The following is a brief summary of the historical comparability of census mobility data, from the 'Place of residence five years ago' question. More detailed information (including references to the 1941 and 1946 Censuses) is available in two user guides and in one technical report: A User's Guide to the 1976 Census Data on Mobility Status, uncatalogued working paper, May 1980; User's Guide to 1986 Census Data on Mobility, November 1990; Mobility and Migration, 1991 Census Technical Report (Catalogue no. 92‑326); and Mobility and Migration, 2001 Census Technical Report (Catalogue no. 92-384-XIE), all of which are available through Statistics Canada.
A. Conceptual changes
The mobility status question on place of residence five years ago has not differed significantly from the five-year questions of previous censuses. Therefore, the mobility data are generally comparable from 1961 on. The question has been based on a five-year reference interval and the census subdivision (CSD) has been used as the migration-defining unit. While the five-year census mobility data are generally comparable from 1961 to 2006, there are some conceptual differences users should be aware of.
- Since 1991, the term 'address' has replaced the term 'dwelling', which had been used in all previous censuses since 1961. The current term 'address' is used in the context of address of usual residence, not mailing address.
- From 1976 on, the primary classification of the population was made on the basis of mobility status (movers, non-movers) while, in some of the earlier censuses, the primary classification was based on migration status (migrants, non-migrants).
- There are also changes in related factors, such as question content, which users should be aware of when analysing mobility data.
Factors affecting conceptual comparability
A number of factors affect historical data comparability of mobility in relation to the conceptual framework. Some of the areas in which changes have occurred are: coverage and universe, question content and structure, and geographic framework.
- Changes in coverage and universe
From 1961 on, the universe for mobility status has included the population 5 years of age and over, with exclusions, which have varied from census to census.
- In 1961, mobility status was reported for the population 5 years of age and over who are members of private households, excluding residents in collective dwellings, temporary residents, overseas military and government personnel and their families, and persons located after the regular census through postal check or re-enumeration. In 1971 and 1976, the universes of population 5 years of age and over excluded Canadian residents stationed abroad in the Armed Forces or in diplomatic services.
- From 1981 on, the mobility universe comprises the population 5 years of age and over residing in Canada, excluding institutional residents and Canadian military and government personnel and their families posted abroad, in households outside Canada. This is in contrast to the 1971 and 1976 data that did include institutional residents.
- In 2006, an additional exclusion from the mobility universe is the employees living in a institutional collective households.
- In 2006, seniors who live in a senior's residence and fill in a long form of the census questionnaires (2B, 2D or 3B) are included in the mobility universe.
- Changes in question content and structure
- From 1961 to 1986, the previous country of residence was not collected for respondents indicating a place of residence outside Canada five years earlier. From 1991 on, respondents who indicated that they had lived outside Canada five years ago were asked to provide the name of the country.
- From 1971 on, internal migrants were asked to specify only the name of their census subdivision of residence 5 years ago, whereas in previous censuses migrants were also asked whether or not their earlier residence was a farm.
- A question on the number of intermunicipal moves was asked only in 1971.
- Since 1986, emphasis was placed on ensuring that Indian reserves were accurately reported in mobility categories. From 1986 on, the answer categories refer to 'city, town, village, township, other municipality or Indian reserve' compared to 'city, town, village, borough or municipality' in 1981 and 'city, town, village, municipality' in 1971 and 1976.
- Instructions in the question referring to write-ins of place names were the same between 1971 and 1976, but they were expanded in 1981 to include examples. The 1981 instruction was repeated in 1986. In 1991, the instruction was revised with new wording and examples. In 1996, the wording and examples in the instruction were again slightly modified. For 2001, they were again slightly modified. There are no changes for the 2006 Census.
- In 1991, revisions were made to both the structure and wording of the 'Place of residence five years ago' question. In addition to the rewording of instructions and the replacement of the term 'dwelling' with 'address', a filter question was introduced to serve as a screen for movers and non-movers. As well, answer categories were reworded and shortened. With these revisions, the basic content is still the same as the five-year questions of previous censuses; in general, historical comparability is retained.
- In 1996, although the filter question was eliminated, the concept remained the same.
- In 1996, the write-in box for county was eliminated. However, at the time of automated coding, the code for county was provided.
- In 2006, postal code write-in boxes were included in the question.
- Changes in geographic framework
- The comparability of the mobility data over the censuses has been affected by both conceptual changes in geography (such as the definitions of rural area [RA], urban area [UA], farm, non-farm, census metropolitan area [CMA] and census agglomeration [CA]) and the changes in census subdivision (CSD), census division (CD), census metropolitan area (CMA) and census agglomeration (CA) boundaries. Because the number of census geographic areas (CSDs, CMAs, etc.) and their boundaries change from census to census, the user must exercise caution when using mobility data over two or more censuses. For example, in 1986 there were 6,009 CSDs, 114 CAs and 25 CMAs compared to 5,710 CSDs, 88 CAs and 24 CMAs in 1981. The number of CSDs was much smaller in 2001, reduced to 5,593, and to 5,418 in 2006, because of the amalgamation of cities. The changing number and boundaries of CSDs from one census to another will, to some extent, affect the comparability of the measure of 'migrants' across censuses (since the volume of migrants is partly a function of the number and size of CSDs). The number of CMAs and CAs increased in 2006 to 33 and 111, respectively. Details of changes affecting the historical comparability of census geography from 1961 to 2006, as well as definitions and descriptions of available maps, are covered in a variety of census products.
- Because of changes in geographic areas between censuses, places of residence five years ago must reflect boundaries of the census in question in order to obtain geographic consistency between current and previous place of residence. For example, when tabulating 2006 data on usual place of residence five years ago by current place of residence, all areas reflect 2006 boundaries, even when referred to as places of residence in 2001.
B. Collection and processing changes
The changes over censuses associated with each of the collection and processing stages have not significantly affected the comparability of mobility and migration data. However, there are some changes in processing that the user should be aware of when analysing mobility data.
- In 1991, autocoding (computerized coding) was introduced for converting write-ins of place names in the mobility question to the Standard Geographical Classification (SGC) codes. In previous censuses, write-ins were coded manually. Some of the manual coding procedures used in 1986 for resolving duplicate place names (e.g., Kingston township vs Kingston city, both in Ontario, but only 'Kingston' reported) were automated during Edit and Imputation (E & I) as an extension of the autocoding system. The increased accuracy obtained with autocoding is expected to improve the quality of mobility data on out-migrants from CSDs, compared with previous censuses. In 2006, postal code information was also used in coding. The evaluation of the use of postal code on the data quality will be done later.
- A significant change in E & I from earlier censuses occurred in 1981. Prior to 1981, non-response (partial/total) to the question on previous place of residence was reported as 'Not stated'. However, for 1981, this 'Not stated' category was dropped. Non-response to the question on previous place of residence was changed to a specific response via a combination of deterministic, family and hot-deck imputation assignments. This imputation was achieved using the SPIDER program, which was introduced in 1981. For the 2001 Census, the Imputation portion of the E & I processing is done using the CANadian Census Edit and Imputation System (CANCEIS). The Edit portion is still being done using the SPIDER program. For the 2006 Census, both Imputation and Edit portions were done using the CANCEIS.
C. Place of residence 1 year ago
The question on the place of residence 1 year ago was asked in the 1991 Census for the first time. At this time, the migration-defining boundary was the province or the territory. From the 1996 Census on, the migration-defining boundary was changed to the census subdivision as in the case of the question on the place of residence 5 years ago. The processing of the 1 year ago data, such as the coding and the edit and imputation, was performed the same way as for the 5 year ago data.