Survey of Digital Technology and Internet Use (SDTIU)

Detailed information for 2013

The Survey of Digital Technology and Internet Use measures the adoption and use of various digital technologies, including the Internet.

Data release - June 11, 2014

• Questionnaire(s) and reporting guide(s)
• Description
• Data sources and methodology
• Data accuracy

Description

The use of the Internet and digital technology has become pervasive in today's society. Like electricity, the Internet has become so ingrained in the way that business operates, it can be considered a General Purpose Technology (GPT) because it affects the way the entire economy functions. The use of digital technologies and electronic commerce, while of great potential, also pose numerous challenges to both businesses and policy makers. In order to understand the effects and impacts of these technologies, it is essential to first understand who is using them and how.

The data from this survey are used by businesses and policy makers to monitor the uptake and adoption patterns of Information and Communication Technologies (ICTs) and assess their impact on the economy. Results are also monitored by international organizations such as the Organization for Economic Co-operation and Development (OECD) for benchmarking purposes and to study the development and the influence of the digital economy.

Statistical activity

Science and technology (S&T) and the information society are changing the way we live, learn and work. The concepts are closely intertwined: science generates new understanding of the way the world works, technology applies it to develop innovative products and services and the information society is one of the results of the innovations.

People are looking to Statistics Canada to measure and explain the social and economic impacts of these changes.

The purpose of this Program is to develop useful indicators of S&T activity in Canada based on a framework that ties them together in a coherent picture.

Reference period: The 12-month calendar year

Collection period: November of the reference year through March of the next year

Subjects

• Business and government internet use
• Information and communications technology

Data sources and methodology

Target population

This survey covers private firms in almost all industrial sectors with the exception of the following NAICS subsectors: 111, 112, 114, 1151, 1152, 238, 55114, 814.

All firms, except for very small firms, were included in the target population. Enterprises that had under $100K or $250K in revenue, depending on the sector, were excluded from the population frame.

The industrial classification assigned to an enterprise engaged in electronic commerce is the industrial classification of the establishment with the highest value-added within that enterprise. For instance, if an Internet transaction were conducted in a retail establishment within a manufacturing enterprise, that activity would be classified as a sale of the manufacturing sector if that establishment provided more value-added to that enterprise.

Instrument design

The survey data was collected using an electronic questionnaire.

The questionnaire was significantly redesigned from the previous version in 2013 to meet the policy needs of the sponsoring partner.

Cognitive testing of the questionnaire content was carried out in two phases in conjunction with the Questionnaire Design Resource Centre based at Statistics Canada in both official languages. The first round of testing concentrated on validating respondents' understanding of concepts, questions, terminology, the appropriateness of response categories and the availability of requested information. During the second round, the content was tested through one-on-one interviews with 18 potential respondents that took place in Gatineau, Montreal and Toronto. The resulting comments and analysis of these interviews led to further revision of the questionnaire to make the questions more relevant to respondents and easier to answer.

Sampling

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

The survey focuses on the use of information and communications technologies such as personal computers, e-mail and the Internet from a sample of Canadian enterprises in the private sector. This is the target population of the survey.

The frame consists of enterprises on the Business Register (BR) developed by Statistics Canada. The sampling frame is first stratified by North American Industrial Classification System (NAICS) at the level required for estimation (NAICS 2 or 3 level). Within each industrial level, three strata by size are built: large units, which are sampled with certainty, and medium and small units, in which the sampling is conducted using a probability of selection. Size is based on the revenue of the enterprise. There's also a take-none stratum. The enterprises that are under $100K or $250K in revenue, depending on the sector, were excluded from the population frame.

The sample unit for the survey is the enterprise which is the organizational unit of a business that directs and controls the resources relating to its domestic organization and for which consolidated financial and balance sheet accounts are maintained. The implication is that the survey collects data on transactions that occur between enterprises, while it specifically excludes intra-firm transactions, i.e. Internet transactions that may occur between two establishments or companies within the same enterprise.

The method used is the Lavallée-Hidirouglou algorithm, which does the stratification and the sample allocation to strata by minimizing the sampling size while attaining the target coefficient of variation (CV) based on the size variable.

A sample of around 17,000 enterprises targets a CV between 2.5% and 5% in the majority of industries.

Once the stratification and the allocation are done a minimum sample size of 20 units is set for the small units in some sectors. As well, sample sizes in key sectors are boosted to ensure more accuracy for the small and medium strata. The sample sizes are adjusted for non-response with an expected response rate of 65%.

All units are selected with certainty in the take-all strata while a random sample is selected in the take-some strata.

Data sources

Data collection for this reference period: 2013-09-30 to 2014-03-14

Responding to this survey is mandatory.

