Lengthy questionnaires should be shortened to enable better compliance in large-scale trials also ensuring adequate measurement precision.
As the number of questions sufficient to create reliable scales may vary considerably depending on the complexity of concepts and purposes, consecutive participants of a large study, the Swedish Obese Subjects (SOS), received a second mailing of original long-form questionnaires after 1 week to be compared with the SOS short-form measures.
Internal consistency, unidimensionality, robustness and reproducibility were tested for scales reflecting social support and life events, personality traits, health perception and psychosocial functioning.
Very brief generic scales could perform satisfactorily for study-specific purposes in diseased populations.
The short-form social support scales yielded satisfactory psychometric properties and the short life events module was best divided into two multi-item variables.
High and consistent metric values were found for the personality, general health and psychosocial functioning scales.
Our method study thus guided in striking a balance between scaling properties/reliability levels and length of questionnaires/subject burden.
We recommend a method study like ours in every instance where patient-based data recordings are among the primary measures of outcome.
Mots-clés Pascal : Obésité, Epidémiologie, Homme, Questionnaire, Validation test, Version revisée, Psychométrie, Support social, Personnalité, Autoperception, Santé, Fidélité test, Méthode étude, Etat nutritionnel, Trouble nutrition, Suède, Europe
Mots-clés Pascal anglais : Obesity, Epidemiology, Human, Questionnaire, Test validation, Revised drafting, Psychometrics, Social support, Personality, Self perception, Health, Test reliability, Investigation method, Nutritional status, Nutrition disorder, Sweden, Europe
Notice produite par :
Inist-CNRS - Institut de l'Information Scientifique et Technique
Cote : 95-0385442
Code Inist : 002A26N06. Création : 01/03/1996.