Commercial and Institutional Consumption of Energy Survey (CICES), December 2005

Appendix B
Scope of survey and methodology

This appendix offers a summary of the methodology used for the Commercial and Institutional Consumption of Energy Survey for 2004, conducted by Statistics Canada on behalf of the Office of Energy Efficiency.

Purpose of the survey

This survey's main objective was to obtain information on energy demand in Canada. It was designed to gather up-to-date statistical data on the energy consumption trends of establishments and enterprises in Canada.

Target population

Table 1 lists the 18 industries initially included in the survey.

Table 1
Description of frame size

Number of the industry within the frame size	Industry	NAICS Codes
1	Warehousing and storage – wholesale trade	41 and 49
2	Retail trade	44-45 (excluding 445)
3	Food retail	445
4	Information and cultural industries	51
5	Finance and insurance; real estate and rental and leasing services; professional, scientific and technical services	52, 53 and 54
6	Ambulatory health care services	621
7	Nursing and residential care facilities	623
8	Social assistance	624
9	Arts, entertainment and recreation	71
10	Accommodation services	721
11	Food services and drinking places	722
12	Other services (excluding public administration)	81 (excluding 813110)
13	Public administration	91
14	Colleges and cégeps	6112
15	Universities	6113
16	Hospitals	622
17	Elementary and secondary schools	6111
18	Religious organizations	813110

The database was created using elements from three mutually exclusive lists. The first list, taken from the Consumption of Energy Survey, 2003, includes universities, colleges and cégeps, and hospitals. The second, taken from the Business Register, includes establishments whose North American Industry Classification System (NAICS) codes match the industries numbered 1 to 13 and 18 in Table 1.The third, taken from a list provide by Statistics Canada's Division of Culture, Tourism and Centre for Education Statistics, includes elementary and secondary schools.

For the list of hospitals, only those with 50 or more employees were included in the survey in order to exclude entities and associations that may have the NAICS code of a hospital but not the same mandate or mission (for example, a Board of Directors). For the same reason, only colleges and cégeps with 20 or more employees were considered in this survey.

Table 2 shows the distribution of the target population, by industry.

Table 2
Size of population, by industry

Number of the industry within the frame size	Industry	Population
1	Warehousing and storage – wholesale trade	64 404
2	Retail trade	102 732
3	Food retail	23 305
4	Information and cultural industries	12 741
5	Finance and insurance; real estate and rental and leasing services; professional, scientific and technical services	182 317
6	Ambulatory health care services	56 162
7	Nursing and residential care facilities	6 383
8	Social assistance	16 818
9	Arts, entertainment and recreation	17 874
10	Accommodation services	10 859
11	Food services and drinking places	60 730
12	Other services (excluding public administration)	80 156
13	Public administration	7 006
14	Colleges and cégeps	222
15	Universities	126
16	Hospitals	736
17	Elementary and secondary schools	16 696
18	Religious organizations	14 923
	Total	674 190

Survey design

The previous edition of the Consumption of Energy Survey was limited to universities, colleges and hospitals. In fact, that survey constituted a census of these categories. Given the broader scope of the 2004 survey – which covers all the industries of the commercial and institutional sector – a census was not feasible. Accordingly, the 2004 edition of CICES constitutes a sample survey.

Statistics Canada used a stratified sampling plan, for which the stratus was the region, in the case of all industries, and the number of employees, in the case of all industries excluding numbers 14 to 17. Tables 3 and 4 describe the Region and Size strata.

The sample size for each domain – for each industry in this survey – was determined in such a way that the quality for each stratum (region and/or size) is equal. The final sample size was 7349 units.⁸ Table 5 shows the size of the sample for each domain (industry).

Table 3
Description of the Region stratum

Region	Province
Atlantic	New Brunswick, Nova Scotia, Prince Edward Island, Newfoundland and Labrador
Quebec	Quebec
Ontario	Ontario
Prairies	Manitoba, Saskatchewan, Alberta
British Columbia	British Columbia

Table 4
Description of the Size stratum

Size	Description
1	Number of employees > 0, and number of employees < 21
2	Number of employees > 20, and number of employees < 251
3	Number of employees > 250

Table 5
Size of sample, by industry

Number of the industry	Industry	Sample size
1	Warehousing and storage – wholesale trade	532
2	Retail trade	566
3	Food retail	452
4	Information and cultural industries	481
5	Finance and insurance; real estate and rental and leasing services; professional, scientific and technical services	610
6	Ambulatory health care services	550
7	Nursing and residential care facilities	399
8	Social assistance	426
9	Arts, entertainment and recreation	455
10	Accommodation services	368
11	Food services and drinking places	445
12	Other services (excluding public administration)	529
13	Public administration	384
14	Colleges and cégeps	151
15	Universities	114
16	Hospitals	201
17	Elementary and secondary schools	250
18	Religious organizations	436
	Total	7349

Data collection, validation and imputation

Survey responses were collected from January to March, 2005. Respondents were given three follow-up calls to encourage them to complete the questionnaire. Table 6 shows a distribution of the sample based on type of response.

