Survey of Drinking Water Plants (SDWP)
Detailed information for 2015
Every 2 years
The Survey of Drinking Water Plants is conducted to provide Canadians with national and regional information related to the production of drinking water.
Data release - November 1, 2017
The survey is a census of drinking water plants serving 300 or more people, and asks for information on volumes of water treated, type of treatment, financial aspects of the operation, as well as raw (source) water quality. The survey results produce a national portrait of treatment processes and costs, and source water quality across Canada. These data will be used to track the state of water stocks on a regional basis in Canada, and will also be used in the development of environmental accounts.
Reference period: Calendar year
Collection period: January through June of the year after the reference period.
- Environmental quality
Data sources and methodology
The target population is composed of drinking water treatment plants that are licensed and regulated by provincial/territorial agencies (excluding First Nations communities) and that draw and process source/raw water from the environment to produce treated/potable water for consumption, serving 300 or more people.
The observed population comes from a frame created in 2007. At that time, Statistics Canada requested the inventories of drinking water plants held by the provinces and territories. It excludes systems that supply water to communities with less than 300 people and other regulated systems that service schools, campgrounds, commercial establishments, provincial parks, etc. This frame is kept up to date using responses from previous survey cycles and through contact with potential respondents prior to mail out of each new cycle.
The survey questionnaire was designed to collect data on volumes of water treated, treatment costs, processes used and source (raw) surface water quality. The questionnaire was developed in collaboration with subject matter experts at Health Canada and Provincial experts. A number of respondents were also consulted through individual meetings to ensure the information being asked was available and that the questionnaire could be filled out within a reasonable time frame.
Environmental Accounts and Statistics Division, in consultation with the Questionnaire Design Resource Centre, conducted testing of the questionnaire in July, September and October 2007 across Canada which included Atlantic Canada (5 locations), Québec (6 locations), Ontario (7 locations) and Western Canada (6 locations). Working group meetings were held with representatives of Environment Canada and Health Canada and final changes to the questionnaire were made based on the results of the field testing. The revised 2011 questionnaire was tested in both official languages in the fall of 2010 at 9 locations (5 in Ontario and 4 in Québec). The revised 2013 questionnaire was not tested because the only changes made were to remove or restrict the scope of some questions. The revised 2015 questionnaire was tested in both official languages in the summer of 2014 at 19 locations (7 in New Brunswick, 6 in Ontario and 6 in Western Canada).
This survey is a census with a cross-sectional design.
Data are collected for all units of the target population; therefore, no sampling is done.
The survey frame was created in 2007. At that time, Statistics Canada requested the inventories of drinking water plants held by the Provinces and Territories. The frame is kept up to date using responses from previous survey cycles and through contact with potential respondents prior to mail out of each new cycle.
Excluding systems that supply water to communities with less than 300 people and other regulated systems that service schools, camp grounds, commercial establishments, provincial parks, etc., a survey frame of approximately 2,000 drinking water facilities serving communities of 300 or more people was compiled, the majority being public (municipal) systems.
Data collection for this reference period: 2016-01-11 to 2016-06-30
Responding to this survey is mandatory.
Data are collected directly from survey respondents and extracted from administrative files.
Respondents are contacted to confirm contact information prior to the survey being sent. The questionnaires are mailed out and respondents are asked to return the completed forms within sixty days of receipt. The surveys are addressed to a contact person who is either responsible for, or has knowledge of, the drinking water plant being surveyed. A letter explaining the purpose of the survey, the requested return date and the legal requirement to respond was included with the mail-out package.
Telephone and fax follow-up are used to obtain data from respondents who returned incomplete questionnaires or who failed to respond. Returned questionnaires are scanned using an imaging system that captures the data for transfer into a database. Some data are manually keyed in. A capture and edit software is applied to run edit checks on the data, which serve to identify real or potential response errors.
The use of administrative data to replace or complement survey data is a priority for Statistics Canada. Efforts to achieve this goal involve the Survey of Drinking Water Plants where respondents in Québec are not surveyed by Statistics Canada. Instead, the data are collected by an existing survey administered by the Québec Ministère des Affaires municipales et de l'Occupation du territoire and the results are shared with Statistics Canada in accordance with section 13 of the Statistics Act. The micro data for Québec respondents are merged and processed with the survey data collected by Statistics Canada.
View the Questionnaire(s) and reporting guide(s) .
Many factors affect the accuracy of data produced in a survey. For example, respondents may have misinterpreted questions, answers may have been incorrectly entered on the questionnaires, and errors may have been introduced during the data capture or tabulation process. Every effort was made to reduce the occurrence of such errors in the survey.
Returned data are first entered and checked using the capture and edit software. This procedure verifies that all core cells have been filled in, that certain values lie within acceptable ranges, that questionnaire flow patterns have been respected, and that totals equal the sum of their components. Collection officers evaluate the edit failures and concentrate follow-up efforts accordingly. Phone follow-ups are performed to verify information in cases where edit checks fail.
