The Use of Cluster Analysis for Development of Categorical Factors in Exploratory Study: Facts and Findings from the Field Research

This paper describes the development of categorical latent variables and their use for the typology development. The main idea of this modeling was to investigate the possibility of using the classification methods instead of factor analysis for development of the final latent variable which can cumulatively explain the set of primary indicators. The modeling is based on the empirical findings from the online retail consumers' behavior study. Selected data allowed confirm the statement that even the small data sets using the classification data analysis methods can display the significant Ecological Validity. The modeling was performed in three steps. First, the number of primary indicators was reduced using the factor analysis. Based on it several latent variables were created. Second, the k-mean cluster analysis instead of secondary factor analysis was used for development of three cluster variables that represent six clusters in total. Third, all three variables were used for development of the final latent variable which is categorical and represents the eight theoretically possible and five empirically confirmed categories.