Census Undercoverage Survey (CUS)
Summary of changes
Activity on this program started: 1961
For the self-response mode, the paper questionnaires were replaced by an electronic questionnaire. Similarly to what was done in the 2006 cycle, Census questions on Indigenous group and identity were added.
The two shorter questionnaires were eliminated. The wording and placement of some questions was changed to match the wording from the Census questionnaire and improve the flow.
For the Census frame, the province of residence was updated using tax information. Also, most deceased persons were pre-identified and classified as out-of-scope. These changes were made to improve sample efficiency.
Estimation - In 2016, the non-response adjustment methodology was improved. The Census frame calibration step was new, as was the use of the bootstrap method for variance calculation.
Instrument design - Two shorter questionnaires were introduced (one for response by the selected person and one for proxy response). These were used for the non-response adjustment sub-sample.
Sampling - In order to improve the accuracy of the estimates of census population undercoverage, the sample size for the provinces increased from 53,648 persons for the 2001 RRC to 67,813 persons for the 2006 RRC. For the territories, the step of matching the entire health care file to the census before selection was new. The sample for the territories consequently decreased from 5,007 persons to 1,938 persons.
Data sources - A major change to the processing sequence was made for the 2006 RRC. Rather than sending all of the SPs for collection, only those who were not found on the census database during a first search were sent for collection.
Collection instrument - The 2006 Census questions on aboriginal identity were added. Questions on students, workers, and people residing outside the country were dropped.
In 1996, the Private Dwelling Study carried out in 1991 to measure overcoverage was incorporated in the Reverse Record Check.