COMPSTAT 2022: Start Registration
View Submission - COMPSTAT2022
A0557
Title: Missing-data analysis with power M-estimators Authors:  Gabriel Frahm - Helmut Schmidt University (Germany) [presenting]
Abstract: Frahm et al. (2020) develop an observed-data M-estimation procedure in order to estimate the location and scatter of incomplete multivariate data. We have evaluated their so-called power M-estimators by performing a simulation study, comparing the observed-data M-estimation approach with single- and multiple-imputation methods of missing-data analysis. Our main objective is to clarify the open question of whether it is better to rely on imputation or not. This question is highly relevant since observations are rarely complete in real-life applications.