9. Recent developments A New Survey Measure of Disability: the Disability Screening Questions (DSQ)
9. Recent developments

9.1. Changes after the 2012 CSD

The 2012 Canadian Survey on Disability (CSD) long version of the DSQ is the one presented here. After the CSD, the short version was created, which incorporated changes that could benefit the long version.

Since cycle 28 of the GSS and the 2013 CIS, one change was made to the short version:

• To determine if the Pain screeners should be asked, at least one filter question should be positive, that is, it also considers the Other filter. To be able to do this, the Pain screeners were moved to the end of the short DSQ, after the Other filters are asked. An updated diagram of the flow of this version of the short DSQ is given in Appendix B.

Since the 2012 CSD, two changes were made to the long version:

• Filter 1 was split into two—Seeing and Hearing.
• The dynamic text for one of the Pain screeners (question S16) was changed so that respondents who said “Yes” to having pain that is always present (question S15) would then be asked (S16): “Do you also have periods of pain that reoccur from time to time?” The word "also" does not appear in S16 for people who said “No” to S15.

9.2. Using the long DSQ on the CSD

As mentioned in Section 7, the long version of the DSQ that will be used on the next CSD will no longer include filter questions. The DSQ filters (except Filter 6) will replace the old filter questions on the 2016 long form Census, and the survey frame for the CSD will include only people who have given at least one positive answer to these filters. The DSQ version on the next CSD will be similar to that used for the 2012 CSD (version described in this report) except that all filter questions have been removed.

The definition of disability in the next CSD will be the same as that in the 2012 CSD:^{Note 1} anyone who reports being “Sometimes,” “Often” or “Always” limited will be considered to have a disability, along with those who report being “Rarely” limited if they also report having “A lot” of difficulty, or “Cannot do at all.” This is the RACXR definition (definition 3 in Section 4.4.1.2).

9.3. Using the long DSQ on general population surveys

In recent months, STC and ESDC discussed the possibility of creating a modified version of the long DSQ for general population surveys willing to include the long version on some cycles.  If two different versions (short and long) of the DSQ are used on general population surveys, they must use the same definition of disability so that the estimates from all general population surveys are comparable.

Because the short version definition of disability is more restrictive, this new long version would have to be adapted to yield the same definition.^{Note 2}  This involves synchronizing the long DSQ and excluding all Rarely answers from the definition of disability.

Advantages of a long DSQ on general population surveys

There are two advantages of including a long DSQ on general population surveys. The first is that the long version allows identification of all 10 disability types, not just the five broad categories (Seeing, Hearing, Physical, Cognitive and Mental health-related). However, identifying the 10 types does not guarantee that estimates for each type will be reliable enough to be released. Although some surveys have considerable sample sizes, the percentage of the sample that represents PWD can be small. Moreover, because the prevalence of some disability types is very low, the sample size for each type may be insufficient.

The second advantage of including a long DSQ on general surveys is that a severity score could be computed. Severity of disability is linked to key factors in the lives of PWD (income, labour market participation, etc.); having a severity score for every respondent is a major value for analysis.   

Timing of long DSQ on general population surveys

Using data from the 2011 LFS quantitative test of the long DSQ, a simulation was performed to determine the impact of these changes to the long version on average interview time.  The simulation estimated how many respondents would have had to answer each question if the filters and screeners had been synchronized and the “Rarely” answers had been excluded from the definition of disability. With information from the "Audit Trail" files of the LFS test, it was estimated that this new version of the long DSQ would take 2 minutes 10 seconds, that is, 10 seconds more than the short DSQ. The average number of questions asked with this new version would be 8.7, slightly above the 7.9 average for the short DSQ (Section 6.7).

Therefore, for an extra 10 seconds of the DSQ, it would be possible to obtain information about 10 disability types and compute a severity score for each respondent.

Next steps

The modified version of the long DSQ for general population surveys, using the short version definition of disability, will be available for inclusion on surveys in the next few months. It is unclear if the short DSQ is still needed, given that the new long version has much more to offer at a low cost—an average of 10 extra seconds.

Moving from the short version to the new long DSQ version should be transparent for users.

• Previous
• Next