Abstract: While social media users are fairly adept at identifying content that contains misinformation, empirical studies show users are more willing to share misinformation. This observation highlights a disconnect between users’ perceptions of truth and their incentives to “share” (i.e., propagate) this content to other users. We provide two related but distinct models of sharing incentives that reconcile this disconnect. The first model adopts a social norms perspective of incentives, where misinformation sharing among co-partisans leads to more flippant and casual sharing of misinformation. The second model considers how misinformation might influence the beliefs of other social media users, and so agents may share misinformation simply because the “ends justify the means”. In this second model, we find that certain equilibria arise where each party fights misinformation spreading from the other party by spreading their own pro-attitudinal misinformation. This effect pushes in the direction of more casual sharing of misinformation when counter-partisans are sharing more misinformation. To test these opposing forces, we design an experiment that presents selective research studies showing whether Democrats or Republicans are more likely to share misinformation on social media. Following the treatment, each participant is subsequently asked about whether she’d share various headlines. Our results allow us to quantify the magnitude of both types of sharing incentives.