PDF Version | Table of Contents | Next Page
This section summarizes the methodology used in the Canadian Vehicle Survey (CVS), conducted by Statistics Canada on behalf of Transport Canada and Natural Resources Canada (NRCan) in 2008. More information is available in the Canadian Vehicle Survey: Annual 2008, produced by the Transport Division of Statistics Canada.12
The CVS is a voluntary survey of vehicles that is conducted quarterly. The survey design allows for calculation of annual estimates based on the data collected during the four quarters. The survey population consists of all motor vehicles registered in Canada at any time in 2008 that have not been scrapped or salvaged. Buses (since 2004), motorcycles, off-road vehicles (e.g. snowmobiles) and special equipment (e.g. cranes, snowploughs) are excluded from the registration lists used in the sample.
The survey population is derived from the vehicle registration lists sent by the 10 provincial and three territorial governments to Statistics Canada three months before the reference period. This population differs slightly from the population of interest because vehicles that were registered during the quarter or fewer than three months before the quarter began are not included in that quarter’s sample (the sample for each quarter is derived from the population of the preceding quarter).
The registration lists received by Statistics Canada undergo a rigorous preparation procedure:
The most recent set of prepared lists is used to select the sample for each quarter. These sets of vehicle lists and the days within the respective quarter constitute the survey population.
The CVS uses a two-stage sample design. A sample of vehicles is selected in the first stage, and a sample of consecutive days within the quarter is selected in the second stage. All vehicles from the survey population are stratified into 78 strata according to vehicle type, jurisdiction and vehicle age. Then a systematic sample of vehicles (first-stage sample) is selected from the survey population to spread the sample over all regions.
In the second stage, a first reporting day within the quarter is randomly assigned to each vehicle that was selected in the first stage. Within each stratum, the first reporting day is spread evenly over the quarter to ensure a uniform number of responses over time and for each day of the week. This step is not applied to the vehicles registered in the three territories because only odometer readings are collected.13
The sample consisted of 33 555 vehicles for the four quarters of 2008, with 21 495 vehicles from the provinces and 12 060 from the territories.14 Table B-1 shows the number of vehicles sampled in the provinces and territories in 2008 by type of vehicle.
Table B-1 Vehicles in the sample by jurisdiction and vehicle type
Jurisdiction | Vehicles | Total | ||
---|---|---|---|---|
Light vehicles | Medium trucks | Heavy trucks | ||
Newfoundland and Labrador | 877 | 204 | 221 | 1 302 |
Prince Edward Island | 544 | 179 | 146 | 869 |
Nova Scotia | 1 072 | 264 | 272 | 1 608 |
New Brunswick | 1 067 | 224 | 270 | 1 561 |
Quebec | 2 277 | 466 | 539 | 3 282 |
Ontario | 2 554 | 654 | 621 | 3 829 |
Manitoba | 1 112 | 333 | 293 | 1 738 |
Saskatchewan | 1 049 | 363 | 401 | 1 813 |
Alberta | 1 626 | 536 | 596 | 2 758 |
British Columbia | 1 755 | 340 | 640 | 2 735 |
Total for provinces | 13 933 | 3 563 | 3 999 | 21 495 |
Yukon | 1 580 | 784 | 1 571 | 3 935 |
Northwest Territories | 3 348 | 1 041 | 789 | 5 178 |
Nunavut | 2 549 | 160 | 238 | 2 947 |
Total for territories | 7 477 | 1 985 | 2 598 | 12 060 |
Total for Canada | 21 410 | 5 548 | 6 597 | 33 555 |
Data collection for the vehicles sampled is conducted differently in the provinces than in the territories. In the provinces, the registered owners of the sampled vehicles are contacted for a Computer-Assisted Telephone Interview (CATI). During the CATI, the following information is collected about each sampled vehicle:
Respondents are asked to complete a trip log. If they agree, the trip log is mailed to them. There are two types of logs: one for light vehicles and one for medium and heavy trucks. Respondents receiving a light-vehicle log are requested to record information for 20 consecutive trips made in the selected vehicle, beginning on the assigned first reporting day. Respondents have to record a new trip each time the driver enters the vehicle or a passenger enters or exits the vehicle.15
Respondents receiving a heavy-vehicle log (medium and heavy trucks) are requested to record information for all the trips made in the selected vehicle over the assigned seven days. A new trip begins if there is a stop that lasts more than 30 minutes, if the driver changes, if the reason for the trip or the use of the vehicle changes, if the truck configuration is modified or if the truck goes from full to empty or the reverse.
