View the most recent version.

Data Processing

(Document last updated May 1, 2001)

Crunching the numbers

Once every five years on a designated day, everyone in Canada takes part in the same national event — the Census of Canada.

The countdown for the 2001 Census began more than five years ago when the first planning meeting was held in December 1995. During the first two weeks of May, 2001, 34,000 census representatives will deliver census questionnaires to every household in Canada. On May 15, 2001, Census Day, respondents will “count themselves in” when they mail back their completed forms.

Following the collection activities, the next phase in the census process begins approximately one month after Census day — translating the responses from approximately 11.8 million households into meaningful data. This part of the census cycle is known as the Processing phase and is divided into five main activities:

• Regional processing
• Imaging
• Interactive verification
• Automated coding
• Edit and imputation
• Glossary of terms

Regional processing

(June 2001 to the end of October 2001)

Regional Processing staff are responsible for the manual preparation and data entry of the completed census questionnaires.

Given the enormous volume of census questionnaires and data to be keyed, Regional Processing contracts the work to Canada Customs and Revenue Agency (CCRA), formerly Revenue Canada. The coming census marks the fifth time Statistics Canada and CCRA have joined together in this working arrangement.

For the 2001 Census, approximately 2,800 CCRA employees will be sworn to secrecy under the Statistics Act and are deemed Statistics Canada employees for the duration of their work on the census. Approximately 13.2 million questionnaires will be keyed using approximately 5 billion keystrokes. CCRA will perform the work at the seven regional tax centres across the country: St. John’s, Summerside, Shawinigan, Jonqui�re, Sudbury, Winnipeg, and Surrey; and at the International Tax Services Office in Ottawa.

As the collection activities wind down, the questionnaires from completed enumeration areas including maps and visitation records (VRs) are shipped in EA boxes to the designated tax centre for the following Regional Processing activities.

Manual coding — The first objective is to prepare the completed questionnaires for data capture, including the manual assignment of numeric codes to written answers for the questions on the 2B questionnaires concerning industry and occupation. All of the other written responses are converted to numerical codes using automated systems during the Automated Coding activities.

Direct Data Entry — Data capture is performed by traditional keying at mainly mainframe terminals. Verification of the accuracy of the keyed data is done by selecting a sample of questionnaires already key-entered and capturing the information directly from the questionnaire a second time. Quality control is achieved by comparing the two sets of captured information.

As the data are key-entered, they are transmitted across dedicated lines to the main CCRA office in Ottawa. Then:

• the data are transferred to cartridges and transported by bonded carrier to Statistics Canada on a daily basis to be loaded on the mainframe computer in Ottawa; and finally,

• the questionnaires are reassembled in their EA boxes for shipment to Statistics Canada’s 2001 processing site in Ottawa.

Imaging

(mid-July 2001 to November 2001)

In previous censuses, the remaining processing steps were accomplished using the paper questionnaires. For the 2001 Census, that process will be replaced by imaging (i.e., scanning) all of the questionnaires and VRs as soon as they arrive at Statistics Canada’s 2001 processing site in Ottawa.

Steps in imaging

• After receiving the EA boxes at the 2001 processing site, the questionnaires and VRs will be sorted and batched.

• The questionnaires and VRs (because they are in booklet format) will be cut into separate sheets to prepare them for the scanners.

• The questionnaire and VRs will be imaged using high volume scanners. Quality control procedures will be performed to ensure that the images are legible.

• The images will reside on disc until the Interactive Verification activities using those images are complete.

• The images will then be written to permanent media (optical platters) for subsequent access through an image retrieval system. At the end of the processing activities, the image retrieval system will be moved to STC’s head office. The images (on platters) will be kept in a secure location and will be accessible only to authorised STC employees from within a secure location. They will not be accessible to individual offices.

Interactive Verification

(mid-July 2001 to mid-November 2001)

The main objective of the Interactive Verification (previously called Head Office Processing) activities is to produce a ‘cleaner’ version of the raw data captured during Regional Processing. This means that any further editing and correction on the data can be done using automated systems without having to refer back to the questionnaires or VRs.

• Structural editing — Edits are performed on the raw data to ensure such things as unique geographic household identifiers, and consistent questionnaire and person counts. Corrections will be determined by referring to the questionnaire and VR images, and the electronic data will be corrected using an interactive PC-based system.

• Block quality control — The objective of this step is to verify the quality of the block number. Automated and manual processes will be used to ensure that the block number that is captured from the questionnaire at CCRA is the same as the block number on the enumerator’s EA map.

• 2Cs and Ships — Questionnaires received from Canadian citizens living overseas or on ships are evaluated to ensure that the people enumerated on these forms were not already included on other census forms.

• Dwelling Classification Study — As a result of this study, adjustments are made to the census counts to account for people living in dwellings that were incorrectly enumerated as vacant, and those living in dwellings for which no questionnaires were returned.

Automated coding

(September 2001 through March 2002)

The 2B long-form questionnaire contains several questions where answers can be checked off against a list, as well as many questions where the respondent must write in an answer in the boxes provided. These written responses must be converted to numerical codes before they can be tabulated for publication purposes. For the 2001 Census, all written responses on the long-form questionnaire, except those for industry and occupation, will undergo automated (computer) coding systems to assign the numerical code. Specially trained coders and experts will resolve cases that can not be assigned automatically. The following is the list of questions that will have written responses. The variables shown in bold are new for 2001:

Edit and imputation

(August 2001 through April 2002 for the 2A data and through October 2002 for the 2B data)

The data collected in any survey or census contain omissions or inconsistencies. These errors can be the result of respondents missing a question, or they can be due to errors generated during processing. For example, a respondent may be unwilling to answer a question, may fail to remember the right answer, or may misunderstand the question. Census staff may code responses incorrectly or may make other mistakes during processing.

The editing process detects errors and the imputation process corrects them. The edit and imputation phase is important because:

• Consistent estimates are essential to users, particularly those counts that are used as official estimates for legislative and administrative purposes.
• If invalid or missing responses are not adjusted, data users would have to tabulate incomplete data or develop their own estimates, which would not be consistent with other results.
• Many data users do not wish to adjust or tabulate incomplete data.
• Correct data are necessary for processing purposes. For example, family patterns are constructed based on information provided by respondents on age, sex, marital status, relationship to Person 1, etc. If these data are missing or inconsistent, family characteristics cannot be compiled.

Weighting — Questions on age, sex, marital status, mother tongue and relationship to the household reference person (Person 1) are asked of 100% of the population. However, the bulk of census information is acquired on a 20% sample basis (with the additional questions on the 2B questionnaire). “Weighting” is used to project the information gathered from the 20% sample to the entire population.

A simple weighting method would be to multiply all the units by five, however, a more sophisticated technique is used. The census weighting procedure starts with initial weights equal to approximately five. These initial weights are then adjusted by as small an amount as possible so that various sample estimates (e.g. number of males, persons aged 15-19, etc.) agree with known population counts.

As one of the final processing steps, all households will be assigned to a DA using their block number. The results of all of these processing steps are the final 2001 census databases that will be the source of data for all publications, tabulations and custom products.

This newsletter is produced by the Census Communications Project at Statistics Canada. Reprinting is welcome.