Examination of Indices of High School Performance Based on the Graded Response Model. Issue 2 (14th March 2019)

• Record Type: Journal Article
• Authors: Allen, Jeff; Mattern, Krista
• Is Part Of: Educational measurement. Volume 38:Issue 2(2019)
• Page Start: 41
• Page End: 52
• Publication Date: 2019-03-14
• Subjects: ACT scores -- college degree attainment -- college readiness -- graded response model -- high school GPA; Educational tests and measurements -- Periodicals; Educational tests and measurements -- United States -- Periodicals; 371.262
• DOI: 10.1111/emip.12250
• Languages: English
• ISSNs: 0731-1745

Abstract

Abstract: We examined summary indices of high school performance (coursework, grades, and test scores) based on the graded response model (GRM). The indices varied by inclusion of ACT test scores and whether high school courses were constrained to have the same difficulty and discrimination across groups of schools. The indices were examined with respect to skewness, incremental prediction of college degree attainment, and differences across racial/ethnic and socioeconomic subgroups. The most difficult high school courses to earn an "A" grade included calculus, chemistry, trigonometry, other advanced math, physics, algebra 2, and geometry. The GRM‐based indices were less skewed than simple high school grade point average (HSGPA) and had higher correlations with ACT Composite score. The index that included ACT test scores and allowed item parameters to vary by school group was most predictive of college degree attainment, but had larger subgroup differences. Implications for implementing multiple measure models for college readiness are discussed.