Defining and operationalizing our urban-geographic variables

---

Our principal SPSS syntax files constructing custom urban geographies for use with the 1921, 1931, 1941 and 1951 CCRI microdata are available for download here (they are .TXT documents that can be renamed .SPS for opening in SPSS) :

1. syntax SPSS – CITY variable.txt
2. syntax SPSS – MEGACMA (metro Montreal & Toronto by zones) variable.txt
3. syntax SPSS – SIZECLASSDYNAMIC variable.txt

Major cities

We wanted to conduct some city-specific analyses of the CCRI microdata. To do this, we had to work within the limitations imposed by Statistics Canada’s disclosure rules. These disallow the release of any data identifying a geographic entity with a population of less than 25,000 (according to CCRI weighted estimates). Throughout the twentieth century, the Dominion Bureau of Statistics/Statistics Canada generally used 30,000 as the lower population limit of a major urban centre. We decided to use 50,000 as the lower limit of a “major city” for the purposes of our study, considering that the disclosure rules also prohibit the release of frequency tables containing cells with small cell sizes (population of less than 5 in the unweighted count).

In the first instance, we identified, for each census year in our study (1921, 1931, 1941 and 1951) all census subdivisions in the CCRI database that were designated as a “City” in the census and that had a population of 50,000 or more. However, over this period, the major cities identified in this first step were evolving into more complex urban areas in which modern transportation technologies linked the core city with “satellite communities” or new residential suburbs that were closely tied to it in economic terms (labour markets, commuting…). Except for some of the large prairie cities (Calgary, Edmonton, Regina, Saskatoon), these outlying suburbs and satellite communities were not within the administrative boundaries of the core city.

Dominion Bureau of Statistics analysts acknowledged in the 1930s that new methods were needed for defining these more complex urban areas, and they began to promote the concept of “Greater” or “Metropolitan City” (Cudmore and Caldwell 1938). While we take this for-granted today, it was an innovation at the time! It was implemented in the 1941 census which defines eight Greater Cities (population 100,000 and over for the city proper i.e. the administrative unit), forming part of a longer list of 12 Metropolitan Areas (the eight Greater Cities plus four more Metropolitan Areas (population 50,000 and over for the city proper). They also defined them retroactively for 1931 (Dominion Bureau of Statistics, [1948] and 1950, Ch. II). In 1951, a standardized vocabulary and method of defining “census metropolitan areas”, was established (Ross 1984).

We decided that this more inclusive concept of major cities – often comprising an agglomeration of the core and adjacent municipalities and sometimes including districts classified as rural – would be more meaningful for our research project than simply using the city proper (i.e. administrative unit). We thus opted to apply the Greater City/Metropolitan Area census concept, as far as possible, to the CCRI database. Implementing this innovative approach was a major methodological undertaking (for more detail see See also: Appréhender l’urbain dans l’ICRS, Powerpoint presentation to QICSS Summer School, 2015 (in French / PDF). ) generating original research infrastructure for historical research on Canadian urban geographies. We are sharing the SPSS syntax files for the CITY variable that we created to apply these urban geographies to the CCRI datasets for 1921, 1931, 1941 and 1951, so that they can be used for future research projects.

For 1921, 1931 and 1941, we created aggregates of CSDs to resemble as closely as possible the composition of the Greater Cities/Metropolitan Areas. For 1951, we created aggregates of CSDs to approximate the newly defined Census Metropolitan Areas.

