Om statens verktyg för skoljämförelser: Vem vill dansa SALSA?

Abstract

Sweden has an indicator system for grade 9 in comprehensive schools building on value added measures. The aim is to stimulate improvement, and not to rank schools in quality. Data are easily available at the webb. A government committe suggested recently that the system should be used by a new, more powerful state inspection. Whatever purpose, the problem with the statistical model is its weakness in taking account of the variance in marks that cannot be related to school composition in terms of parents’ education, ethnicity, and gender distribution among pupils. Explained variance is about 40 per cent, and the rest (the residual) is taken as the indicator of a school effect. The article show’s some ways to improve this: Structural equation modelling, and pupil questionnaire data about the transfer of cultural capital in families. By these means the explained variance is increased to 55 per cent. By using school adjustement as another value added measure explained variance was 68 per cent. Variation between schools is markedly higher with school adjustement than with marks. Still, there are heavy problems with the model. The article concludes by discussing what functions this indicator system may have in the steering system as a whole.