Evaluation of Missing Data Imputation Methods for Human Osteometric Measurements

Using William W. Howells' Craniometric Data Set and the Goldman Osteometric Data Set, the authors of this study evaluated the performance of multiple popular statistical methods for imputing missing metric measurements. Results indicated that multiple imputation methods outperformed single imputation methods, such as Bayesian principal component analysis (BPCA). Based on the findings of this study, the authors suggest a practical procedure for choosing appropriate imputation methods. As many quantitative multivariate analyses cannot handle incomplete data, missing data imputation or estimation is a common preprocessing practice in biological anthropology for data pertaining to incomplete bioarcheological or forensic skeleton specimens. Multiple imputation with Bayesian linear regression implemented in the R package norm2, the Expectation–Maximization (EM) with Bootstrapping algorithm implemented in Amelia, and the Predictive Mean Matching (PMM) method and several of the derivative linear regression models implemented in mice, perform well regarding accuracy, robustness, and speed. (Published Abstract Provided)