Abstract
The research findings presented in this paper illustrate how the “value added” of schooling can be assessed empirically using cross-sectional data. Application of the regression-discontinuity approach within a multilevel framework produces both an estimate of the absolute effect of 1 year schooling and an estimate of the variation across schools of this effect. In the study reported here, the approach was applied to both a cross-sectional and a longitudinal dataset. The research findings indicate to what extent different results are produced when cross-sectional as opposed to longitudinal data are analysed.
Notes
1. Note that these percentages are not really comparable with one another. Both figures express different aspects of the same phenomenon. It seems possible to convert both percentages to effect sizes that have been defined in relation to interventions in which there is a control and an experimental group (see Tymms, Citation2004). We will briefly return to this matter in the final section, but a detailed coverage of this topic is beyond the scope of this article.
2. The estimated effects of grade in the models with quadratic effects of age included (see ) indeed hardly differ from the effects of grade in models that include an interaction effect of age with grade. Details on these analyses can be obtained from the first author.