Skilvirkni skimunar með Talnalykli : leit að slökum nemendum í stærðfræði

Abstract

The utility of screening for mathematical difficulties was studied in the normative sample (7V=1914) of Talnalykill, a criterion- and norm-referenced standardized test of mathematics achievement. The aim was to use one or two subtests to find students that were one standard deviation or more below the mean on the test as a whole. The subtest Arithmetic had little accuracy in first grade. However, about 90% of low-achieving students (one standard deviation or more below the mean) could be found in grades 2 to 7 when using a criterion of 9 or below (M=10, sd=3) on the subtest. This would result in 60% reduction in the number of assessments with the test as a whole. Two step screening with both Arithmetic and Numbers with 90% of the low-achieving students found, resulted in assessment of all students with Arithmetic, 43% with Numbers and 33% of the students with the test as a whole. If 80% discovery rate for low-achieving students is adequate, this could be achieved by assessment of 100%, 31% and 21% of the students with Arithmetic, Numbers and the test as a whole, respectively. A discovery rate of 67%, by comparison, required assessments of 100%, 20% and 15% of the students respectively with the two subtests and the test as a whole. The results confirm that low-achieving students in mathematics can be found cost-effectively by using Talnalykill. Screening for the total score by one or two subtests has high utility and reasonable cost. The screening method is also very flexible, since the information provided enables the fine-tuning of the screening criterion in order to optimize the trade-off between discovery rate and cost. This flexibility enhances the value of the results for use within individual schools or school systems.