Les sondages moins rigoureux sont-ils moins fiables?

Bill C-83 proposes to regulate the publication of polls and their methodology during electoral campaigns. It would be necessary to determine the type of pertinent methodological information in order to judge the quality of the poll. This article presents results of a study on the relation between methodology used in polls published in Quebec during the May/June 1997 federal electoral campaign and the quality of the estimates of voter intentions. Analysis verifies that, when the methodology used was less rigorous, the spread between the polls and the true voter intentions was larger and less stable even when taking account of errors due to sample size. These results underline the necessity to enforce requirements about the publication of the methodology of the surveys reported during electoral campaigns and of revising Bill C-83 accordingly.