**Abstract:** Debates exist about the efficacy of misinformation corrections, and whether they can sometimes “backfire” causing people to become more entrenched in misinformation. Using ostensibly real tweets from politicians, neutral organizations, and lay people, we demonstrate how partisanship not only biases belief in misinformation, but also poses a risk factor for corrections backfiring. Across three experiments (N = 1,217 Prolific participants), we find that participants are susceptible to misinformation from political ingroup members far more than outgroup members– a partisan bias most prominent when viewing content shared by famous politicians. Moreover, while corrections worked on average, the effects were small, with backfires occurring 23% of the time and being 52% more likely when the correction came from a political outgroup member. Conservatives were more prone to backfiring than liberals. In sum, corrections have small effects compared to partisan identity congruence, and sometimes backfire–especially if they come from a political outgroup member.