Missing data in multilevel research.

Autor/inn/en	Grund, Simon; Lüdtke, Oliver; Robitzsch, Alexander
Titel	Missing data in multilevel research.
Quelle	Aus: Humphrey, Stephen E. (Hrsg.); LeBreton, James M. (Hrsg.): The handbook of multilevel theory, measurement, and analysis. 1st ed. Washington, DC: American Psychological Association (2019) S. 365-386 PDF als Volltext Link als defekt meldenVerfügbarkeit
Sprache	englisch
Dokumenttyp	online; gedruckt; Sammelwerksbeitrag
ISBN	978-1-4338-3001-3
DOI	10.1037/0000115-017
Schlagwörter	Datenerhebung; Mehrebenenanalyse; Korrektur; Computerprogramm; Fehlertoleranz; Datenanalyse; Implementationsforschung; Messfehler; Modellierung; Statistische Methode
Abstract	Multilevel data are often incomplete, for example, when participants refuse to answer some items in a questionnaire or drop out of a study that involves multiple measurement occasions. This chapter [II.16] provides a general introduction to the problem of missing data in multilevel research, and presents two principled methods for handling incomplete data: multiple imputation (MI) and maximum likelihood (ML) estimation. It discusses how these procedures can be used to address missing data in multilevel research, and considers their commonalities as well as their individual strengths and weaknesses. The ML and MI may be regarded as state-of-the-art procedures for handling missing data. A brief computer simulation study is used to illustrate the statistical behavior of the parameter estimates obtained from these methods. Finally, the chapter illustrates their application in a data analysis example and provides the syntax files and computer code needed to reproduce our results. (Orig.).
Erfasst von	DIPF \| Leibniz-Institut für Bildungsforschung und Bildungsinformation, Frankfurt am Main
Update	2021/2