Analisis Spasial untuk Data Pekerja Anak di Pulau Jawa dan Faktor-faktor yang Mempengaruhinya

Abstract

The population of Indonesia were 237.6 million people and about 82 million of whom are children population (SP,2010). Child is a person under 18 years old, which have the potential to be developed in order to active role in the development of the nation in the future. In Law No.23 of 2002, mentioned rights acquired children population, but in the field of employment are 3.4 million children aged 10-17 years old who worked (BPS 2012). Child labor is a social problem for a country, because it will have a negative impact on the quality of a country's future successor. Three contributor to the province's largest number of child workers in the provinces of West Java, Center Java and the East Java, suspected in a region close relationship problem of child labor. Thus this study aims to see the relationship closeness in the area of child labor problems in Java, which saw the city/regency where become a hotspot and coldspot area, and analyzes the factors that influence it. The spatial correlation in this study seen based on the value of the moran indexs, and to see the visually of spatial correlation is based on the moran scatter plot. Variables that are used in the preparation of the model are variables that are not multikolinear. The child labor data is a count and have a poisson distibution, so model of Poisson Regression Analysis is a model that is used to view the amount of the alleged value of the parameter of each explanatory variable. Breusch Pagan Test (BP-Test) was conducted to determine the presence of spatial diversity. BP-Test results indicate the existence of spatial diversity means Poisson model of analysis results can not be used in general for all cities/regencies in Java, therefore the local model analysis used Poisson Geographically Weighted Regression (GWPR). The results showed the model Poisson Regression Analysis results formed AIC has a value of 385293, while the local model GWPR AIC has a value of 207010, it means that the best model is the local model GWPR the child labor problem in Java. GWPR local models also has the assumptions of nonstationary, it is seen by comparing the standard error Poisson Regression Analysis models and the Inter Quartile Range (IQR) of GWPR. Seven variables used in the preparation of the Poisson regression analysis models and local models GWPR already represent all aspects (child, family and common ). The child aspects explained by the variable numbers of children do not attend school, primary school gross enrollment rate, gross enrollment rate of SMP, and children who are married. Family aspect is represented by the variable average monthly household expenditure (expand), while the common aspect is represented by variable illiterate and Total of SMA.