Indeks Sosio-ekonomi Menggunakan Principal Component Analysis

In household survey, we could measure socio-economic status through income, expenditure and ownership of valuable goods. Measuring income and expenditure in developing countries has many weaknesses, therefore many researchers prefer to use the ownership of valuable goods as proxy of socio-economic status. Using ownership of valuable goods as proxy indicator creates another problem of having many variables for the socio-economic proxy. To show how to simplify many variables of ownership of valuable goods into 1 socio-economic index. Using prinicpal component analysis with Stata. Using Indonesia Demographic & Health Survey 2002-2003 data, 7 binomial variables of ownership of valuable goods and 3 ordinal variables of housing condition to construct socio-economic indices using principal component analysis (PCA), tetrachoric and polychoric correlation.We used Stata to construct the socio-economic index. Correlation matrices were derived using tetrachoric command for tetrachoric correlation and polychoric command for polychoric correlation. Two socioeconomic indices were constructed, 1 index was based only on 7 binomial variables of ownership of valuable goods and 1 index was based on 7 binomial variables of ownership of valuable goods and 3 ordinal variables of housing conditions. PCA was used to construct those 2 indices. In 7 variables model, the socio-economic index could explain 57% variance and in 10 variables model, the socio-economic index could explain 54% variance. We also showed the use of xtile command to regroup the subjects based on quintile of socio-economic indices. PCA, tetrachoric and polychoric correlation could be used to construct socio-economic indices based on information of ownership of valueable goods and housing conditions. Key words: Socio-economic indices, principal component analysis, tetrachoric correlation, polychoric correlation