For 1931 and 1941, the definitions of eight Greater cities (Vancouver, Winnipeg, Windsor, Hamilton, Toronto, Ottawa, Montréal and Québec) are given in Dominion Bureau of Statistics. 1950. Eighth Census of Canada, 1941. Volume I: General Review and Summary Tables. Table 8, and the same table also gives definitions for 1931. The four additional metropolitan areas, whose core city has a population of 50,000-99,000 (Victoria, London[iii], St. John and Halifax) are defined in Dominion Bureau of Statistics. ([1948]). Population of the ‘Greater’ Cities by Sex, Age, Racial Origin, Religion, Etc. Eighth Census of Canada, 1941. Population [Bulletin] No. A-13. As far as possible, we extrapolated the 1931 definitions back to 1921, which may have resulted in some then-rural areas being included in our metropolitan city geographies. For 1951, when the concept of “Census Metropolitan Area” was standardized and applied consistently, we used a published census table (Dominion Bureau of Statistics. 1953. Ninth Census of Canada, 1951. Volume 1: Population. General Characteristics. Table 12).

We could not exactly recreate these DBS definitions, however, because they often include only certain parts (the urbanized parts) of CSDs located at the urban-rural fringe whereas the CCSD did not provide geographic coding below the CSD level. Consequently, we had to decide whether or not to include the entire subdivision including its rural parts. The CMA descriptions and maps in a review by Statistics Canada of the evolution of the concept from 1941 to 1981 (Ross 1984) were invaluable in helping us make these judgment calls.[iv]

Urban/rural distinctions

Throughout the twentieth century, and today, Canadian census dissemination practice has insisted on the importance of distinguishing urban from rural populations. However, the threshold for an urban area has varied between, and even within, censuses. The predominant practice from 1871 to 1941 was to designate as “urban” places that were cities, towns, villages and parishes incorporated under provincial laws, regardless of size. However, as early as the 1931 census, DBS analysts began to recommend setting a minimum population threshold of 1000 for an incorporated place to be categorized as “urban”, arguing that smaller places, even if incorporated, were more akin to rural communities in terms of their economies and population trends (Cudmore and Caldwell, 1938: 9). This became the official census practice in 1951 (Statistics Canada 1984).[i] A third practice, initiated with the 1931 census on account of the increasing proportion of rural dwellers with non-agricultural occupations (e.g. extractive industries or providing services to urban residents at the fringe of nearby cities, in the context of the rapid growth of larger urban areas), was to distinguish the “non-farm” and “farm” populations of rural areas.

We opted for a variant of the 1,000-population threshold criterion, as Baskerville (2001) had done for analysis of 1901 census microdata. We coded as “urban” all census subdivisions that the CCRI team had identified as a city, town or village in the variable CSD_TYPE (based on information transcribed from census records) AND whose population was 1000 and over. (The exception, of course, is that if a town or village of less than 1000 inhabitants fell within a metropolitan area, as defined by the procedure explained above, we counted it as urban). Our decision was motivated by the desire for a consistent definition over time and one that would reduce the impacts of provincial variations in the criteria for the population threshold for incorporation.[ii]

Urban size class

Our next step was to create a variable that would allow analyses according to the size of the urban area. Larger cities are likely to be more diversified in terms of their employment structure and housing stock, and thus to offer a wider range of choice of living arrangements for the unmarried individuals who are the focus of our research. Taking into account the size distribution of Canadian urban areas during our study period of 1921 to 1951, we opted for the following distinctions:

• Metropolises of half a million or more inhabitants
• Other very large cities (100,000 – 499,000)
• Other major cities (50,000 – 99,999)
• Mid-sized cities (10,000 – 49,999)
• Small urban centres (1,000 – 9,999)

We created two versions of this variable: SIZE_CLASS_DYNAMIC, in which the size class value is specific to the census year; and SIZE_CLASS_CONSTANT, in which the city is coded according its size in 1941.

Syntax file for size_class_dynamic

Intra-metropolitan geographies of Montréal and Toronto

Finally, for Canada’s two metropolises, Greater Montréal and Greater Toronto, we created a simple three-fold intra-urban geographic classification so as to compare unmarried people living in the suburbs with those in central areas. We categorized each census subdivision (CSD) of these metropolitan areas into one of the following zones for each census year (1921, 1931, 1941 and 1951):

• Central core
• Inner suburbs
• Urban fringe.

