Marozzi presented a simple method to reduce the dimension of indicator variables, compared it to PCA and showed that is markedly simpler to be used and requires milder assumptions. An application to university student satisfaction was discussed. In this paper we present further research about this method. Firstly, we propose two alternative ways for reducing the dimension of indicator variables of which one resembles regression forward selection, and apply them to the data considered in Marozzi (2008). General indications on which method choose are given. Secondly, we evaluate how the choice of the correlation coefficient influences the results.
Some Notes on Indicator Variable Reduction
MAROZZI, Marco
2009-01-01
Abstract
Marozzi presented a simple method to reduce the dimension of indicator variables, compared it to PCA and showed that is markedly simpler to be used and requires milder assumptions. An application to university student satisfaction was discussed. In this paper we present further research about this method. Firstly, we propose two alternative ways for reducing the dimension of indicator variables of which one resembles regression forward selection, and apply them to the data considered in Marozzi (2008). General indications on which method choose are given. Secondly, we evaluate how the choice of the correlation coefficient influences the results.
I documenti in ARCA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
