Imputing missing data values in repeated measurement within-subjects designs

Bingham CR, Stemmler M, Petersen AC, Graber JA (1998)

Publication Type: Journal article

Publication year: 1998

Journal

Methods of Psychological Research Institute for Science Education

Publisher: Institute for Science Education

Book Volume: 3

Pages Range: 131-155

Journal Issue: 2

Abstract

Introduces a procedure for inputing missing data values in repeated measurement within-subjects designs. Two sets of simulations were performed. The first used a single data set to examine the influence of the proportion of cases with missing values, the number of repeated measures, and the shape of the individual response curve on the discrepancy between the imputed and observed data, and to compare this method with mean-substitution. A Monte Carlo simulation was also conducted which examined the proportion of cases with missing values, the shape of the individual response curve, and the sample size in relation to the discrepancy between the observed and imputed standard errors and moments. Results indicated that the imputed data closely approximated observed values, that estimation improved as the number of repeated measures increased and proportion of cases with missing values decreased, and that sample sizes greater than 100 provided equivalent results.

Authors with CRIS profile

Mark Stemmler Lehrstuhl für Psychologische Diagnostik, Methodenlehre und Rechtspsychologie

Involved external institutions

University of Michigan US United States (USA) (US) Columbia University US United States (USA) (US)

How to cite

APA:

Bingham, C.R., Stemmler, M., Petersen, A.C., & Graber, J.A. (1998). Imputing missing data values in repeated measurement within-subjects designs. Methods of Psychological Research, 3(2), 131-155.

MLA:

Bingham, C. Raymond, et al. "Imputing missing data values in repeated measurement within-subjects designs." Methods of Psychological Research 3.2 (1998): 131-155.

BibTeX: Download