Analysis of Preference Data Using Intermediate Test Statistic

Abstract

Intermediate statistic is a link between Friedman test statistic and the multinomial statistic. The statistic is based on ranking in a selected number of treatments, not necessarily all alternatives. We show that this statistic is transitive to well-known test statistic being used for analysis of preference data. Specifically, it is shown that our link is equivalent to the multinomial test statistic when only one alternative is selected for ranking but to Friedman statistic when all alternatives are selected for the purpose. Having used intermediate statistic in knowing whether all the alternatives on why people choose to be educated have the same probability with respect to their preferability, it was discovered that at least one alternative is significantly preferable to the other.

Keywords:Preference data, Friedman statistic, multinomial test statistic, intermediate test statistic.