Data are collected directly from survey respondents.

Data are collected through two modes, primarily through an electronic questionnaire application and also through a paper questionnaire. Businesses are initially contacted by telephone during a pre-contact phase to ask whether they used the Internet and also to have them provide an appropriate contact and e-mail address within the business. Only those firms that could not be reached or did not provide an e-mail address during pre-contact are mailed a paper questionnaire. Those firms that received an electronic questionnaire could also request a paper questionnaire on an adhoc basis.

Follow-up because of non-response, inconsistent or missing data is done by phone on a priority basis.

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

Error detection

At data collection, some edits are applied to each questionnaire such as rules of consistency and validity. During processing additional consistency edits are applied. Outlier detection is done on the variable "Sales over Internet" using Statistics Canada's Generalized System for edit and imputation (BANFF). The outlier detection is done for each industry. Additionally, historical data were used to validate reported "Sales over Internet" values.

Imputation

Imputation is used for partial non-response. Because there was overlap in the sample, and some of the questions on the questionnaire were repeated there are historical data available for some units. Historical data were used to impute for missing data where possible. Historical imputation was done using a few different methods: direct replacement, replacement with a ratio adjustment, and donor imputation where the donor and recipient had the same response in historical data.

Tax data were used to impute the "Sales over Internet" variable in combination with the "percentage of sales conducted over the internet." Imputation using administrative data is used to impute the question referring to the number of employees.

Donor imputation is used in the remaining cases to replace missing or inconsistent values with those of the nearest respondent according to characteristics such as size, industrial classification and key variables from the questionnaire. A procedure in Banff is used to do the nearest neighbour donor imputation. Imputation is conducted within homogeneous groups, the initial imputation group corresponding to the stratum. If there is not a sufficient number or proportion of donors in a group, or if imputation from all available donors would result in questionnaire inconsistencies, we move to a more aggregated imputation group.

Outlier enterprises are excluded from the donor pool. When imputation is done for the sales over the internet, the donor's revenue must be within 75% to 125% of the recipient's revenue. When imputation is over, the initial edit rules are reapplied to assure the consistency of all the questionnaires going into the estimation process.

Estimation

Non-response weight adjustment is used for complete non-response. The weighting is done in two phases because a screening question was asked in pre-contact. Statistics Canada's Generalized Estimation System (GES) is used for estimation. There are three types of estimates produced: in the case of percentage variables (P), a ratio is used to derive an estimate, in the case of categorical variables (C), again a ratio is used and in the case of numerical variables (Y), the usual estimator of the total is used.

Quality evaluation

Survey results are analyzed at both the micro and macro level. At the micro level, checks are performed on the data to verify internal consistency and identify extreme values. At the macro level, the data are subjected to a detailed quality review process, including a comparative analysis to prior year. Material errors are thereby identified and corrected.

Disclosure control

Statistics Canada is prohibited by law from releasing any information it collects that 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.

In order to prevent any data disclosure, confidentiality analysis is done using the Statistics Canada Generalized Disclosure Control System (G-Confid). G-Confid is used for primary suppression (direct disclosure) as well as for secondary suppression (residual disclosure). Direct disclosure occurs when the value in a tabulation cell is composed of or dominated by few enterprises while residual disclosure occurs when confidential information can be derived indirectly by piecing together information from different sources or data series.

Revisions and seasonal adjustment

This methodology does not apply to this survey.

Data accuracy

While considerable effort is made to ensure high standards throughout all stages of collection and processing, the resulting SDTIU 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 interpretations of questions and mistakes in recording, coding and processing data are other examples of non-sampling errors.

Coverage error results from inadequate representation of the intended population. This error may occur during selection of the survey population, or during data collection and processing. In order to avoid such errors, a number of sources describing the population of the industry are used and compared.

The response rate at estimation was 75.0%.

Response error may be due to many factors, including design of the questionnaire, interviewers' or respondents' misinterpretation of questions, or respondents' faulty reporting. Frequent changes in company personnel may also lead to response error. Several features are in place to help respondents complete the questionnaire, including logic and consistency checks, and a glossary of terms and concepts. Responses are compared from year to year and any significant deviations are queried by analysts to ensure their accuracy. However, even with these checks, the quality of data depends on the respondent's willingness to consult their records.

Non-response error occurs because not all respondents cooperate fully. To alleviate the impact on the survey, respondents are usually asked to provide key variables and the others are estimated.

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. The sampling error is measured by a quantity known as the standard deviation. The latter indicates the expected variability of the estimate that would be produced if we sampled repeatedly. The actual value of the standard deviation is unknown, but it can be estimated from the sample.