The purpose of the Canadian Survey on Disability (CSD) is to provide information about Canadian adults whose daily activities are limited because of a long-term condition or health-related problem. This information will be used to plan and evaluate services, programs and policies for adults with disabilities to help enable their full participation in Canadian society.
Data release – December 3, 2013
The Canadian Survey on Disability (CSD) gathers information about Canadians aged 15 and over whose daily activities are limited due to a long-term condition or health-related problem.
Information from the CSD will be used by all levels of government, as well as associations for persons with disabilities and researchers working in the field of disability. Data may be used to plan and evaluate policies and programs for Canadian adults with disabilities to help enable their full participation in society. In particular, information on adults with disabilities is essential for the effective development and operation of the Employment Equity Program. Data on disability are also used to fulfill Canada's international agreement relating to the United Nations Convention on the Rights of Persons with Disabilities.
The survey collects information on: type and severity of disability, use of aids and assistive devices, help received or required, educational attainment, labour force status, experiences and accommodations at school or work, and ability to get around the community.
The population covered by the CSD includes all adults aged 15 and over (as of Census/NHS day, May 10, 2011) who had an activity limitation or a participation restriction associated with a physical or mental condition or health problem and were living in Canada at the time of the Census/NHS. This includes persons living in private dwellings in the ten provinces and three territories. The population living on First Nations reserves is excluded, as are people living in collective dwellings. Since the population living in collective dwellings is excluded, the data, particularly for the older age groups, should be interpreted accordingly.
The target population of the CSD is a subset of the covered population, which consists of persons who were identified as a person with a disability (based on the social model of disability) during the CSD interview.
The CSD questionnaire was developed by Statistics Canada in collaboration with Employment and Social Development Canada (ESDC - formerly Human Resources and Skills Development Canada). Input for the survey was obtained from the ESDC Technical Advisory Group on persons with disabilities which consists of experts in the field of disability, including academics and representatives from various community associations across Canada. The questionnaire was tested in both official languages in the spring of 2012.
This is a sample survey with a cross-sectional design.
The sampling frame for the CSD includes all persons aged 15 and over, as of May 10 2011 (Census/NHS day), who answered "Yes" to at least one of the National Household Survey (NHS) filter questions on activity limitations and who lived in Canada at the time of the survey. The sampling unit is the person.
Sampling design and stratification
The sample design can be viewed as a three-phase design where the first two phases were for the selection of the NHS sample itself and the third phase was for the selection of the CSD sample. In the first phase, sample selection of the NHS itself corresponds to a systematic sample of approximately one dwelling out of three across Canada in non-remote areas (N1 regions). The second phase corresponds to subsampling of NHS non respondents in N1 regions, a new procedure that was put in place to mitigate the potential effect of non-response bias due to the higher non-response that resulted from the NHS. In this phase, a subsample of non-respondent dwellings was selected for non-response follow-up (NRFU).
In the third phase, the CSD sample was selected from the group of individuals who responded to the NHS (including the NRFU subsample) and reported activity limitations within the NHS. The sample was selected so that there would be a sufficiently large sample in each estimation domain.
One of the survey objectives was to allow for the estimation of the number of individuals with a disability by province/territory and of various age groups in the population. The CSD estimation domains were formed by crossing the province and the following age groups: 15 to 24 years old, 25 to 44 years old, 45 to 64 years old, 65 to 74 years old and 75 years and over. For each of the three territories, the estimation domain includes a single age group (15+). For Prince Edward Island, the first two age groups had to be combined because of their very small population sizes.
Each estimation domain was further sub-divided into six strata to take into account both the severity of the activity limitation according to the NHS and the NHS sample design. To control for severity, those people who answered "Yes, often" at least once to the NHS filter questions were placed in the "Often" stratum, while those who answered "Yes, sometimes" at least once but never "Yes, often" were placed in the "Sometimes" stratum.
To control for the NHS sample design, two characteristics were considered: (1) being among the NHS's initial respondents as opposed to the non-response follow-up respondents; and (2) living in a remote region or not. Controlling for these characteristics effectively grouped people with similar preliminary (NHS) weights.
The sample sizes were determined in such a way that, for each estimation domain, a minimum proportion could be estimated with a maximum coefficient of variation (CV) of 16.5%. At Statistics Canada, 16.5% is often used as the upper limit for the CV of an acceptable estimate. The minimum proportion that could be estimated was set at 9% for the 15 to 24 years old, 7.5% for the 25 to 44 and the 45 to 64 years old, and 11% for the 65 to 74 year olds as well as 75 years and over. For Prince Edward Island, the minimum proportion to be estimated for the 15 to 44 years old was set at 9%, while for 15 year olds and over in the three territories, it was set at 8%.
An optimal allocation method among the substrata of a given area was used. It took into account various types of sample size losses, such as expected non-response and the expected proportion of false positives (i.e., individuals who reported an activity limitation in the NHS but did not report a disability in the CSD).
