Assessing Partial Triadic Analysis with MaximumEntropy Bootstrap: an application to BES Italianeducation indicators

Abstract In this paper we analyse a set of Italian equitable and sustainable well-being indicators, related to the education and training domain, considered at a regional level over different years. To this end, the Partial Triadic Analysis - a multiway approach which allows to properly deal with time-dependent data structures - will be used in conjunction with a bootstrap method to carry out the assessment of the outcomes obtained. The adopted resampling scheme - called Maximum Entropy Bootstrap - proved to be an appropriate analytical tool, given its ability to preserve the longitudinal data structure without incurring in the usual restrictive assumptions of more traditional bootstrap methods. To the best of the authors' knowledge, this is the first time a bootstrap method for longitudinal data has been employed in PTA.