Open Access
January 2017 Comparison of Imputation Methods for Missing Values in Longitudinal Data Under Missing Completely at Random (MCAR) mechanism
Lotsi ANANI, Louis ASIEDU, Johnson KATSEKPOR
Afr. J. Appl. Stat. 4(1): 241-258 (January 2017). DOI: 10.16929/ajas/241.213

Abstract

This paper compared the performance of five (5) techniques of imputing missing values under the assumptions of MCAR mechanisms. The study compared the techniques for solving missing values using the Generalized Estimating Equation (GEE) model for the complete dataset, the coefficient of determination, mean squared error (MSE) and root mean squared error (RMSE). The pairwise deletion is the best under MCAR mechanism. Listwise deletion and the hot deck imputation methods performed poor under the MCAR mechanism.

Dans cet article, nous comparons les performances des cinq (5) techniques d'imputation de valeurs sous les hypothèses de données manquantes de manière complètement aléatoirement (MCAR). La comparaison se fait la base du modèle des Equations Généralisées d'Estimation (GEE) pour la base complète, le ceofficient de détermination, de l'erreur quadratique moyenne (MSE) et du coefficient RMSE. Notre étude conclut que la méthode d'élimination appairée est la meilleure sous l'hypothèse (MCAR). Les performance des méthodes Liswise et hot deck se révèlent relativement faibles.

Citation

Download Citation

Lotsi ANANI. Louis ASIEDU. Johnson KATSEKPOR. "Comparison of Imputation Methods for Missing Values in Longitudinal Data Under Missing Completely at Random (MCAR) mechanism." Afr. J. Appl. Stat. 4 (1) 241 - 258, January 2017. https://doi.org/10.16929/ajas/241.213

Information

Published: January 2017
First available in Project Euclid: 16 May 2019

Digital Object Identifier: 10.16929/ajas/241.213

Subjects:
Primary: 62-07
Secondary: 62-J05

Keywords: ceffcient of determination , GEE , longitudinal data , missing data

Rights: Copyright © 2017 The Statistics and Probability African Society

Vol.4 • No. 1 • January 2017
Back to Top