Het Ongelijk van Netelenbos?

Within the context of school success of ethnic minority children in the Netherlands, it is currently debated what criteria should be applied by the Ministery of Education in assigning extra funds to children at risk. To this date, two criteria are used: level of parental education and parents' country of origin. As both criteria do no longer fully comply with the needs, it is proposed to test the children entering Kindergarten, with respect to their language proficiency in Dutch and/or their intelligence. This article reports the results of a longitudinal study on the language and school development of monolingual Dutch and bilingual ethnic minority children who were followed from entering Kindergarten at the age of 4, to the end of grade 6 in primaty school at the age of 12. During the first four years, Dutch language tasks ( ) were annually administered. Besides, children took an intelligence test (RAKIT) at the age of five. In grade 6, at the age of 12, a test battery of Dutch language proficiency, writing, reading, mathematics, science and information processing tasks were administered (Cito Eindtoets) to operationalise 'school success'. Finally, several background data were gathered, such as level of parental education and home language use. The results of regression analyses tentatively show that a language test administered in Kindergarten is the best predictor of school success (Multiple R= .432), that the intelligence test adjusts 4% to that, and that both home language use and parental education add another 2%. For the ethnic minority children alone, the Dutch language test (R2=.275) and home language use (adding 5%) are the only two predicting factors. However, if we look into the data as if we were teachers and had to decide whether or not to assign extra funds to a specific child on the basis of his/her test scores in Kindergarten, 29% of those assignments had been wrong: 14% wrongly not assigned ('false negatives'), and 15% wrongly assigned ('false positives'). For only 11% of the children the assignment of funds would have been predicted correctly. From this perspective, testing children in Kindergarten is not the most appropriate way to assign extra funds to children at risk.