The total sample size for the CSD was 45,443 individuals.
Data collection for this reference period: 2012-09-24 – 2013-01-13
Responding to this survey is voluntary.
Data are collected directly from survey respondents and derived from other Statistics Canada surveys.
The questions in the CSD were administered using Computer Assisted Telephone Interviews (CATI). Accommodations were made to maximize participation by sending an introductory letter to all respondents which explained the purpose of the survey and the importance of their participation. A brochure was included in the introductory package with an email address to reach us for more information, a TTY telephone number (Telecommunications Device for the Hearing Impaired) and a Braille insert containing contact information for the visually impaired. A link was also provided to direct respondents to a webpage specific to CSD on the Statistics Canada site. This webpage contained general survey information for respondents.
A PAPI questionnaire was also used to conduct in-person interviews with a sample of respondents in the Northwest Territories who did not have a telephone and could not otherwise have been contacted.
Interviews by proxy were allowed in some circumstances, such as when the selected respondent was absent for the duration of the survey, spoke neither English or French, was not able to participate due to mental or physical health problems or when a parent/guardian insisted on responding for his or her 15 to 17 year old child.
In order to reduce the number of questions asked on the CSD, Statistics Canada combined information collected in the 2011 NHS to the information provided during the CSD interview.
In addition, a sample of approximately 124,000 NHS respondents was selected and added to the analytical file. These were respondents who answered no to both of the 2011 NHS filter questions and they are considered to not have a disability. They were not contacted to participate in the 2012 CSD. The extra sample was needed to compute disability rates and to allow comparisons to be made between persons with and without a disability using NHS variables. Survey weights are available for this stratified sample which takes the NHS sample design into account.
Responses to survey questions were captured directly by interviewers using a computerized questionnaire. This reduced processing time and costs associated with data entry, transcription errors and data transmission. Data from the CSD paper questionnaires were also entered into the CATI system once forms were returned from the field and were thus electronically captured for further processing.
Some editing of data was done directly at the time of the interview. Specifically, where a particular response appeared to be inconsistent with previous answers or outside of expected values, message screens on the computer prompted the interviewer to confirm answers with the respondent, and, if necessary, to modify the information.
Once the data were received at head office, they were processed through a series of steps to convert the questionnaire responses from their initial raw format to a high-quality, user-friendly database involving a comprehensive set of variables for analysis.
During processing, analyses were performed on the data to identify gaps, inconsistencies, extreme outliers and other potential problems in the data. CSD data were evaluated both for internal consistencies across variables and for external consistencies in relation to the respondents' linked NHS data. Customized edits and imputations were performed to resolve any problems found.
As an example of internal edits, there were a number of edits to ensure that responses related to age throughout the questionnaire were not contradictory. As another example, edits ensured that the number of hours a respondent reported that they usually worked per week did not exceed 168 (24 hours x 7 days) and prompted interviewers to confirm the hours with respondents whenever reported hours per week exceeded 80.
In terms of external edits, analyses were performed comparing CSD and NHS data for those indicators where the concepts across surveys overlapped in meaningful ways and where inconsistencies would have the potential to present analytical problems of interpretability for users. Over 20 indicators were analyzed, primarily in the areas of labour force participation, educational attainment and family status, and inconsistencies were resolved.
Identified discrepancies, logical inconsistencies and missing information were resolved wherever possible either automatically using customized deterministic editing rules or through the intervention of more complex modelling.
In one particular situation, a problem was found in the flow of the electronic questionnaire for some respondents due to missing data on a key input variable for these cases. Imputation was used to replace the missing data on the mistakenly skipped question with a new value. Imputation was based on two logistic regression models for predicting the probability of possible responses. The model was constructed based on the responses of other respondents with a disability. In a second similar situation, where no adequate donor pool could be identified for imputation, values were left missing on the final database.
In the case of discrepancies between CSD and NHS data, the CSD data was maintained as the most current data and NHS values were assigned a special missing value.
In a sample survey, each respondent represents not only himself/herself, but also other people who were not sampled. For that reason, each respondent is assigned a weight which indicates the number of people that he or she represents. To maintain data coherence and ensure that the results accurately represent the target population and not just the individuals sampled, that weight must be used to compute all estimates.
There were several steps in calculating the weights for the CSD. The first step was to assign each unit selected for the CSD an initial weight based on the sample design. The initial weight was the inverse of the probability of inclusion. For the CSD, the initial weight was the product of two factors: the NHS final weight and the inverse of the CSD subsampling fraction.
In the second stage of the weighting process, a number of adjustments were made to the weights to control for non-contact and non-response after contact.
The final step was to calibrate the survey weights on the NHS estimated totals and then to proceed with an additional calibration to account for units that were in scope during the May 2011 selection process but out of scope at the time of the survey in 2012.
