engleski

Independence Testing of Variables with Nominal Measure - Customs Component in Croatian Brand

The most common procedure of testing independence of variables with nominal measure in classical statistic sense is Chi-square. It is rather complex procedure and sometimes it does not lead to final conclusions, because of statistic-technical backsets. There are two typical situations when statistic-technical backset appears: a great number of variations of nominal variable/s that on artificial way increases degrees of freedom and a great number of expected counts less than 5. Both situations result with wrong, usually opposite conclusion about independence of variables with nominal measure. That is reason why a much more precise procedure based on proportions testing is established in this paper. Namely, authors have perceived that in such kind of testing valid results can be realized using hypothesis testing about difference of two proportions. During quantitative researches in social science authors have recognized it as the most appropriate approach to survey research analysis. So, in this paper a survey research on customs component in Croatian brand is presented to verify those statements. Case study is very interested and useful in scientific sense by itself, not only as the practical example used to sustain mathematical-statistic analysis conclusions. This research was initiated in order to discover possibilities of including Zabiokovlje, as a part of Split-Dalmatian County, in Croatian brand in sense of sustainable development. A survey has been carried out in this area and more than 600 questionnaires have been analysed. Comparative research results analysis by proportions testing confirms the same conclusions as Chi-square of independence.

independence testing; comparative survey research; Croatian brand

nije evidentirano