Producing official statistics via voluntary surveys -- the National Household Survey in Canada

Abstract

Statistics Canada conducts over 350 business, social and institutional surveys a year. Of all social or household type surveys, only one in addition to the Census of Population is conducted on a mandatory basis, the Labour Force Survey. By their very nature, voluntary surveys will achieve lower rates of response and are thus exposed to higher risks of bias. For the 2011 Census of Population program, the long form census was for the first time collected on a voluntary basis as the National Household Survey. The survey content was basically the same as that of previous Census long forms and covered various socio-demographic topics that are of high importance to a wide variety of stakeholders in Canada. Given that one of the key characteristics of a census is to produce data for small regions and for subgroups of the population, collecting the survey on a voluntary basis introduced several challenges. Statistics Canada, based on its extensive experience with voluntary surveys, developed a number of processes and approaches to ensure the highest data quality possible. This paper describes what these measures were for data collection, data processing and estimation. It also provides a brief description of the quality assurance processes underlying the release strategy of the 2011 survey.