Commercial and Institutional Consumption of Energy Survey

Detailed information for 2004





Record number:


The purpose of this survey is to collect detailed information on the energy demand and consumption patterns of Canadian businesses, institutions and organizations.

Data release - February 22, 2006


The purpose of this survey is to obtain information on the demand for energy in Canada. This survey will provide updated statistical information on the energy consumption patterns of Canadian businesses, institutions and organizations. This sector is a key focus of Natural Resources Canada's programming and analysis of energy efficiency. These data provide much needed insight into the patterns of energy consumption across this sector and will give Natural Resources Canada the ability to develop or refine its programs to promote energy efficiency in Canada.

Reference period: Calendar year


  • Energy
  • Energy consumption and disposition

Data sources and methodology

Target population

The target population for this survey is businesses and institutions -- at the establishment level -- within the NAICS codes outlined in the additional information link below, with a minimum of 1 employee and a business location that is not a residence. As with the previous survey, small home-based businesses and businesses with no employees (such as independent consultants or researchers) will be out of scope. Unless otherwise indicated, the Business Register was the survey frame for each industry.

Instrument design

The questionnaire was initially designed for Manufacturing, Construction and Energy Division's Industrial Consumption of Energy Survey. Its use for this population was tested via a pilot survey in 2002. Revisions to the 2003 questionnaire were based upon the pilot survey results, as well as input from the Questionnaire Design Resource Centre. There were 3 versions of the 2004 survey questionnaire, one for commercial businesses which included a question on the number of employees, one for hospitals which included a question on the number of beds and one for education institutions which included a question on the number of students.


This is a sample survey with a cross-sectional design.

The survey frame was created from three non-overlapping lists. The first list is the frame used for the 2003 Consumption of Energy Survey, which covered colleges and CÉGEPs, universities and hospitals. Hospitals were drawn from the Business Register using NAICS codes 622111 General (except paediatric) Hospitals; 622210 Psychiatric and Substance Abuse Hospitals; and, 622112 Paediatric Hospitals. "Health boards" and "associations" were removed from the list as these are not hospitals per se, but administrative bodies. Also removed were "research institutes" that are already covered by in-scope hospitals. "Non-hospital" units were minimized by applying an employee threshold of 50 or more employees. The university population was taken from a list provided by Public Institutions Division. Here, NAICS code 611310 Universities is considered in-scope. This list is campus-based, which is the unit of measurement for universities. The college population used NAICS code 611210 Community Colleges and C.E.G.E.P.s from the Business Register. Colleges were also treated on a campus basis, similar to universities. Colleges with fewer than 20 employees were excluded.

The second list comes from the Statistics Canada's Business Register and includes NAICS categories 41 Wholesale Trade, 44 and 45 Retail Trade, 49 Warehousing, 51 Information and Cultural Industries, 52 Finance and Insurance, 53 Real Estate and Rental Leasing, and 54 Professional, Scientific and Technical Services, 621 Ambulatory Health Care Services, 623 Nursing and Residential Care Facilities, 624 Social Assistance, 71 Arts, Entertainment and Recreation, 721 Accommodation Services, 722 Food Services and Drinking Places, 81 Other Services (except Public Administration, and 91 Public Administration.

The third list is the frame of elementary and secondary schools maintained by the Culture, Tourism and Centre for Education Division of Statistics Canada, excluding the following school types: Aboriginal or Band Operated School, Hospital School, Prison / Jail / Detention Centre / Offender Facility School, Distance Education (Virtual / Correspondence School), Home School, and Nursery, Daycare Centre, E.C.S. (Early Childhood Services).

The sampling unit was the establishment with the exception of universities and colleges (where the unit is the campus as noted above) and churches (NAICS 813110) where the location level is used.

A stratified sample design was used with stratification based on the region for all industries, and on the number of employees for all industries except universities, colleges and CÉGEPs, hospitals, and elementary and secondary schools (where the number of employees was not available on the frame). The stratification keys were the regions Atlantic, Québec, Ontario, Prairie, and British Columbia, and the employee size categories 1 to 20 employees, 21 to 250, and over 250 employees.

The sample was selected so as to generate estimates of equal quality (variance) for all strata. The sample size for the 2004 survey was 7349 units.

