ANALISIS REGRESI LINIER BERGANDA PADA DATA SURVEY UNTUK PEMODELAN TOTAL PENGELUARAN DI JAWA TENGAH, INDONESIA

Survey data is collected using a sampling design to capture Population phenomena. One of the sampling techniques used is complex design. Complex design is a sampling technique other than Simple Random Sampling (SRS). Sampling weight is used for the estimate from the complex design to result in unbiased estimation. Regression analysis using survey data needs to consider complex design. This study models total expenditure in Central Java province using March 2021 Susenas data, with a sample of 29,870 households, 3,005 primary sampling units, and 65 strata. The best model produced from this study is 25.51%. Based on the regression model, the characteristics of households with higher expenditure on categorical independent variables include households living in urban areas, heads of households that are male, married, graduates/postgraduates working in the service sector, and households not receiving Government assistance. Meanwhile, the positive coefficient of continuous variables include the age of the head of the household, hours worked during one week, and the number of households living in one house