4. The new DSQ A New Survey Measure of Disability: the Disability Screening Questions (DSQ)
4. The new DSQ

This section presents the questions retained in the final^{Note 1} version of the DSQ. The goals of the DSQ are reviewed, followed by a list of disability types covered. The wording of each question and answer category is given.

4.1. Goals of the new DSQ

The DSQ were developed with three main goals:

1. The questions had to cover as many disability types as possible, including questions on activity limitations for each type to follow the social model of disability;
2. The questions had to cover all the information needed to calculate a severity score; and
3. They had to be short enough to be administered in two minutes or less, on average, per respondent in a general population survey.

Development of the DSQ was challenging, as the goals are contradictory—the first two imply that a considerable amount of information must be collected, but the third requires that the number of questions be kept to a minimum.

4.2. Disability types covered

Table 1 lists the 10 disability types covered in the long DSQ. “Other/Unknown” is not counted as one of the 10 types; it is a “catch-all” category for people who reported a health problem or condition that limits their daily activities, but does not fit one of the 10 disability types.

The DSQ do not cover “Communication” disabilities. Questions to identify persons with a Communication disability were tested during qualitative interviews and did not work as expected, ^{Note 2} owing largely to the evolving nature of communication, notably, the growing importance and use of social media and technology.

Unlike earlier instruments, the DSQ split Agility into two separate types: Flexibility and Dexterity. This was done at the request of the TAG and was also based on comments from users, noting differences in barriers faced by those with dexterity and flexibility disabilities. Although such disabilities may affect a similar population (mostly older adults), they require different aids and assistive devices.

4.3. DSQ filter and screener questions

As described in Section 3.2 (Second round of qualitative testing – March 2011), filter questions were added at the beginning of the DSQ to reduce average interview time. These filters identify people who are not likely to have a disability and allow them to skip all the subsequent screener questions.

4.3.1. Filter questions

The filter questions must be kept to a minimum, as they are asked of every respondent. They must be brief, inclusive enough to avoid filtering out persons who have a disability, and yet not too inclusive to reduce the number of false positives.

The current version of the DSQ contains six filter questions to cover five categories of disability types:  Sensory, Physical, Cognitive, Mental health-related, and Other/unknown (Table 1). There is one question for each category, with an extra question for the Other/unknown category. The DSQ filter questions are numbered from 1 to 6 with the prefix F (for Filter), which distinguishes them from the screeners presented later.

4.3.2. Developing the filter questions

Because the goal of the filter questions is to prevent people who say “No” to all of them from going through the screeners, it would seem that the answer categories “Yes” and “No” would suffice. The choice of answers is meant to give respondents some flexibility by allowing for “softer” responses. This reduces the risk of filtering out persons with a less severe disability who may be hesitant to answer “Yes” if they only have some difficulty sometimes. It might also be argued that answer categories should be related to the intensity of difficulty (“No difficulty,” “Some difficulty,” “A lot of difficulty,” . . .). However, during qualitative testing, many respondents answered before the categories were even read out loud, and the answers tended to be “Well, sometimes.” Testing indicated that people think more of frequency than of intensity when asked these types of question.

To minimize the number of filters, each combines several disability types. For example, Filter 1 combines Seeing and Hearing. In qualitative testing, respondents did not hesitate in replying when the two concepts were together in a single filter question. When asked about this, most respondents indicated that “ideally, the two would be split,” but that they “hadn’t really had any difficulty.” Covering more than one type in a single question is not ideal, because someone who “sometimes” has difficulty with one type and “often” with the other may be unable to say precisely how often they have difficulty. However, the goal is to identify persons who do not have difficulty with any of the disability types covered by each question. If difficulty exists, its frequency does not matter at this point.

Previous versions of Filter 1 (about Seeing and Hearing) that were tested included the concept of communication, which was removed from the final version.^{Note 2}

Because the filter questions must be short, they are rather vague. For example, no mention is made of the use of aids for seeing and hearing. However, interviewers were instructed to clarify this if need be. The consequence of not being specific in the questions about the use of aids is that some people may report difficulty although they do not have any when using their aids. When they get to the screeners, this is resolved.

The DSQ do not have a specific filter to identify people with pain. Depending on the type and cause, people with pain may be captured as part of mobility, concentration, or even vision difficulties (for example, migraines). During qualitative testing, most people with pain said “ Yes” to one of the first four filters, while a few others said “Yes” to the other filter (the “catch-all” question).

Filter 6 is asked only of people who say “Yes” to Filter 5. In previous versions of the DSQ, Filter 5 asked about the presence of any other health problem or condition that limits daily activities. The double-barrelled nature of that question was problematic. Qualitative testing revealed that two separate questions were needed—one to ask about the presence of any other health problem or condition, and another to ask if it limits daily activities. Filter 6 was added to avoid filtering in too many people with minor health problems that do not limit their daily activities.

