STATISTICAL ANALYSIS OF MULTIPLE RESPONSE ITEMS

Abstract

The present paper proposes some methods of statistical analysis for so called multiple response items (mentioned as “Mult Response” in SPSS manual (Hull and Nie [3])), which has been analyzed only descriptively. The proposed methods include asymptotic interval estimation and hypothesis testing for proportions of categories within one item and between two items, some association measures and their asymptotic interval estimation. The asymptotic theory is developed under the assumption of a multinomial distribution for frequencies of category combinations by using the δ-method.