ABSTRACT
Inclusion of sociodemographic characteristics of children and primary caregivers as well as demographic characteristics of the household and country adds levels of nuance to the interpretation of the results. This chapter analyses how the representations of different process, person, and context systems fare relative to one another in predicting the early childhood development outcomes. It employs machine learning methods to assemble all the variables together in an atheoretical model to gain an understanding of the relative contributions of each variable in predicting the five domains of early childhood development. The five domains of early childhood development include child literacy and numeracy development, child socio-emotional development, child physical health development, early childhood development of approaches to learning, and overall early childhood development. The chapter focuses on the dichotomous measures of the Early Childhood Development Index and subindexes; that is, the information about which children are on-track or off-track according to United Nations International Children's Emergency Fund criteria.
