Retention and science performance in Portugal as evidenced by PISA

Abstract

Problem Statement: The question of academic retention has emerged as a powerful discourse in educational policy in Portugal. International assessment programs, such as the Program for International Student Assessment (PISA), have fueled the controversy, and retention rates are sometimes implicated as an underlying cause for the generally poor performance observed. Purpose of Study: The data from PISA 2006 provide an opportunity to examine this question with a large sample of 15-year-old students. We report associations between retention and science performance, an analysis of the characteristics of retained students, and a hierarchical linear model of the effect of retention on performance, controlling for economic-social-cultural status (ESCS). Research Methods: The 2006 PISA sample in Portugal consisted of 5109 students in 173 schools, all between the ages of 15 years 3 months and 16 years 2 months. Missing data resulted in 5013 usable cases for this study. Since PISA has no single variable to indicate a student’s status as “retained at least once” or “never retained”, the first step was to operationalize retention with the available data. Mean analyses were based on 80 weighted replicas of 5 plausible values of science performance and its sub-domains. Findings: Portugal is among the PISA participants with highest retention. Only 4 (of 57) PISA participants revealed a higher proportion of the age-based sample attending the 7th grade. Students who were retained at least one year differ systematically, on several social dimensions, from those not retained. The retained are more likely to be boys, immigrants, in public schools, from small towns and villages, and of lesser economic means. Mean performance in science is directly associated with the number of student retentions (though this is confounded with other variables). Being retained is a more powerful predictor of science performance than is ESCS. Conclusions: Retention practices in Portugal are outside de norms of other OECD countries. Neither curriculum nor teaching practice seem implicated in the poor science results observed in Portugal. Low performance is a partial artifact of age-based sampling coupled with high retention.