Annual Industrial Consumption of Energy Survey (ICE)

Detailed information for 2010





Record number:


The survey provides estimates of energy consumption by manufacturing establishments in Canada.

Data release - December 1, 2011 (preliminary)


The Industrial Consumption of Energy survey, which is funded by Natural Resources Canada and Environment Canada, provides estimates of energy consumption by manufacturing establishments in Canada. These estimates serve as an important indicator of Canadian economic performance and are used by all levels of government in establishing informed policies in the energy area.

The survey results are used by Natural Resources Canada to track energy efficiency improvements and by Environment Canada to calculate carbon dioxide emissions. Industry also uses the information to monitor the results of their energy reduction efforts and to measure their contributions to Canada's climate change goals. Within Statistics Canada, the data are used as an input into the environmental accounts and statistics as well as into the annual Report on Energy Supply and Demand in Canada.

Natural Resources Canada and Environment Canada are provided with data on a regular basis. Statistics Canada has also entered into data sharing agreements with various agencies and government departments for this survey.

Reference period: year

Collection period: January to March


  • Energy
  • Energy consumption and disposition

Data sources and methodology

Target population

The target population comprises manufacturing establishments in Canada. Under the North American Industry Classification System (NAICS), manufacturing establishments are classified to NAICS 31, 32 and 33. The ICE questionnaire is also sent to some units outside of the manufacturing sector, such as mining, oil and gas extraction and utilities.

Instrument design

The questionnaire requests data on the consumption of various energy commodities such as electricity, natural gas, propane, diesel, wood and steam. It also asks information on the different usages of energy commodities: as fuel, to produce steam for sale, to produce electricity and for non-energy use.


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

The Industrial Consumption of Energy (ICE) survey began in 1995 with both annual and quarterly components. The quarterly component was discontinued in 2003. Since then, all units have been collected on an annual basis. The cost savings were reinvested into the survey to improve data quality by adding 1,000 units to the sample.

The sampling strategy was streamlined for the 2006 reference year while maintaining the same sample size, this allowed production of provincial estimates for use as input into the annual Report on Energy Supply and Demand in Canada. Of the 88 industries of interest, provincial estimates are produced for 35 of them.

The frame used for sampling is Statistics Canada's Business Register. The statistical unit is the establishment. The survey population includes all manufacturing establishments above certain thresholds that vary by industry and by reference year. To minimize the collection of data from smaller establishments and reduce their response burden, the smallest establishments (or "take-none" stratum) in each of the industries of interest in terms of their value of shipments or gross business income are excluded from the ICE sample.

Establishments are stratified by industry and by size based on their shipment value. Four strata are defined by size: one take-all, two take-some and a take-none. "Take-alls" are selected based on their uniqueness, their size and their importance in their industry.

The sampling for the take-some strata is done using Statistics Canada's Generalized Sampling System (GSAM). All sampled units are assigned a sampling weight. The sampling weight is a factor that indicates how many similar units the sampled unit represents in the population. This weight allows estimates for the population to be produced.

Data sources

Data collection for this reference period: 2011-01-04 to 2011-03-31

Responding to this survey is mandatory.

Data are collected directly from survey respondents.

The collection period begins in January with the mailing of the questionnaires to the selected establishments. Phone and fax follow-ups begin in February for establishments that have not yet responded. The Regional Office collects data from late reporters using Computer Assisted Telephone Interviews. The collection period ends in March.

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

Error detection

The following methods of error detection are used:

Edits are performed during data capture to ensure that keying errors are corrected. For example, a boundary edit identifies large changes in fuel consumption from one year to the next; and historical edits identify whether an establishment is using a fuel it has never reported consumption for in the past, or whether it is not using a fuel that it has reported past consumption for.

During and after the collection period, year-over-year validity of data is verified for large contributors to the industry and fuel estimates. Unusual, unexplained movements are questioned and, if need be, respondents are called to confirm the data.

All cases where the unit of measure has not been specified, or where the respondent has reported in the "other fuel" category, are looked at manually.

