Setting a standard for low reading proficiency. A comparison of the bookmark procedure and constrained mixture Rasch model.

Autor/inn/en	Feseker, Tabea; Gnambs, Timo; Artelt, Cordula
Titel	Setting a standard for low reading proficiency. A comparison of the bookmark procedure and constrained mixture Rasch model.
Quelle	In: PLOS ONE, 16 (2021) 11, 23 S. PDF als Volltext kostenfreie Datei (1); PDF als Volltext kostenfreie Datei (2); PDF als Volltext kostenfreie Datei (3) Link als defekt meldenVerfügbarkeit
Sprache	englisch
Dokumenttyp	online; Zeitschriftenaufsatz
ISSN	1932-6203
DOI	10.25656/01:26331 10.1371/journal.pone.0257871
URN	urn:nbn:de:0111-pedocs-263315
Schlagwörter	Bildungsforschung; Lesefertigkeit; NEPS (National Educational Panel Study); Deutschland
Abstract	In order to draw pertinent conclusions about persons with low reading skills, it is essential to use validated standard-setting procedures by which they can be assigned to their appropriate level of proficiency. Since there is no standard-setting procedure without weaknesses, external validity studies are essential. Traditionally, studies have assessed validity by comparing different judgement-based standard-setting procedures. Only a few studies have used model-based approaches for validating judgement-based procedures. The present study addressed this shortcoming and compared agreement of the cut score placement between a judgement-based approach (i.e., Bookmark procedure) and a model-based one (i.e., constrained mixture Rasch model). This was performed by differentiating between individuals with low reading proficiency and those with a functional level of reading proficiency in three independent samples of the German National Educational Panel Study that included students from the ninth grade (N = 13,897) as well as adults (Ns = 5,335 and 3,145). The analyses showed quite similar mean cut scores for the two standard-setting procedures in two of the samples, whereas the third sample showed more pronounced differences. Importantly, these findings demonstrate that model-based approaches provide a valid and resource-efficient alternative for external validation, although they can be sensitive to the ability distribution within a sample. (DIPF/Orig.).
Erfasst von	DIPF \| Leibniz-Institut für Bildungsforschung und Bildungsinformation, Frankfurt am Main
Update	2024/1