Family size and educational outcomes: empirical evidence through multilevel approach

The aim of this work is to test if family size is correlated with educational outcomes of pupils, analyzing data from the survey INVALSI 2011 on the 5th grade of Italian primary school. Using a multilevel regression model we verified that children with only one sibling get math scores slightly higher than only children, even with the same other observable characteristics, while children of large families (three or more siblings) obtain significantly worse average results. Furthermore, we also found the presence of a contextual effect that leads to worse scores for students attending classes with many children from large familie