Analysis of Longitudinal Regression Model Using the Generalized Estimating Equation (GEE) for the Child Welfare Composite Index (CWCI) in West Java

Abstract

The purpose of this study was to determine the Generalized Estimating Equation (GEE) analysis as one of the longitudinal data regression methods for composite index data on children's welfare (CWCI) in West Java Province. GEE analysis uses four different correlation structures to determine the best model, namely: independent, exchangeable, AR (1), and unstructured. The determination of the best model of the four structures uses the Quasilikelihood Information Criterion (QIC). The West Java CWCI longitudinal data used consists of 5 independent variables, each of which is a dimension of survival, protection, growth and development, participation, and identity in 27 districts and cities in West Java for the period 2015 – 2017. This research is important as an alternative approach of longitudinal data regression analysis when we found a correlation problem in the condition of observed data.