[Internal statistical validation of a quality of life questionnaire]

In this paper we present two different statistical approaches to evaluate the psychometric properties of a quality of life questionnaire. First the study of the factorial structure is briefly exposed. Then we present the unidimensional classical models. They are based on the linear relationship between the observed score and the true score. The definition of the reliability was first addressed in this classical framework. Its estimation with the Cronbach alpha coefficient is one important feature of the evaluation of an instrument. More recently, modern response theory gives a better statistical framework to deal with unidimensional latent traits. These models describe the probability of positive answer to an item as a function of the actual value of the latent trait and an item parameter. We expose the principles of the Rasch model: hypothesis, estimations methods and fit tests. Finally practical applications to the validation process of a questionnaire are explored with data from a study of a short French version of the SIP questionnaire.