Analysing databases of quantitative and qualitative data with missing qualitative values

Abstract

The analysis of databases with qualitative and quantitative data iscomplex. A new procedure was applied for the estimation of quali-tative missing values: we first applied optimal scaling (OS) to com-plete data; next, correlations between variables were studied; third,the relationships between variables were explored using Path Anal-ysis; and, finally, a homogeneity analysis based procedure was fol-lowed in order to obtain the imputations that were used in the finalOS analysis. The OS analysis was performed a second time, and theresults were interpreted using CATPCA and Property Fitting.