For establishments that have no previous data to compare with, an outlier detection method using Statistics Canada's generalized system BANFF compares the energy consumption of establishments with similar shipments within the establishment's industry to identify whether the data are reasonable or not.

Finally, for some industries data reconciliation at the micro-level is done with other survey results.


Imputation is used to determine plausible values for all variables that are missing or inconsistent with historical data. A number of statistical techniques are employed for this purpose that use survey data collected during the current and previous cycles.

For records for which historical data are available, missing fuel consumption values are calculated using a trend that represents the change in fuel consumption of the industry to which the establishment belongs.

Starting in 2006, an automated donor imputation program is used in addition to the other imputation methods mentioned above, using Statistics Canada's generalized system BANFF. Donor imputation involves identifying a respondent record that is similar to a non-respondent based on information that is available for both establishments (such as industry, value of shipments, and types of fuel consumed). The data available for the donor establishment are then used for the non-respondent.


The Generalized Estimation System is applied to calculate energy consumption estimates for each of the 88 manufacturing NAICS of interest. The most recent shipment values for these industries available from the Annual Survey of Manufactures are used to help correct sampling errors and to include an adjustment for the uncovered portion of each industry that was excluded from the sample, that is, the take-none stratum.

Starting in 2006, coal, coal coke, petroleum coke, coke on catalytic cracking catalyst, refinery fuel gas, butane and steam purchased are weighted, while they were not previously. This change allows the production of estimates that represent the population.

Quality evaluation

Throughout the collection and processing of the data, every effort was made to ensure that the results would be of superior quality. As part of the quality evaluation stage, initial survey results are validated by industry experts such as the Canadian Industry Program for Energy Conservation and the Canadian Industrial Energy End-use Data and Analysis Centre. Natural Resources Canada and Environment Canada are also important partners in this data validation stage. Data quality indicators are included in the CANSIM tables in order to provide users with information about the reliability of the data.

Disclosure control

Statistics Canada is prohibited by law from releasing any information it collects which could identify any person, business, or organization, unless consent has been given by the respondent or as permitted by the Statistics Act. 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.

Confidentiality analysis includes the detection of possible direct disclosure, which occurs when the value in a tabulation cell is composed of a few respondents or when the cell is dominated by a few companies.

Data accuracy

While considerable effort is made to ensure high standards throughout all stages of collection and processing, the resulting estimates are inevitably subject to a certain degree of error. These errors can be broken down into two major types: non-sampling and sampling.

Non-sampling error is not related to sampling and may occur for many reasons. For example, non-response is an important source of non-sampling error. Population coverage, differences in the interpretation of questions, incorrect information from respondents, and mistakes in recording, coding and processing data are other examples of non-sampling errors.

For the 2010 Annual Industrial Consumption of Energy Survey, the non-sampling errors were controlled through a careful design of the questionnaire, the use of a minimal number of simple concepts and consistency checks. Of the units contributing to the estimate, the response rate was 85%.

Sampling error occurs because population estimates are derived from a sample of the population rather than the entire population. Sampling error depends on factors such as sample size, sampling design, and the method of estimation. An important property of probability sampling is that sampling error can be computed from the sample itself by using a statistical measure called the coefficient of variation (CV). The assumption is that over repeated surveys, the relative difference between a sample estimate and the true value that would have been obtained from an enumeration of all units in the universe would be less than twice the CV, 95 times out of 100. The range of acceptable data values yielded by a sample is called a confidence interval. Confidence intervals can be constructed around the estimate using the CV. The standard error is calculated by multiplying the sample estimate by the CV. The sample estimate plus or minus twice the standard error is then referred to as a 95% confidence interval.

For the 2010 Annual Industrial Consumption of Energy Survey, CVs were calculated for each estimate. Generally, the more commonly reported variables obtained very good CVs (of less than 10%), while the less commonly reported variables were associated with higher but still acceptable CVs (under 25%). Some data might not be released because of poor data quality.

Date modified: