Selected Methods of Categorical Data Analysis and Their Application in Consumer Behaviour Research

This paper explores the methods of categorical data analysis, with special focus on their pros and cons and their possible application in the research of Czech consumers' shopping habits. To that end, a large survey mapping shopping preferences of Czech consumers (mainly the shopping behaviours of different genders in relation to their income, education, age, place of residence etc.) has been used as a source of data. The survey results were analyzed using the contingency tables analysis, including the Pearson's chi-squared test of independence in contingency tables. The dependence intensity was identified via Pearson's Contingency Coefficient and the impact of individual categories was assessed with the post-hoc residual test. The influence of several combined factors on the payment method was tested through logistic regression; the model parameters were estimated by the Maximum Likelihood Estimation and the ROC curve allowed for the assessment of the model's quality. Every method proved a statistically significant dependence of the most frequent payment method on the education and age of respondents, number of household members and the size of respondents' place of residence.