What factors make science test items especially difficult for students from minority groups?

Abstract

Substantial gaps in science performance between majority and minority students are often found instandardized tests used in primary school. But at the item level, the gaps may vary significantly. Theaims of this study are: (1) to identify features of the test items in science (grade 5 and grade 8 students) that can potentially explain group differences; and (2) to analyze what factors make test itemsespecially difficult for minority students. Explanatory variables such as reading load, item difficulty,item writing load, and use of the multiple-choice format are found to be major factors. The analysis reveals no empirical relationships between performance gap and either item subject domain, item test location, or the number of illustrations used in the item. Subtle issues regarding the design ofitems may influence the size of the performance gap at item level over and above the main explanatory variables. The gap can be reduced significantly by choosing “minority friendly” items.