The following information is recorded for each trip:
Since 2004, when NRCan became co-sponsor of the CVS, respondents have been asked to continue recording fuel purchases until they reported two fill-ups or five purchases or until the 28-day reporting period ended.
Less information is collected in the territories. Statistics Canada sends a questionnaire that asks for an odometer reading at the beginning of the quarter and another at the end, so the distance travelled during the quarter can be identified. Information is also collected on the vehicle’s ownership status (still owned, sold or scrapped), body style and type of fuel used.
After all the information for the survey has been 25 collected, Statistics Canada conducts a series of computerized and manual verifications to ensure that the records are consistent and that there are no errors as a result of data capture.
Missing values and data found to be in error are imputed by another automated system using different imputation rules depending on the vehicle, available information and type of data to be imputed. For example, data can be imputed based on responses to other questions or by using data from similar vehicles. The imputed data are examined again for completeness and consistency.
Statistics Canada defines the CVS response rate as the number of vehicles for which the respondents have provided full or partial answers to the questions concerning vehicle-kilometres only, divided by the number of vehicles in the sample. Table B-2a and
Table B-2b show the response rates obtained for each quarter by vehicle type.
Table B-2a CVS response rate – Provinces
2008 (Quarter) |
CVS response rate – Provinces (percent) | ||
---|---|---|---|
Light vehicles | Medium trucks | Heavy trucks | |
Quarter 1 | 58.2 | 60.7 | 62.4 |
Quarter 2 | 59.1 | 60.3 | 57.9 |
Quarter 3 | 58.2 | 62.3 | 62.7 |
Quarter 4 | 60.9 | 65.7 | 66.8 |
Annual | 59.1 | 62.3 | 62.4 |
Table B-2b CVS response rate – Territories
2008 (Quarter) |
CVS response rate – Territories (percent) | ||
---|---|---|---|
Light vehicles | Medium trucks | Heavy trucks | |
Quarter 1 | 14.6 | 10.8 | 12.1 |
Quarter 2 | 16.9 | 11.2 | 9.7 |
Quarter 3 | 15.9 | 10.9 | 13.0 |
Quarter 4 | 12.9 | 10.4 | 9.1 |
Annual | 15.1 | 10.8 | 10.9 |
The response rate for the fuel component of the CVS is lower than the response rates in the preceding tables. Therefore, the data on fuel consumption has a high imputation rate, which helps explain the lower quality of fuel consumption estimates in this report.
Estimates are based on the principle that each vehicle in the sample represents a certain number of vehicles in the population of interest. A sample weight is assigned to each vehicle in the sample, and the purpose of the final set of weights is to reflect as closely as possible the characteristics of the vehicle population during the reference period.
All estimates for 2008 presented in this report were produced by using an estimate module developed by Statistics Canada. This module also calculates the coefficient of variation (CV) that reflects the quality of each estimate. The CV takes into account variability due to sampling and variability due to non-response and imputation.
For example, a variance due to relatively high imputation has a negative effect on the quality of fuel consumption estimates. Estimates with a CV of more than 35 percent are not reliable enough to be published. Table A-1 in Annex A describes the indicators used in this report to describe the quality of estimates.
For more information on the methodology used in the CVS, contact the Transport Division, Statistics Canada, at
Transport Division12 Statistics Canada, 2009, Canadian Vehicle Survey: Annual 2008, Catalogue No. 53-223-X, www.statcan.ca/bsolc/english/bsolc?catno=53-223-X.
13 Less information is collected in the territories because respondents there are asked to participate in several surveys a year.
14 A larger sample in the territories enables Statistics Canada to compensate for a lower response rate in these jurisdictions.
15 This definition has been used as of the first quarter of 2004 and is different from that used in previous versions of the CVS.