6533b832fe1ef96bd1299ff5

RESEARCH PRODUCT

Validité et fiabilité des questionnaires d'évaluation de la qualité de vie : une étude appliquée aux accidents vasculaires cérébraux

subject

description

This article deals with various means, which could be used by designers in order to estimate error terms in questionnaire tests results, thanks to psychometry. The study must be conceived so as to minimize systematic errors (validity) and random errors (reliability). Methods settled to test the validity of a study are very subjective, even if some of them seem to be more "scientific", being based on the principle of correlation. However the last-mentined don't stand up to some theoretic weakness and other explicit subjective methods should be chosen instead : for instance experts consensus.Methods of reliability evaluation have been developed under strong hypothesis to estimate a reliability coefficient. A very commonly-used method is Cronbach's alpha, but some criticisms have spread concerning the right interpretation of the coefficient. The article concludes on the incompleted aspect of the theory of error measurement. However it's necessary to go on justifying each study thanks to the means which are available : that is to say the expert consensus to validate the choose of items and the alpha coefficient to attest their reliability. Nevertheless, we must be careful when interpreting the results. The last part of this work deals with the empirical results which have been achieved thanks to these methods in a survey about the quality of life post stroke.