Literaturnachweis - Detailanzeige
Autor/inn/en | Rapp, Joel; Allalouf, Avi |
---|---|
Titel | Evaluating Cross-Lingual Equating. |
Quelle | (2002), (25 Seiten)
PDF als Volltext |
Sprache | englisch |
Dokumenttyp | gedruckt; online; Monographie |
Schlagwörter | Quantitative Daten; College Applicants; College Entrance Examinations; Equated Scores; High Stakes Tests; Higher Education; Test Format; Test Items; Testing Programs; Translation; Verbal Tests |
Abstract | This study examined the cross-lingual equating process adopted by a large scale testing system in which target language (TL) forms are equated to the source language (SL) forms using a set of translated items. The focus was on evaluating the degree of error inherent in the routine cross-lingual equating of the Verbal Reasoning subtest of the Psychometric Entrance Test (PET), a high-stakes scholastic aptitude test that is adapted from Hebrew into five different target languages. A research plan inspired by the double-linking method, but modified to fit the context of cross-lingual testing, was used. The error in cross-lingual equating of the verbal subtest of PET was estimated by the difference between two equating conversion functions, each linking the SL verbal section and a parallel verbal section in the TL. A number of test forms were used, with sample sizes ranging from 3,843 to 7,861. The average difference found by the same method over a number of test forms would reflect the degree of overall instability that exists in the cross-lingual equating process. Findings indicate that the differences between the conversion functions in the two alternative links were high. On average, the difference between the conversions was between 1 to 2 raw score points in the equating process of one of the target languages to the source language, and between 0.5 to 1.0 raw score points in the equating process of the other target language to the source language. These differences were obviously caused by a real and systematic problem that underlies the cross-lingual equating process. Some possible reasons for these differences are discussed. (Contains 1 table and 28 references.) (SLD) |
Erfasst von | ERIC (Education Resources Information Center), Washington, DC |