Violations of the molecular clock have long been thought to indicate phylogenetic model misspecification, misspecifications that have recently gained greater relevance due to the abundant data of the phylogenomics era. However, the degree to which lineage-specific rate heterogeneity is indicative of model misspecification and the degree to which any such misspecification misleads inference are unknown. Here we propose a two-part study, uniting results from simulation with those from a synthesis of published data to determine the effect of rate heterogeneity on the accuracy of phylogenetic inference.