Appendix C: Data Documentation and Sources for Maps,Tables, and Figures

Planning for Prosperity: Globalization, Competitiveness and the Growth Plan for the Greater Golden Horseshoe relies on data from Statistics Canada to analyze the dynamics of economic change in the Greater Golden Horseshoe.

Employment statistics for Ontario and Canada were retrieved from the CanSIM database.[1] A custom table derived from the Labour Force Survey[2] representing the nine Census Metropolitan Areas (CMAs) found in the Greater Golden Horseshoe (GGH) is used to report on region-wide statistics.

All mapping and employment numbers for specific geographies in the GGH were based on customized data tables that were created from the 2001 and 2006 long-form Census and 2011 National Household Survey (NHS) from data on the Employed Labour Force 15 years and older Having a Usual Place of Work, or Place of Work (POW) for short. These data exclude people who work from home, those who work outside Canada, and those who have no fixed workplace address (such as certain tradespeople or freelancers who work from their clients' locations). The breakdown for 2011 is shown below:

Place of Work Status

Total GGH

Total jobs

4,304,590

Worked at home

287,400

Worked outside Canada

19,150

No fixed workplace address

465,665

Usual place of work

3,532,370

The NHS has been criticized for its inadequate coverage of certain groups and certain less-populated areas in Canada. The Chief Statistician for Statistics Canada[3] recently weighed in on warranted and unwarranted criticism of the NHS. Nevertheless, in the absence of alternative sources of employment data with the necessary regional coverage, we have used NHS to map employment patterns for 2011, compared the mapped variables with those in the 2001 and 2006 censuses, and compared data totals based on different geographic census units and different aggregations of detailed industry and occupation codes to best understand how the NHS-based POW data compares to the previous censuses (see the section Comparing Three Censuses for discussion on how census totals compare).

Data on Place of Work were used to map jobs (workplaces of workers) by industry and occupation based on the North American Industry Classification System (NAICS) and the National Occupation Classification (NOC). NAICS uses a five level hierarchy of 2- to 6-digit codes that break down industrial sectors, while the NOC system is made up of a hierarchy of 2- to 4-digit codes that break down occupations into detailed categorizations. Ultimately, 2-, 3-, and 4-digit NAIC codes were combined to create 10 "planning categories." This was done to create meaningful industry categories that relate economic structure and change to land use. (See Table A1.)

The NOC codes were combined to map occupations in the STEM grouping. The NAICS and NOC systems have changed between 2001 and 2011, so we had to adjust our categorization to account for these changes. The changes are noted in the sources for each map.

This appendix contains a list of the NAICS or NOC codes that were used for each map.

We have also drawn on commuting data from the Transportation Tomorrow Survey (2011). The TTS does not cover the entire Greater Golden Horseshoe (it omits Haldimand and Wellington counties and the northern part of Peterborough County).

Comparing three censuses: 2001, 2006, and 2011

Since census geographies (i.e., Dissemination Areas (DAs), Census Tracts (CTs),

Census Subdivisions (CSDs), etc.) change over time, a common geography was needed to compare the three years. For this study, the standard 2011 CTs and CSDs were used. In order to join the 2001 and 2006 census data with the 2011 geography, the 2001 and 2006 geography IDs needed to be customized by Statistics Canada. Essentially, this allowed the custom 2001 and 2006 geography IDs to correspond with the 2011 geography. It should be noted that the process of obtaining custom geographies increases the suppression threshold for all years.

Suppression and loss of data

Suppression rules are applied to census products to ensure the confidentiality of respondents' information. Typically, suppression is applied to areas in which the population is less than 40 or 100, depending on the year of the census. More specifically, in using a standard geography (2011) the suppression threshold is 40. However, since the 2001 and 2006 geographies are customized, the suppression threshold is increased to 100.

Suppression also occurs when the data is explored in finer detail. For example, in examining NAICS data, higher levels such as the total and 2-digit codes have less suppression, but in the more detailed 3- and 4-digit codes suppression is more evident. This issue was more predominant in the 2011 data - possibly owing to changes in how the NHS survey was conducted. Table A2 compares employment data totals (total number of jobs based on POW data) for all three years based on aggregations of different levels of geography (CSD, CT, DA) and different categorizations of the NAICS codes, for example NAICS 2-digit codes versus planning categorization codes which is based on 2-, 3- and 4-digit NAICS codes. Although for all three years, 2001, 2006, and 2011 there is an absolute and percentage difference in totals based on aggregations at different levels of geography and different levels of detailed categorization, the difference is greatest in the 2011 NHS.

Random rounding

In addition to suppression as a means of ensuring confidentiality of respondents' information, random rounding is also used. All census data are subject to a random rounding algorithm, which rounds raw counts to end in either 0 or 5. This process is carried out based on a predetermined frequency. However, raw counts that already end in 0 or 5 are excluded and therefore remain the same.