Do Ethnicity, Religion and Gender Matter on Low-Stakes Assessment Tests: A Field Experiment

Abstract

Standardized assessment tests with low stakes for the student are increasingly used as an important policy and research tool. Thus, it is important that they reflect ability accurately. We examine how 8th graders perform, based upon gender and ethnicity, on a real-life high-stakes test, in comparison to a low-stakes test. Overall, Jews have a significantly smaller grade difference between the high-stakes and the low-stakes test (0.7 STD) than Arabs (1 STD). The smaller grade difference suggests that 60% of the Jewish-Arab performance gap in national assessment tests can be attributed to effort differences in the test itself rather than in student ability. This study shows that educational reforms and educational and economic policy must use caution when assessing the policy results across population groups according to low-stakes assessment tests.