4.3.3. Screener questions

The DSQ screeners determine if a respondent has a disability and if so, the type(s) and severity. The screeners are presented for each disability type and in the order used in the questionnaire. They are numbered from 1 to 33 with the prefix S (for Screener) to ensure that they are not confused with the filter questions.

4.3.4. Developing the screener questions

The screener questions are structured in one of two ways, depending on the type of disability:

1. For seeing, hearing, mobility, flexibility and dexterity, one (or more) question(s) assess(es) the intensity of difficulty in doing certain tasks, followed by one question (if there is difficulty) about how often daily activities are limited by this difficulty.
2. For pain, learning, developmental, mental health-related and memory, one (or more) question(s) assess(es) the presence of a health problem or condition, followed by one question (if there is a problem or condition) about how often daily activities are limited by this problem or condition, and then followed by one question (if daily activities are limited) on the intensity of difficulty.

The reason for structuring the questions differently according to the disability type is that in the first case (seeing, hearing, mobility, flexibility, and dexterity), difficulty is assessed by asking about certain tasks. For the other types (pain, learning, developmental, mental health-related, and memory), the questions are not task-oriented, so it was not possible to ask about the amount of difficulty. A question was needed to assess whether the respondent had a certain health problem or condition, then another question was needed to verify if it limited their daily activities, and if so, another question was asked to assess the amount of difficulty.

This subtle distinction between how questions are asked depending on the disability type implies that for the second group, the question on intensity of difficulty was not asked of those who reported that their health problem or condition never limited their daily activities. This is not a problem for the DSQ, because the focus is on people whose activities are limited. However, this complicates analyses of the level of difficulty that people with a health problem or condition have, whether or not they are limited by it.^{Note 6}

The questions on frequency of activity limitation are similar from one disability type to another, but types differ in the assessment of difficulty or presence of a problem or condition. This is discussed in more detail for each set of screeners.

The answer categories for the activity limitation questions (“Never,” “Rarely,” “Sometimes,” “Often,” “Always”) differ from the categories used for the filters (“No,” “Sometimes,” “Often,” “Always”). “Rarely” was inserted between “Never” and “Sometimes” in the DSQ screening questions early in the process. It allowed greater precision and was useful for people who are hesitant to say that they are “Never” limited.

Seeing and hearing screeners

Seeing and hearing are the only two disability types that ask about ability rather than difficulty (although the response categories refer to the amount of difficulty). Testing revealed that the ability question was the most effective. To be consistent, a similar approach was used for hearing.

Questions combining the use of an aid and ability to see/hear were tested (for example, "Do you have difficulty seeing, even when wearing glasses or contact lenses?"), but respondents did not always give the answer that would be expected based on their situation. Such wording made the questions too long; respondents would sometimes remember only "even with your glasses or contact lenses" and would answer as if the question was "Do you wear glasses or contact lenses?"

Testing that used the contexts of seeing up close versus at a distance to try to capture level of difficulty indicated that seeing up close was not always well interpreted. For example, a respondent said that he did not need glasses to see up close, but he did need them to read. To avoid such confusion, seeing up close versus at a distance was not used in the final question wording.

The increasing use of cochlear implants in Canada^{Note 7} influenced the decision to refer to them in the hearing screeners. People who wear one know what it is, while others answer that they do not wear one, and it does not pose a problem.

Mobility screeners

The preamble to the mobility screeners mentions the ability to move around even when using an aid such as a cane. Other aids such as crutches or a walker were tested, but people misinterpreted the question thinking of such things as wheeling around and using elevators. For this reason, only the use of a cane is mentioned.

The mobility screeners comprise two separate questions to assess difficulty: one about difficulty walking, and the other, difficulty with stairs. Combining the two concepts in the same question did not work well. Also, context was needed for each task—respondents requested more precision, for example, how many stairs? “Climbing stairs” was changed to walking up or down a flight of stairs. Various iterations of distance for the walking question were tested; football fields (often used in other instruments) meant very little to most respondents; city blocks vary greatly; and measures of distance in both yards and metres are necessary because many older people conceptualize in yards. The most consistent measure was time, which worked well.

Flexibility and dexterity screeners

The flexibility screeners comprise two separate questions to assess difficulty: difficulty bending down and difficulty reaching in any direction. The dexterity screeners use only one question to assess difficulty: difficulty grasping small objects. The PALS questionnaire used seven questions to assess difficulty with agility (which corresponded to the combination of flexibility and dexterity). PALS data showed that the questions chosen for the DSQ captured the vast majority of people with agility issues.