Data sources

Responding to this survey is voluntary.

Data are collected directly from survey respondents.

These data were collected via a paper questionnaire using a mail-out, mail-back approach. Completed questionnaires are captured via an internally developed MS Access application. For the 2004 survey, various elements of the 2003 survey frame, such as contact person, telephone number and mailing address were updated; otherwise, there was no pre-contact with respondents. Telephone follow-up was conducted for partially completed or problematic cases.

View the Questionnaire(s) and reporting guide(s).

Error detection

Error detection was applied at the processing stage of the survey. Once all questionnaires were captured, micro data analysis is performed using SAS, Excel or Access. Here, the main consideration is the energy intensity of each responding unit. Energy intensity is the energy consumed divided by the gross area of a building, expressed in energy units (normally gigajoules) per square metre. This is undertaken for each type of energy consumed (e.g. electricity, natural gas, etc.). Because the sample units used in the survey are diverse, such as a university or college campus or a hospital complex, the energy intensity measure is used to standardize consumption across the many different units. Any records with an extreme value for intensity were flagged and three variables are examined: their total energy, total area and number of employees (or number of beds if the responding unit is a hospital, or number of students for a school). If any of these three variables has a questionable value, that variable was flagged for imputation.


Automatic imputation was used to fill blank values for total energy consumption, total area, and number of employees/beds/students. The distribution of total energy consumption to each energy type was also imputed. Donor records were chosen using the "hot deck" method based on the following matching criteria:

. Industry Type, composed of groups of NAICS categories
. Energy Type, an eight character binary string showing which of the eight different energy types are used
. Employee Size Category, based on Business Register records
. Region
. Industry Group, composed of groups of Industry Types
. Measure Type -- i.e. beds, students or employees, indicating that the donor unit is a hospital, a school, or other establishment

The matching criteria are applied hierarchically beginning with the combination Industry Type, Energy Type, Employee Size Category, Region. If no match is found the Region criterion is dropped; if there is still no match, then Industry Group is substituted for Industry Type, and so on.

In addition, manual Imputation was used for records where an appropriate donor record could not be found in the automatic process. Using the nearest neighbour method, donor records are chosen which contain similar energy types (though not all the same energy components), are of a similar type of industry or have total energy values that were reasonably close in value. Once the donor records are found the blank values are filled using calculations similar to those in the automatic imputation method.


Estimation was performed at the national level for this survey. The micro data file was adjusted for non-response and then units were re-weighted for the purposes of estimation. The estimates are calculated using Statistics Canada's Generalized Estimation System as applied to a stratified sample.

Quality evaluation

The quality evaluation process includes historical trend analysis in order to gauge the consistency of reporting.

Disclosure control

Statistics Canada is prohibited by law from releasing any data which would divulge information obtained under the Statistics Act that relates to any identifiable person, business or organization without the prior knowledge or the consent in writing of that person, business or organization. Various confidentiality rules are applied to all data that are released or published to prevent the publication or disclosure of any information deemed confidential. If necessary, data are suppressed to prevent direct or residual disclosure of identifiable data.

Given that there is a micro data sharing agreement with Natural Resources Canada for this survey, all respondent identifiers are removed from the shared file. Identifiable cases are removed from the file to avoid any direct or residual disclosure. In certain situations, when there are not enough respondents to protect confidentiality at a given geographical level, NAICS codes are collapsed into a composite code.

Revisions and seasonal adjustment

This methodology does not apply to this survey.

Data accuracy

Upon completion of this survey, coefficients of variation were calculated for all energy types consumed. The overall response rate for this survey was 70 %. Given the highly variable nature of our unit of observation (i.e. university campuses, hospital complexes, etc.), a somewhat more generous letter grading system was developed in order to accommodate such natural differences.

Data Quality Indicator
A: Coefficient of Variation (CV) less than 20 percent
B: Coefficient of Variation (CV) between 20 percent and 29.99 percent
C: Coefficient of Variation (CV) between 30 percent and 39.99 percent
D: Coefficient of Variation (CV) between 40 and 49.99 percent; USE WITH CAUTION
F: Coefficient of Variation (CV) greater than or equal to 50 percent; ESTIMATE TOO UNRELIABLE TO BE PUBLISHED

Date modified: