Improving Spanish classroom assessment via logistic regression: lessons from the Rasch model. (2nd January 2020)

• Record Type: Journal Article
• Authors: Wallace, Glen H.
• Is Part Of: Journal of Spanish language teaching. Volume 7:Number 1(2020)
• Page Start: 51
• Page End: 63
• Publication Date: 2020-01-02
• Subjects: Language assessment -- classroom language assessment -- Rasch modeling -- testing and evaluation in language teaching -- Spanish Language Teaching (SLT); Evaluación de lenguas -- evaluación de lenguas en el aula -- el modelo de Rasch -- evaluación en la enseñanza de lenguas -- Enseñanza del Español como Lengua Extranjera o Segunda (ELE/EL2); Spanish language -- Study and teaching -- Periodicals; 468
• DOI: 10.1080/23247797.2020.1771009
• Languages: English
• ISSNs: 2324-7797

Abstract

ABSTRACT: Each year in the United States, hundreds of thousands of undergraduate students enroll in introductory Spanish courses. For many students, passing these classes is a prerequisite for graduation. Consequently, it is imperative that the exams used in these courses are of the highest possible quality. For decades, the Rasch model (a special case of logistic regression) has been used to evaluate high-stakes language assessment, but has rarely been applied to exams used in classrooms. The current study employs the Rasch model to evaluate a midterm exam administered in the fourth week of an introductory Spanish class ( n = 399 across 17 sections). This model identifies which questions best evaluate students at different proficiency levels and shows that the test reliably assigns scores near the pass-fail cutoff. The results of this study suggest that for this sample, items testing article-noun agreement were the most appropriate for the average student.