Résumé : Both the Dutch-and French-speakingBelgian educational systemsarecharacterized by a high level of segregation, placing a strong emphasis on academic, socioeconomic, and ethnic make-up of theirstudent population.In this context, school composition is expected to have an effect,howeverthe choice of measurementis notstraightforward. Multilevel analysis was conducted on a Dutch-speaking subsample of 4,618students in 156schools using PISA 2012datainorder to explore different modellingoptionsof multiple compositions. Factorisation is then explored, in combinationwith multilevel modelling, to see whether it provides an adequate strategy to deal withthelimitations of multiple compositionsvariables. In addition, a French-speaking subsample of 2,759 students in 95 schools was used to test measurement invariance,check whether the construct has the same meaning in both groups,and assess whether it can be used in multilevel models. Modelling multiple measures of compositional effect: does factorisation simplify the picture in Belgium?. Available from: https://www.researchgate.net/publication/309742570_Modelling_multiple_measures_of_compositional_effect_does_factorisation_simplify_the_picture_in_Belgium [accessed Nov 8, 2016].