We investigated the impact of different approaches for handling missing data on inferences using a molecular evolution case study. Using a fish trait dataset, we aimed to identify traits that were significantly associated with molecular rates. Multivariable regressions were performed using: 1) a complete-case dataset, and datasets imputed using 2) trait-only (non-phylogenetic) imputation and 3) phylogenetic imputation methods. Results were compared to assess the impact of the missing data handling approach on the significance level of the association for each trait. Results indicated that the model fitted to the phylogenetic imputed data aligned with complete case, while also revealing new insights about molecular rate correlates in fishes. These results are supported by those of previous studies and suggest that the missing data handling approach can have a considerable impact on biological conclusions.