Exit Exams - neither a panacea nor a curse
Cosentino de Cohen, Clemencia
Program for Evaluation and Equity Research (PEER) The Urban Institute Washington, DC, USA
Testing regimes at the end of secondary school vary greatly throughout European educational systems, despite the push to harmonize certification requirements across jurisdictions. Testing efforts which rely on different types of exams to certify completion of secondary school and often to give access to university studies may have important consequences for student learning, future life chances, and social inequality. This research presents a comparative analysis of the relationship between testing regimes and student achievement. Using a micro dataset created from individual records of students tested in the Third International Mathematics and Science Study and relying on a thorough qualitative review of testing regimes this paper explores the relationship between exit exams (particularly external and high stakes) and student achievement. A guiding concern of this research is whether exit exams are associated with greater inequalities in learning outcomes, signaling that inequalities may be generated by, or maintained through, existing examination regimes. Unlike prior research on this topic, which generally models the relationship between exams at the end of secondary school and achievement in 7th or 8th grade, this work aligns achievement and testing appropriately: both measured at the end of secondary school. Results of multivariate linear models (weighted least squares and quantile regressions) indicate that external exit exams have a small positive association with student achievement. Exams that are high stakes are negatively associated with student achievement, unless those exams are external. Exit or graduation exams that are external and high stakes display a positive association with student achievement. Against expectations, the results suggest that the influence of exams is not greater among higher achieving students. On the contrary, exams may have a stronger (and positive) effect among lower achievers. Also contrary to expectations, external exit exams are not associated with higher variance in student achievement. The paper concludes with a discussion of the implications of these results, particularly in the context of possible inequalities in student achievement. The paper also includes a discussion of methodological challenges encountered when attempting not to model using the nation as the unit of analysis. The nation proved to be an inescapable dimension of analysis.