Empirische Grundlage, Stichprobenausfall und Adjustierung im PISA-Längsschnitt 2012-2013

This article presents the empirical basis for the PISA longitudinal study 2012–2013. The composition of the additional national sample of students of the 9th grade is illustrated. Findings with regard to non-participation (dropout) for the second point of measurement will be presented at the level of schools and students. The results show that the dropout is related with characteristics of family background, the level of competence in the three domains (mathematics, reading and science) in 2012, and with other student and school demographic variables at the level of schools as well as students. These findings are taken up in order to derive weight adjustments, which could counteract a systematic bias of the results caused by the selective dropout. The limits of the possibilities for correction are discussed. The PISA longitudinal study is characterized by a mandatory assessment for the first point of measurement followed by a voluntary test for the second point of measurement. Thus, the present article provides valuable information about probable causes of non-participation in voluntary studies and the associated risk of systematic bias for population estimates.