People prefer simpler explanations and judge them more satisfying, where simplicity is defined as the number of unexplained causes in the explanation (“root simplicity”). But what are the effects of this preference on cognition and why do people like simple explanations? We consider the answers to these questions with three experiments. Our findings suggest that the participants who select simpler explanations when the simpler explanation conflicts with available data are more likely to overestimate how often they saw data points consistent with their chosen explanation. We show that choosing a simpler explanation causes this data distortion. We find that the preference for root simplicity is stronger when root causes greatly alter the probability of their effects. This aligns with arguments suggesting a preference for root simplicity results from the desire to efficiently encode observations and to support effective interventions. These findings have implications for theories of inference and explanation.