If a record had no response for at least one mandatory cell after editing, the record was not processed any further and was considered a total non-response.
Further data checking is performed by subject matter officers who review returned data that have been identified statistically as outliers. Comparison with data from previous years is carried out to determine if the differences between years are reasonable. In some instances, collection officers are asked to confirm responses with the respondents. Subject matter officers also research drinking water plants (annual reports, web sites, etc.) in an effort to verify information submitted by respondents.
Outlier values were identified after collection by the Hidiroglou and Berthelot method (1986) through the use of the BANFF generalized system and reviewed by the client division for verification. Only real outliers were removed from the imputation process.
Statistical imputation is used for partial non-response records. Seven methods of imputation are used for the Survey of Drinking Water Plants:
- deductive imputation (only one possible value for the field to impute)
- imputation by linear regression
- trend imputation
- imputation by ratio
- donor imputation (using a "nearest neighbour" approach to find a valid record that is most similar to the record requiring imputation in terms of treated water volume and other characteristics)
- imputation by historical value (use of data from previous cycles) and
- manual imputation.
The criteria for ratio and donor imputation are various combinations of water treatment type, source water type and geographical location (province, region, or Canada). All of these methods are implemented using the Statistics Canada generalized edit and imputation system (BANFF). No imputation is conducted on water quality variables.
In the estimation process, the response values are multiplied by a factor adjustment (weight) to account for plants in the population who could not be contacted or were unable to participate in the survey. No estimation is conducted on water quality variables.
Totals, ratios, percentages were estimated with Horvitz-Thompson and the Generalized Estimation System (GES).
Micro data evaluation and error detection are important processes used to ensure good quality data. However, the final estimates obtained through the use of this micro data must also be evaluated in order to ensure accuracy. The quality of the estimates produced from a survey can be assessed through comparison to the trends obtained from other data sources and/or through a historical comparison to data obtained previously through the same survey. Estimates for the Survey of Drinking Water Plants were compared with the estimates from previous reference periods. This historical comparison was made to ensure that the estimates were coherent.
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.
A discretionary disclosure order (DDO) pursuant to paragraph 17(2)(a) of the Statistics Act was obtained to allow increased disclosure of aggregate information. The DDO permits the release of drinking water data, allowing the dissemination of a complete national profile of information related to the production of drinking water.
Revisions and seasonal adjustment
Revisions are made for the previous survey reference period, with the initial release of the current data, as required. The purpose is to address any significant issues with the data that were found between survey cycles. The actual period of revision depends on the nature of the issue. For the most current data please refer to CANSIM tables 153-0105 to 153-0108, and 153-0124 to 153-0128. The data are not seasonally adjusted.
Sampling error can arise when the information obtained from a sample of a population is used to derive an estimate for the entire population. Since the Survey of Drinking Water Plants is a census, sampling error is null.
Response error may be due to questionnaire design, the characteristics of a question, inability or unwillingness of the respondent to provide correct information, misinterpretation of the questions or conceptual problems. These errors are controlled through careful questionnaire design and testing and the use of simple concepts and consistency checks.
Processing errors may occur at various stages of processing such as data entry, editing and tabulation. All efforts are undertaken to minimize non-sampling errors through extensive edits, quality control steps and data analysis, but some of these errors are outside the control of Statistics Canada.
Non-response errors result when respondents refuse to answer, are unable to respond or are too late in reporting. Missing data items are imputed for partial non-responses.
Total non-response is dealt with by adjusting the weights assigned to units that did respond, such that one responding unit might also represent other non-responding units with similar characteristics (i.e. province, drainage region, source water type, size of population served). The pattern of total non-response, the estimation method, the number of respondents and the variability associated with each measured variable determines total non-response error. If non-respondents are assumed to be randomly "selected" from the population, then responding units may be treated statistically as a random sample. Under this assumption, a measure of total non-response error is the coefficient of variation (CV). It represents the ratio of the standard error of a survey to the estimate itself. Thus, although the survey is a census, getting a nonzero CV is possible. For this survey, CVs are calculated for the major variables and are indicated on the data tables.
A legend of the symbols used for the CV in our tables follows (table symbols are found in the link below):
A excellent data quality (coefficient of variation is 0.01% to 4.99%)
B very good data quality (coefficient of variation is 5.00% to 9.99%)
C good data quality (coefficient of variation is 10.00% to 14.99%)
D acceptable data quality (coefficient of variation is 15.00% to 24.99%)
E use with caution (coefficient of variation is 25.00% to 49.99%)
F too unreliable to be published (coefficient of variation greater than 49.99% - data are suppressed for the purpose of publication and made available only upon user request)
The 95% confidence interval associated with an estimate (assuming a normal distribution) is given by:
Estimate ± 1.96 * (CV in %) * Estimate/100.
The response rate for the survey was 82% in reference year 2015.
NOTE: Readers are advised, with the change in data collection for Québec respondents, comparisons from a time series perspective should be made with caution.