Fitting directed acyclic graphs with latent nodes as finite mixtures models, with application to education transmission

(2015) - Antonio Forcina, Salvatore Modica

Journal: arxiv:1209.0876
Link:: http://arxiv.org/abs/1209.0876
DOI::
Links::
Tags:: #paper #NCDS #Attainment
Cite Key:: [@forcinaFittingDirectedAcyclic2015]

Abstract

This paper concerns estimation, inference and computation of causal effects in nonlinear structural equation models containing multiple latent variables. We apply our methods to the study of intergenerational education transmission. Education transmission involves two unobservables, the endowments of parents and child. Existing analyses control for one, leaving the other out of the model. This paper proposes a discrete finite-mixture model whose system of equations includes simultaneously family and child unobservables. The maximum likelihood estimates are obtained via an efficient EM algorithm, and analysis of causal effects follows and implements Pearl’s model of causality.

Notes

“flow of education is not automatic from parents to children, but it goes through, to a non-negligible extent, if family pushes.” (Forcina and Modica, 2015, p. 1)