Methodology for identifying homogeneous consumer groups based on qualitative data

Abstract

In this article, the author proposes a method for identifying homogeneous consumer groups based on qualitative data. The problem is that when researching the end-user market, information is often presented not in quantitative but in qualitative form. The random variables with which mathematical statistics deal are usually assumed to be numeric. Therefore, among researchers there is an opinion that achieving at least an interval level of measurement is always desirable, since it expands the researcher capabilities, giving him grounds to use data mathematical and statistical analysis traditional methods. Sociologists, on the other hand, emphasize the qualitative data enormous role in the respondents’ study. The presented methodology is based on cluster analysis, differs from the applied market segmentation methods in that it uses cluster analysis algorithms developed concerning qualitative indicators, and involves a proximity measure use that allows one to determine the natural weights between clustering variables. Also, the technique provides for the optimal partition determination based on the changes’ graph in the average internal communication, depending on the selected clusters’ number. The optimal among the partitions set is considered to be a partition in which the average internal connection increases sharply in comparison with the previous partition. Provided that the clusters’ number in each subsequent partition in comparison with the previous one is greater by one. Thus, the methodology allows identifying the existing market structure.