Assessing spelling competence development in the National Educational Panel Study.

Autor/inn/en	Jarsinski, Stephan; Frahm, Sarah; Blatt, Inge; Bos, Wilfried; Kanders, Michael
Titel	Assessing spelling competence development in the National Educational Panel Study.
Quelle	Aus: Blossfeld, Hans-Peter (Hrsg.); Maurice, Jutta von (Hrsg.); Bayer, Michael (Hrsg.); Skopek, Jan (Hrsg.): Methodological issues of longitudinal surveys. Wiesbaden: Springer VS (2016) S. 427-439 PDF als Volltext Link als defekt meldenVerfügbarkeit
Beigaben	Literaturangaben
Sprache	englisch
Dokumenttyp	online; gedruckt; Sammelwerksbeitrag
ISBN	978-3-658-11992-8; 978-3-658-11994-2
DOI	10.1007/978-3-658-11994-2_24
Schlagwörter	Sekundarstufe I; Schüler; Design; Rechtschreibung; Schreibkompetenz; Item; NEPS (National Educational Panel Study)
Abstract	In the National Educational Panel Study (NEPS), spelling competence is a stagespecific measure for secondary school (Stage 4). This is a challenging task because the current state of research does not offer a theory-based and empirically proven instrument that would allow for longitudinal measurement. In this paper, we present results from NEPS pilot studies in order to illustrate our work. For our research, we used an anchor-item design and IRT-based methods to verify the suitability of the test for the longitudinal survey of spelling competence. We demonstrate the reliability of the tests with regard to dimensionality and discrimination. The person parameter and the item parameter provide an insight into the item difficulty and the students' abilities, thereby guaranteeing that the tests account for individual demands throughout secondary school. Moreover, deviance and the correlational structure of the data are taken into consideration. In summary, the selected test design and the choice of anchor items ensure that an adequate test is administered to each student. We removed about 15 - 30 % of the items in each grade to obtain a fit model. As the results for Grades 5 to 7 are nearly identical, it can be assumed that the test is reliable. The assumed five-dimensional model proves to be most adequate to measure the subskills for all grades because the deviance is lower. (Orig.).
Erfasst von	DIPF \| Leibniz-Institut für Bildungsforschung und Bildungsinformation, Frankfurt am Main
Update	2020/3