Une méthode pour faire valider la traduction d'une questionnaire

Abstract

This paper describes Haccoun's (1987) technique for validating a translated questionnaire. This method is based on the idea that if a questionnaire is well translated, bilingual subjects will provide equivalent responses to questions in either language. A single group of bilingual subjects is given both language versions of the questionnaires at two different times in random order. Subsequently: 1) test-retest reliability coefficients are computed for the original and translated versions; 2) correlation coefficients between the original and translated versions of the instrument are computed and compared (simultaneous correlations between languages); 3) the correlations between the original version at time 1 and the translated version at time 2 and vice versa are computed and compared (cross-correlations); and 4) the cross-correlations are compared to the test-retest reliabilities within each language. The final step indicates whether the translated version of the instrument is equivalent to the original. The authors use Haccoun's technique to demonstrate that a French translated version of Alcock et al.'s (1990) questionnaire on nurses' perceptions of nursing research is reliable and statistically equivalent to the original.