Application of confirmatory composite analysis with correlated error on multidimensional poverty

Errors of indicators in a construct of the measurement model are not correlated with each other. Practically, however, there are a number of cases in empirical research in which uncorrelatedness of measurement errors may not hold. This study aims to compare the measurement model with uncorrelated and correlated errors applied to multidimensional poverty measurement. Multidimensional poverty is an alternative approach in measuring poverty by using ten indicators which are divided into 3 dimensions, namely health (2 indicators), education (2 indicators), and living standards (6 indicators). By using confirmatory composite analysis it is obtained that the discrepancy measures values i.e. geodesic distance, standardized root mean square residual (SRMR), and squared Euclidean distance for both models with uncorrelated and correlated indicator errors are all below the corresponding critical value, i.e. model is fit. However, if a comparison is made between the two, the discrepancy value for the model with the correlated indicator error is smaller than for the uncorrelated model. In other words, it can be concluded that the model with correlated error indicator is adequately fit better with the collected data. Or it can also be said that the model can captures more available information in the data acceptably.