Measuring beginner reading skills: An empirical evaluation of alternative instruments and their potential use for policymaking and accountability in Peru

Based on analysis of reading performance data from 475 third-graders in Peru, this study makes recommendations on improving reading tests, choice of reading standards, and how to present the results at the school and individual levels. The paper reviews the literature on using reading skills measurement in the early grades to guide policymaking, strengthen accountability, and improve education quality. It uses data generated from the same students using two common approaches to measuring reading skills: an individually-administered oral fluency test, and a group-administered written comprehension test designed by the Ministry of Education for the 2006 universal standard test of second grade reading comprehension. These two approaches have sometimes been presented as competing alternatives, but the paper shows that it is better if they are used together, as complements. Based on psychometric analysis, the paper shows that both the oral and written tests adequately measured students– reading abilities. The results show that reading fluency and comprehension are correlated: fluent readers are more likely to understand what they read than non-fluent readers. The strength of the fluency-comprehension relationship depends on the level of fluency, the difficulty of the questions, and social characteristics of the school. The paper recommends using improved versions of both tests to evaluate early grade reading skills, as a central element of a system of accountability for results. It proposes a model for reporting test results desgned to highlight the importance of reading standards, mobilize the education community to reach them, track progress, and identify students in need of extra support.