Table 6
Distribution of sample, by type of response

Type of response	Distribution
Complete	1009
Void (duplicate)	22
Out of business	141
Out of scope	442
Partial	687
Refusal	646
No response	4135
Unable to locate	267
Total	7349

The response rate is calculated as follows:

Response rate =	Number of responding units (complete and partial)
Number	of resolved cases surveyed + Number of unresolved cases

The overall response rate is thus calculated as follows:

Response rate =	1009 + 22 + 687
	4135 + 646 + 267 + 1009 + 22 + 687

The response rate is therefore 25.4 percent. Table 7 shows the response rate for each of the 18 industries initially included in the survey.

Table 7
Response rate, by industry

Number of the industry	Industry	Response rate
1	Warehousing and storage – wholesale trade	21.8%
2	Retail trade	16.5%
3	Food retail	16.3%
4	Information and cultural industries	21.4%
5	Finance and insurance; real estate and rental and leasing services; professional, scientific and technical services	19.4%
6	Ambulatory health care services	19.2%
7	Nursing and residential care facilities	33.7%
8	Social assistance	26.8%
9	Arts, entertainment and recreation	21.8%
10	Accommodation services	20.6%
11	Food services and drinking places	11.9%
12	Other services (excluding public administration)	21.4%
13	Public administration	31.7%
14	Colleges and cégeps	59.1%
15	Universities	66.4%
16	Hospitals	57.0%
17	Elementary and secondary schools	34.7%
18	Religious organizations	34.9%
	Total	25.4%

Control rules ensured the validity and internal consistency of responses. The data were reviewed manually.

The situation in which a respondent fails to answer certain questions is called a partial response. In such cases, the missing data were imputed using a hierarchical hot-deck technique. The criteria for matching a data donor were: type of industry, type of energy, category of number of employees (based on data in the Business Register), region, and number of employees/students/beds. In addition, 50 records were imputed manually.

The estimates are based on the principle that each establishment in the sample could represent a certain number of establishments in the target population. Consequently, each respondent establishment was assigned a weighting coefficient indicating how many establishments within the population are represented by this one establishment.

Data suppression

For some industries, the response rate was too low to permit disclosure of data, as the resulting estimates did not meet Statistics Canada's quality standards. Accordingly, some categories were merged to improve the quality of the estimates. In order to publish the survey results, the 18 industries initially included in CICES 2004 were compressed into 8 categories and 7 subcategories. Table 8 shows how the activity sectors were categorized.

Table 8
Correspondence of the CICES 2004 categories with the industries in the survey design

Number of the industry	Industry	Categories established for publication purposes
1	Warehousing and storage – wholesale trade	Wholesale trade and warehousing
2	Retail trade	Non-food retail trade
3	Food retail	Food retail trade
4	Information and cultural industries	Information and cultural industries
5	Finance and insurance; real estate and rental and leasing services; professional, scientific and technical services	Office sector
6	Ambulatory health care services	Non-hospital health care
7	Nursing and residential care facilities	Non-hospital health care
8	Social assistance	Non-hospital health care
9	Arts, entertainment and recreation	Other
10	Accommodation services	Accommodation and food services
11	Food services and drinking places	Accommodation and food services
12	Other services (excluding public administration)	Other
13	Public administration	Office sector
14	Colleges and cégeps	Colleges and cégeps
15	Universities	Universities
16	Hospitals	Hospitals
17	Elementary and secondary schools	Elementary and secondary schools
18	Religious organizations	Other

The coefficient of variation, which indicates the reliability of the data, is used to determine which estimates may be published. Estimates where the coefficient of variation exceeds 50 percent are not reliable enough for publication. In addition, it is important to keep in mind that the calculation of the coefficients of variation does not take into account the fact that some data were imputed. Table 9 shows the various quality indicators associated with the coefficients of variation.

Table 9
Indicators associated with the coefficients of variation

Coefficient of variation	Indicator	Quality of estimate
Less than 20%	A	Excellent
From 20% to 29%	B	Good
From 30% to 39%	C	Acceptable
From 40% to 49%	D	Use with caution
50% or more	F	Too unreliable to be published

Estimates graded "A" or "B" are considered precise enough for most uses. Data where the coefficient of variation is higher – graded "C" or "D" – are precise enough for a few uses, but should be used with caution.

⁸ By way of comparison, the sample size for this survey is more than 40 percent larger than that of the similar Commercial and Institutional Building Energy Use Survey, 2000. This approach, based on the creation of a huge sample, was selected mainly because there had been no preliminary contact with each of the sample units. The various strata were thus oversampled in order to compensate for a foreseeably larger number of ineligible units.