To establish the limits of each zone for each census year, we took into account population densities and public transportation connections to the downtown area. We wish to thank Professor Richard Harris (Geography, McMaster University) for his major contribution to creating the Toronto zones.

See map examples, Intra-metropolitan geographies for Toronto, 1921, 1931, 1941 and 1951

See map examples, Intra-metropolitan geographies for Montréal, 1921, 1931, 1941 and 1951

References cited

Baskerville, Peter. 2001. “Familiar Strangers: Urban Families with Boarders, Canada, 1901.”  Social Science History 25 (3):321-346.

Cudmore, S.A., and H.G. Caldwell. 1938. Rural and urban composition of the Canadian population, Census monographs. Ottawa: Dominion Bureau of Statistics.

Dominion Bureau of Statistics. 1950. Eighth Census of Canada, 1941. Volume 1: General Review and Summary Tables. Ottawa: Dominion Bureau of Statistics.

Dominion Bureau of Statistics. [1948]. Population of the ‘Greater’ Cities by Sex, Age, Racial Origin, Religion, Etc. Eighth Census of Canada, 1941. Population [Bulletin] No. A-13. Ottawa: Dominion Bureau of Statistics.

Moldofsky, Byron. 2013. “Exploring historical geography using census microdata: The Canadian Century Research Infrastructure (CCRI) project.” In Historical GIS research in Canada, edited by Marcel Fortin and Jennifer Bonnell, 271-286. Calgary: University of Calgary Press.

Moldofsky, Byron. 2014. “The CCRI geographical files: introduction and examples.” In The Dawn of Canada’s Century: Hidden Histories, edited by Gordon Darroch, 19-48. Montreal & Kingston: McGill-Queen’s University Press.

Ross, Grafton. 1984. Census Metropolitan Area/Census Agglomeration Program. A Review, 1941-1981. Ottawa: Statistics Canada, Geography Division (Working Paper No. 8 – GEO 84).

Statistics Canada. 1984. Urban Growth in Canada. Census monographs, 1981 Census of Canada. Ottawa: Statistics Canada.

St-Hilaire, Marc, Byron Moldofsky, Laurent Richard, and Mariange Beaudry. 2007. “Geocoding and mapping historical census data: the geographical component of the Canadian Century Research Infrastructure.” Historical Methods: A Journal of Quantitative and Interdisciplinary History 40 (2):76-91. doi: 10.3200/hmts.40.2.76-91.

---

[i] The criterion of municipal status was scrapped in 1971 and the definition has since evolved to take account of population density (Statistics Canada, 1984; Statistics Canada 2012).

[ii] In contrast, the CCRI team chose the first option when creating the CCRI_URBAN_RURAL variable in the microdata files; all CSDs which the DBS had labelled as City, Town, Village or Parish are coded as urban by this CCRI variable and all CSDs where the DBS left this field blank are coded rural (Moldofksy 2013 and 2014).

[i] Prior to 1921, the terminology used in publications reporting census findings was “sub-district”, a term drawn from the electoral geography used for census-taking. The terms “census division (CD)” and “census subdivision” (CSD), referring to municipal administrative entities, not necessarily coinciding with electoral unit boundaries, were established for census publications beginning in 1921, and the term “sub-district” disappeared from 1931 on (Moldofsky 2013: 275 and note 19).

[ii] To do such aggregation, it is first necessary to identify the code corresponding to the required CSD for each census year. The procedures to follow are given in the following three sections of the CCRI online User Manual (read them in this order!):

https://ccri.library.ualberta.ca/enoverview/geography/index.html
https://ccri.library.ualberta.ca/endatabase/geography/ccriud/index.html
https://ccri.library.ualberta.ca/endatabase/geography/digitizedpublictables/index.html

[iii] See Note 3.

[iv] For instance, Ross (1984) indicates that in 1941 the Greater City of London (Ontario) only included small parts of two predominantly rural townships. Consequently for 1941 we opted not to include them in our definition of the major city of London.