This article provides evidence regarding differences in the performance on academic tests in secondary school between students who had attended preschool and those who had not. A non-parametric method based on the generation of counterfactuals to decompose gaps between observable and unobservable factors is used to analyse data gathered by the Programme for International Student Assessment (PISA) between 2009 and 2012 for a number of Latin American countries. This analysis reveals the existence of socioeconomic segregation in terms of access; considerable score differentials (conditional on the controls used), which increase in line with the length of time spent in early childhood education; and significant differences across countries with regard to those differentials, which also tend to be greater in the case of reading test scores than in mathematics test scores.
