that were rated “false” by a major fact-checking outlet. Yet, an interrupted time series alone may fail to account for other factors that could affect politicians’ tendency to make certain claims. This study addresses these concerns by using i) two types of difference-in-differences analyses, each using claims that were rated “true” and claims that were not fact-checked; ii) a placebo test which evaluates the trend in false claims during randomly chosen, hypothetical timeframes; and iii) a topic model which assesses if presidential candidates happen to switch to a new agenda immediately after the fact-check for reasons unrelated to the fact-checking itself. While the effects vary across candidates, this paper demonstrates an important channel through which fact-checking matters in presidential campaigns by showing that fact-checking deters politicians from repeating debunked claims.