This paper assesses how the integration of ICT in education has affected the mathematics test scores for Italian students measured by the Programme for International Student Assessment 2012 data. The problem of endogeneity that affects survey data in this area, is addressed by applying the Bayesian Additive Regression Trees (BART) methodology as in Cabras & Tena Horrillo (2016). The BART methodology needs a prior and likelihood functions using the Markov Chain Monte Carlo (MCMC) algorithm to obtain the posterior distribution. Controlling for socio-economic, demographic and school factors, the predicted posterior distribution implies an increase, on average, of 16 points in the test scores. The result indicates that the use of ICT at school has a positive and strong impact on mathematic test scores.
ICT; Bayesian additive regression tree; Posterior distribution; PISA