Development of model poverty in Java using Meta-Analysis Structural Equation Modeling (MASEM)

Indonesia still has many fundamental problems in terms of government, religion, socio-cultural issues and in particular the problem of poverty. Java is one of the Indonesian islands with 6 provinces and the largest population among the other islands of Indonesia, therefore the problem of poverty on Java must be given greater attention. Specific data on poverty in Java must be collected from several populations. In parallel with the development of statistics, the generalization of the population is not only based on the results of a study but also requires a lot of research, so that the development of statistical methods is necessary. The meta-analysis (MA) is a method that can combine a variety of studies to enable optimal research. The meta-analysis associated with Structural Equation Modeling (MASEM) is one of the developments of statistical methods to examine the influencing factors. The Structural Equation Modeling (SEM) model including poverty as an endogenous variable and economic factors, human resources (HR) and health as exogenous variables have been studied in each province of Java. From the results of these studies, will be developed by the MASEM method through the generalized least squares (GLS) approach in order to show the adequate poverty model on Java. The results of T-test model with the implementation of the MASEM GLS on MatlabR2018 (GLS-M) have shown that poverty was affected by health and HR with coefficient being -0.2761 and -0.2111 but was not affected by Economy. The accuracy of the poverty model, influenced by exogenous variables of economy, health, and human resources, is 50.6% with a chi-square value of 637.0749. While the performance of MASEM with TSSEM on RStudio is better than GLS-M or GLS on RStudio at the model fitting stage.