|
|
Un deuxième exemple : depuis le Recensement de 1981, le formulaire complet, lequel comprend une cinquantaine de questions et que reçoit un ménage sur cinq, ne vise plus les pensionnaires d’un établissement institutionnel (un hôpital, une maison de repos, une prison, etc.), car il est difficile d’obtenir des réponses fiables sur ceux-ci de la part des services administratifs responsables.
|
|
|
The nature and extent of the errors can also vary from one census to the next and jeopardize the comparability of data. The census universe and the collection method may also undergo fairly significant changes that can affect the comparability of data for all or some of the variables. Thus, we have moved on from the traditional method of collection by census enumerators who went from door to door to gather information on each member of the household, to adopt a method of self-enumeration that is more respectful of privacy with the 1971 Census. The new method involved the census enumerator delivering the form to the house, where members of the household, often a single person, would complete it themselves before mailing it. This change in the collection method had a marked effect on responses to the question on ethnic origin; in fact, it prompted a very significant increase in English origin at the expense mainly of Scottish and Irish origins. A second example: since the 1981 Census, the long form, which comprises about 50 questions and is given to one in five households, is no longer used for institutional residents (in hospitals, nursing homes, prisons, etc.), because it is difficult to obtain reliable answers from the administrators in charge. The short form is always used in these cases; it contains less than ten questions and is given to four in five households. Since then, the total data (100%) provides estimates that are always higher than those taken from sample data (20%), which are obviously weighted. Thus, in the 2006 Census, the sample data base estimated the population at 31,241,030 while, according to the total data, the population was 31,612,895, including 371,865 institutional residents (since the 2001 Census, this subpopulation includes the staff and their families who live there), or 1.2% of the total.
|
