Jeanrie, C., Laval University, Canada and Michaux Granier-Deferre, C., Universit Ren-Descartes, France
The translation and the inter-cultural use of tests has been identified many years ago as a major problem for the validity of test scores interpretation. Indeed, as much as translation might take away the initial meaning of an item, the use of a test in a population not initially represented in the sample can raise doubts about the equivalence of the construct's meaning in both populations. The traditional translation method has been, for years, the backward translation, which is based on a first translation from the original version to the target version and then on a second one from the target language to the original language. Both versions in the original languages are being compared to evaluate the quality of the translation process. This method has been criticized, however, because its assessment is based on the original language only and not on the target versions, which can still be inadequate. Other translation methods has been developed, in the later years, to bring a solution to this problem. The conceptual translation method, which is based on the meaning of items instead than solely on its language equivalence, and the decentration method that is based on the simultaneous production of a test in two different languages have been considered as potential improvement to test generalization. This study compares the effect of these three methods on cultural and linguistic equivalence of a scale translated in France and in Quebec. Equivalence is assessed through construct bias and item bias analyses.