ABSTRACT
This chapter explores ways of analysing questionnaires using item response theory (IRT). IRT is a statistical technique that allows for the analyses of the items in questionnaire in relation to their latent traits. The chapter looks at fundamentals of IRT for both dichotomous and polytomous data items and how probabilistic techniques can be used to evaluate the latent structures within questionnaires. The chapter explores study on organic food to illustrate how 1-parameter and partial credit models are used to calculate category and item characteristic curves. When looking at ordinal responses with two or more categories then partial credit models are an adaption of the Rasch model for dichotomous items. The chapter also explores how differential item testing and the information function are used to investigate items in relation to latent model structure. The information function checks reliability of an item in relation to the latent model structure. The chapter investigates validity and reliability issues associated with collapsing Likert scale categories.