Weights were also computed for the 124,000 persons in the NO sample that were added to the analytical file (but for whom no collection was done). The initial weight for this sample was the product of the NHS final weight and the inverse of the CSD subsampling fraction. A calibration of these weights was done first on the NHS estimated totals, and then an additional calibration was done to account for the estimated number of units who became out of scope between Census/NHS day and time of CSD collection.
The bootstrap method was used to calculate variance. None of the existing methods were appropriate for the 2012 CSD's complex design. To calculate variance only, the NHS design was deemed to be a three-phase design. The first phase consisted of sampling one in three households in the N1 regions. The second phase involved subsampling non-respondent households in the N1 regions to follow up on non-response. The third phase was non-response from the households selected for non-response follow-up. To simplify the problem of calculating variance, the different NHS phases were combined into a single phase.
CSD sampling was then considered to be the second phase, and the generalized bootstrap method for two-phase sampling developed in 2006 for the Aboriginal Peoples Survey (APS) was used (see Langlet, É., J.-F. Beaumont, and P. Lavallée. (2008), "Bootstrap Methods for Two-Phase Sampling Applicable to Postcensal Surveys". Paper presented to Statistics Canada's Advisory Committee on Statistical Methods, May 2008, Ottawa.)
Quality assurance measures were implemented at every collection and processing step. Measures included recruitment of qualified interviewers, training provided to interviewers for specific survey concepts and procedures, observations of interviews to correct questionnaire design problems and instruction misinterpretations, procedures to ensure that data capture and coding errors were minimized, and edit quality checks to verify the processing logic. Data are verified to ensure internal consistency and they are also compared to other sources.
The new set of Disability Screening Questions (DSQ) used in the CSD have never been used before which means that no other data source exists that would allow us to compare estimates from the CSD. Nevertheless, distributions of persons with a disability crossed by several characteristics from the CSD were compared with similar distributions from the 2006 Participation and Activity Limitation Survey (PALS) to ensure that there was some consistency. Even though the two surveys are not based on exactly the same concept of disability, it was possible to see that the CSD results are consistent with what was expected.
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.
To reduce risk of disclosures, all estimates are rounded to the nearest 10 units.
Two types of errors occur in surveys: sampling errors and non-sampling errors.
The sampling error measure used for the CSD is the coefficient of variation (CV) of the estimate, which is the standard error of the estimate divided by the estimate itself. In this survey, when the CV of an estimate is less than or equal to 16.6%, the estimate can be used without restriction. When the CV is greater than 16.6% but less than or equal to 33.3%, the estimate will be accompanied by the letter "E" to indicate that the data should be used with caution. When the CV of an estimate is greater than 33.3%, the cell estimate will be replaced by the letter "F" to indicate that the estimate was suppressed for reliability reasons.
Non-sampling errors arise primarily from the following sources: non-response, coverage, measurement and processing. Total non-response will produce a bias if non-respondents have different characteristics from respondents and if non-response is not corrected properly. Non-response adjustments, combined with a relatively high response rate, helped reduce this risk of bias substantially. Non-response to specific questions is often due to difficulty understanding the questions. Thorough quality reviews and questionnaire testing were carried out before the survey, which reduced the extent of partial non-response.
Collection for the CSD ended with an overall response rate of 74.6%. This response rate is the number of complete respondents divided by the number of cases sent to collection minus the out-of-scope cases (cases which did not meet eligibility criteria for the survey). Hence, this rate reflects the percentage of cases that completed the interview relative to the number of cases that should have completed it.
Response rates for the provinces ranged from 68.4% for New Brunswick to 79.6% in Saskatchewan. In the territories, the response rates ranged from 64.2% in Nunavut to 73.3% in the Yukon. By age group, the response rates ranged from 65.8% for 25 to 44 year olds to 84.7% for 65 to 74 year olds.
Coverage errors occur when there are differences between the target population and the sampled population (or survey population). In particular, undercoverage can be problematic. Since the CSD sample is selected from National Household Survey (NHS) respondents, NHS non-respondents could not be sampled for the CSD. Aside from the coverage issues related to the Census and the NHS, non-response bias in the NHS can also translate to a coverage bias in the CSD. To limit the scope of this problem, special non-response follow-up procedures were used in the NHS to reduce the potential bias for at-risk populations and special weighting strategies were used to reduce potential bias. For more information about the quality of NHS data, see the NHS User Guide (Catalogue number 99-001-x2011001).
Measurement errors occur when the response provided differs from the real value. Such errors may be attributable to the respondent, the interviewer, the questionnaire or the collection method, for example. For the 2012 CSD, every effort was made to develop questions that would be understandable, relevant and appropriate for respondents. Other measures were also taken, including the use of skilled interviewers, extensive training of interviewers, and observation and monitoring of interviewers.
Processing errors may occur at various stages, including data capture, coding and editing. Quality control procedures were applied at every stage of data processing to reduce this type of error.