Second-Stage Sample Allocation
The allocation of the 4,750 buildings for second-stage sampling was done by region (Atlantic, Quebec, Ontario, Prairies, British Columbia) in proportion to the square root of the population of the targeted CMAs and CAs. A few adjustments were made to provide a sample that was representative of all building types (see "Stratification and Sampling in Each EA"). Table A1 gives the anticipated and final second-stage sample sizes by region.
TABLE A1. SECOND-STAGE SAMPLE SIZE BY REGION
Population of targeted CMAs and CAs | Square root of population | Anticipated sample size | Final sample size | |
Region | ||||
Atlantic | 1,000,000 | 1,000 | 534 | 547 |
Quebec | 4,246,000 | 2,061 | 1,100 | 1,083 |
Ontario | 7,353,000 | 2,712 | 1,448 | 1,452 |
Prairies | 2,765,000 | 1,663 | 888 | 1,160 |
British Columbia | 2,136,000 | 1,462 | 780 | 519 |
Total | 17,500,000 | 8,898 | 4,750 | 4,761 |
Allocation was done using population data because they were the only information available for the five regions when sampling began. Ideally, the listing results should have been used, but they were not all known at the time of allocation. In addition, allocation done in proportion to the predicted number of unique addresses seemed too imprecise.
Stratification and Sampling in Each EA
To produce reliable estimates by building size, type and age, stratification must take these three variables into consideration. To do so, a random sample of buildings, stratified based on these three variables, was taken in each of the EAs chosen in the first stage (see Tables A2 through A4).
TABLE A2. LIST OF BUILDING SIZES
Description | |
Size | |
1 | 93 to 464 square metres (1,000 to 4,999 square feet) |
2 | 465 to 929 square metres (5,000 to 9,999 square feet) |
3 | 929 to 4,645 square metres (10,000 to 49,999 square feet) |
4 | 4,645 to 9,290 square metres (50,000 to 99,999 square feet) |
5 | 9,290 to 46,451 square metres (100,000 to 499,999 square feet) |
6 | 46,451 square metres or larger (500,000 square feet or larger) |
Description | |
Code | |
100 | Commercial accommodation |
110 | Entertainment, leisure and recreation |
120 | Building with private, financial or professional offices |
131 | Food retail business |
132-139 | Non-food retail business |
141 | Food services |
142-149 | Non-food services |
151 | Enclosed shopping centre |
152-159 | Strip mall and other shopping centre |
160 | Transportation and maintenance facility |
170 | Wholesale trade and warehouse |
200 | Institutional accommodation |
210 | Administration |
221-226,229 | Education (except colleges and universities) |
227-228 | Colleges and universities |
230 | Hospitals and resident patient care |
240 | Walk-in clinic |
250 | Public meeting |
999 | No classification |
TABLE A4. LIST OF CONSTRUCTION PERIODS
Description | |
Period | |
1 | 1994 and later |
2 | 1993 and earlier |
Depending on the region, certain groupings (by size or type of building) may have been grouped together to limit the number of strata.
Strata containing one, two or three buildings were considered "take-all" units. In the other strata, the sample was divided in proportion to the strata's size, with a minimum of three units per stratum as a constraint.
Atlantic
A final sample of 547 buildings was taken in the Atlantic region using the Generalized Sampling System (GSAM). The following groupings were used to limit the number of strata:
The following strata were merged to offset the already known non-response:
Strata with one, two or three units were considered "take-all" units (three units were taken in all other strata).
Note: The stratum number appears in the form sssssssstttag, where
Quebec
A sample of 1,083 units was taken in Quebec. The following groupings were used:
The following strata were merged to offset the already known non-response:
Strata with one, two or three units were considered "take-all" units (three units were taken in all other strata).
Ontario
The final second-stage sample in Ontario included 1,452 units. The following sizes and types were grouped:
The following strata were merged to offset the already known non-response:
Strata with one, two or three units were considered "take-all" units. In the remaining strata, the sample was divided in proportion to the stratum's size, with a minimum of three units as a constraint.
The following EAs were listed partially (in a partial listing, the size, age and area of the building may not have been recorded): 35021202, 35044710, 35049015, 35063065 and 35087120. In these EAs, if the building type was missing, it was coded 888; similarly, if the age or area was missing, it was coded 8. New strata created were treated like the others for the sample.
Prairies
The initial allocation anticipated 888 units for the Prairies. This number was insufficient, given the constraint of a minimum of three units in each stratum. In addition, the initial allocation anticipated 780 units in British Columbia, which was more than enough. Allocation of the total number of 1,668 units in the Prairies and British Columbia was therefore done jointly for the two regions, so that strata with one, two or three units were considered "take-all" units and the rest of the sample was divided among the remaining strata in proportion to the size of the stratum (with a minimum of three units per stratum as a constraint). The final sample resulted in 1,160 units in the Prairies and 519 units in British Columbia.
The following types were grouped:
The following EAs were partially listed: 46014018, 46014058, 48006255, 48006259 and 48026071.
Stratum 4801215799933 (four units) was a "take-all" unit.
British Columbia
A sample of 519 buildings was selected. The following groupings were used:
EA 59028804 was partially listed.
Strata 5903230899931 (6 units) and 5902906199931 (10 units) were "take-all" units.