Consumption of Energy Survey for Universities, Colleges and Hospitals, 2003

PDF Version | Table of Contents | Next Page

Appendix A

Scope of Survey and Methodology

Target population and sample design

The Consumption of Energy Survey (CES) constitutes a census survey of university, college and hospital campuses or complexes in Canada. A questionnaire was sent to each of the campuses or complexes in order to obtain data on the consumption of electricity, natural gas, heavy fuel oil, other middle distillates, propane, steam and wood in each sector. A copy of the questionnaire can be obtained from Statistics Canada.

The questionnaire also asked for data on the floor area of campuses or complexes, as well as the number of hospital beds or students enrolled at the universities and colleges. These data are used to calculate energy intensity indices for each of the sectors covered by the survey. In this analysis, gross energy intensity is used. For example, the gross energy intensity for a region in a specific sector is defined as the total energy consumption of all campuses or complexes of the sector, divided by the total gross floor area of the campuses or complexes consuming energy. This report does not consider average energy intensity, i.e. the average of the intensities of all campuses or complexes, which can be used to compare individual campuses or complexes.

The survey population was extracted from comprehensive lists developed or used by Statistics Canada. A questionnaire was sent in January 2004 to 1098 campuses and complexes. The initial survey frame comprised 126 university campuses belonging to 111 institutions, 230 college campuses belonging to 178 institutions, and 742 hospital complexes. There were 763 respondents, 325 refusals, non-responses or partial responses, and 61 out-of-scope units. The response rate was 70 percent. The data were therefore estimated from 95 university campuses belonging to 86 institutions, 168 college campuses belonging to 138 institutions, and 500 hospital complexes.

Validation of data and imputation

Control rules ensured the validity and internal consistency of responses. The data were reviewed manually. No data were imputed. An adjustment was made for non-response (refusals and unable to locate) and responses that could not be used.

To that end, each campus and complex was attributed a weighting coefficient indicating how many campuses or complexes of the population are represented by each statistical unit. The CES is a census: each unit has an initial weight of one (1). An adjustment of the weighting was applied to compensate for unresolved responses. After removing institutions that no longer operated and weighting the respondents' data, this survey covers 123 university campuses, 228 college campuses and 729 hospital complexes.

Data sharing and confidentiality

Regional blocks ensured the quality of the data and preserved confidentiality. The Atlantic region block is Newfoundland and Labrador, Prince Edward Island, Nova Scotia and New Brunswick. The Prairies block is Manitoba, Saskatchewan and Alberta. The Canadian Territories – Yukon, Northwest Territories and Nunavut – were grouped with British Columbia for the survey.

Certain measures ensured that the estimates based on CES data are reliable enough for publication. The letters used in the tables indicate the degree of sampling error, represented by a coefficient of variation of the estimate. The letter "A" indicates that the estimate has a very low coefficient of variation; the letter "B" indicates a somewhat higher coefficient of variation, and so on. Estimates graded "A" or "B" are considered accurate enough for most uses. Data whose coefficient of variation is higher – graded "C" or "D" – are accurate enough for a few uses, but should be used with caution. The letter "F" means that the coefficient of variation of the estimate exceeds 50 percent. These estimates are not published because they may present too great a sampling error.

Previous Page | Table of Contents | Next Page