The predictive power of Dutch end of primary school tests for educational attainment in ninth grade

Abstract

Dutch secondary education is a multi-tiered system, and many students are placed in a single track in the first year (seventh grade) of secondary education. As part of the placement procedure, all sixth-grade students take an end of primary school test. Although these standardised attainment tests (SATs) are considered high-stakes tests, the predictive accuracy of many of these SATs has not been studied comprehensively. In the current study, three SATs commonly used in The Netherlands (CET, IEP and R8) were compared on their predictive accuracy of educational attainment in ninth grade. The predictive accuracy was also investigated for particular track recommendations, using multiple cohorts of population data, each of around 160,000 students. For the two most recent cohorts, CET was overall a more accurate predictor of educational attainment in ninth grade than R8, which in turn was a more accurate predictor than IEP. All three attainment tests had only mediocre to acceptable predictive power overall. Furthermore, all three tests had very poor predictive power for certain single-track recommendations. For these track recommendations, many students attained a lower or higher educational level in ninth grade. This study is the first to compare the predictive accuracy of these SATs. While statistical noise is normal when using tests, since different tests are used that vary widely in predictive accuracy, there is much more noise than is acceptable. As long as the predictive accuracy is poor to mediocre, these SATs should not have a large role to play in the Dutch placement procedure for secondary education.