Problems in Census Taking

Abstract

While this is too early a date to produce for discussion a tentative questionnaire for the 1951 Census, with all the necessary definitions, we, in the Dominion Bureau of Statistics, are greatly pleased to have the opportunity of placing before you some of the problems that we will have to solve before our army of enumerators begin their house to house canvass to gather the information. Along with the discussion of these problems, I would like to mention some of the improvements that we propose to make in the 1951 Census as compared with previous censuses.The taking of a census is a very important and costly operation. The Advisory Committee on the United States Census stated in 1921: “Of all the peace-time activities of the federal government, taking and compiling the census is the largest.” Because of its high cost, it is necessary to make it yield its maximum both in quality and quantity of returns. The problem of determining the optimum quantity, keeping in mind that quantity very often works against quality, is possibly one of the most difficult ones to solve. Just before the census is taken we receive suggestions from a very large number of persons and groups for additional questions to be placed on questionnaires. These persons reason quite correctly that the most costly part of our organization is the one which makes it possible to cover every nook and cranny of this vast country of ours; and they also reason, not quite so correctly, that since our enumerators visit every home anyway, there is no harm is asking a few more questions. Mind you, the great majority of these questions are worthwhile and would produce information which is not available anywhere else. We are, therefore, placed in the unhappy position of having to appear to be unco-operative or to load our questionnaires beyond the point where it is possible to obtain reliable information. The first point, then, is to determine how long census questionnaires should be to provide as much information as possible without placing the quality of the statistics obtained in jeopardy.