Estimation of plausible values using background variables with missing values. A data augmented MCMC approach.

Autor/inn/en	Aßmann, Christian; Carstensen, Claus H.; Gaasch, Christoph; Pohl, Steffi
Institution	Leibniz-Institut für Bildungsverläufe
Titel	Estimation of plausible values using background variables with missing values. A data augmented MCMC approach.
Quelle	Bamberg: Leibniz Institute for Educational Trajectories (LIfBi) (2014), 15 S. PDF als Volltext kostenfreie Datei Link als defekt meldenVerfügbarkeit
Reihe	NEPS working papers. 38
Beigaben	Literaturangaben
Sprache	englisch
Dokumenttyp	online; Monographie; Graue Literatur
Schlagwörter	Schätzung; Validität; Stichprobe; Schuljahr 05; Schüler; Datenanalyse; Datenerfassung; Markowsche Kette; Mathematische Kompetenz; Kompetenzentwicklung; NEPS (National Educational Panel Study)
Abstract	The National Educational Panel Study (NEPS) provides data on the development of competencies across the whole life span to educational researchers and politicians. Plausible values as a measure of individual competence are estimated by explicitly including background variables capturing individual characteristics into the corresponding Item Response Theory (IRT) models. Despite tremendous efforts in field work, missing values in the background variables can occur. Adequate estimation routines are needed to reflect the uncertainty stemming from missing values in the background variables in the estimation of plausible values. To achieve this, we propose to adapt an estimation strategy based on Markov Chain Monte Carlo (MCMC) techniques that simultaneously addresses missing values in background variables in the estimation of plausible values for the competence scores. The resulting hybrid sampling scheme establishes a one-step approach for the estimation of plausible values using IRT models that incorporate background variables with missing values. In a simulation study allowing to control the mechanism causing missing values, we evaluate the validity of our approach with respect to statistical accuracy. The results show that the proposed approach is capable to recover the true regression parameters describing the relationship between latent competence scores and background variables. The approach is illustrated on an example using data from the NEPS on mathematical competencies of fifth grade students. (Orig.).
Erfasst von	DIPF \| Leibniz-Institut für Bildungsforschung und Bildungsinformation, Frankfurt am Main
Update	2020/2