APPLICATION OF MIXED EFFECT MODEL FOR ASSESSING DEMOGRAPHIC INDICATORS

Abstract

Mixed effect models have been implemented widely by researchers who are doing experiments. Most population studies are done by observational studies such as survey research, or by secondary data analysis from census data. Independent variable of province determined by a researcher may be considered as fixed effect component, while random samples of districts or cities drawn from each province may be considered as random effect component. Three provinces in Java Island, East Java, Central Java, and West Java, are considered as fixed effect component. Fourteen random samples that represent district or city are drawn from each province. A number of demographic indicators from each district or city are collected. The data are obtained from population census performed in 2010 in three provinces. A mixed effect model is constructed and parameters of random effect component are estimated.