A question long used to measure agility was dropped during qualitative testing of the DSQ—difficulty cutting one’s own toenails. This question aimed to measure flexibility and dexterity, but some people reported that they had diabetes and had been instructed not to do it owing to risks of cuts and bleeding. Others reported that because of their weight, they could not bend down far enough to cut their toenails.

For dexterity, a question about difficulty pressing buttons on an electronic device was tested, but functioned poorly, as it depended on the device, some having larger buttons than others. Also, the "pliers" example from the PALS was dropped. It was not something to which most people could relate, and was associated with gender.

Pain screeners

The preamble to the pain screeners explains that the questions are about pain due to a condition that has lasted or is expected to last for six months or more. Specifying pain due to a condition ensures that respondents will not report pain caused by a recent injury. Also, to say "pain that has lasted or is expected to last for six months or more" would be misleading, because respondents might think that they have to be in pain every day for that period.

Two questions assess the presence of pain: one asks about pain that is always present; the other asks about recurring periods of pain in order to capture cyclical or episodic pain. These two questions are similar to those used in the PALS, but the concept of "discomfort" used in the PALS was dropped for the DSQ. The variation of responses during qualitative testing suggested that “discomfort” is too vague to yield accurate data.

Another difference between the DSQ and the PALS is that in the latter, people who reported pain that is always present were not asked about recurring pain. The DSQ ask both. These questions were further improved in the short DSQ version ^{Note 8} (created after the 2012 CSD) with the use of dynamic text. ^{Note 9} This is explained in Section 9.

Learning screeners

Two questions determine if respondents have a learning disability. The first pertains to a condition that respondents think they might have and gives examples. The second asks if respondents were ever told by a teacher, doctor or other health care professional that they have a learning disability. The two questions are needed to take generational differences into account. As part of qualitative testing, people with a learning disability participated in focus groups, during which they were asked how they found out that they had a learning disability. Most “younger” participants had an assessment or diagnosis during their early school years. Older participants found out either by accident, or because of another event (for example, seeking help for depression, job loss, diagnosis of child exhibiting similar characteristics). A learning impairment is identified with a positive response to either of the two questions. The combination of these questions works as expected. Although the first uses medical terms, the question tested very well. Respondents are identified as having a learning disability only if it limits their daily activities.

Developmental screeners

One question determines the presence of a developmental disability—the same question that was used in the PALS. The question is designed to identify those with significant developmental conditions, the prevalence of which is fairly low among people aged 15 or older living outside of institutions. Despite the use of technical terms, the question was well understood in qualitative testing and functioned as expected.

The DSQ have two other screeners for developmental disability, one about the frequency of activity limitations, and the other about the intensity of difficulty due to the condition. However, reporting an activity limitation is not necessary to identify developmental disability. Developmental disabilities are, by definition, very limiting. Interviews for people with a developmental disability are often by proxy because they are often unable to answer on their own. This is the only disability type that does not require the reporting of activity limitations to identify disability.

Mental health-related screeners

The screening question used to assess the presence of a mental health-related condition is the same as the fourth filter question at the beginning of the DSQ, except for the answer categories. Other versions of the mental health-related filter were tested and did not work well: respondents would reply “No” to the filter, but would ultimately report a mental health-related condition to the screener. It seemed that the examples in the screener question (but not in older versions of the filter) helped identify them. Thus, to increase the chances that people would report their mental health-related condition, the screener was also used as a filter.

Several examples are provided with this question. They were taken from the PALS screeners, but slightly changed. Phobias were removed because the term tends to be used to refer to something that one really dislikes rather than something that provokes morbid fear or aversion. Schizophrenia was removed because respondents said that depression and anxiety did not belong in the same category. People with more serious mental illnesses would still see themselves in the list of examples, which contained the most prevalent mental health-related issues.

When this mental health-related screener is used as a filter, the answer categories are the same as those of other filters (“No,” “Sometimes,” “Often” and “Always”), but as a screener, the answer categories are “Yes” or “ No.” Also, if respondents gave a positive answer to the filter, the same question would not be asked again as a screener. Instead, the question on frequency of activity limitations would be asked, but starting with: “You mentioned earlier that you have an emotional, psychological or mental health condition . . . ." For people who said “No” to the mental health-related filter, the question is asked again as a screener. A certain number of respondents who said “No” to the filter said “Yes” to the screener.

The use of technical and medical terms in this question did not seem to pose a problem, regardless of respondents’ socioeconomic backgrounds. No complaints about wording or sensitivity emerged.

Memory screeners

The memory screeners start with a question about the presence of ongoing memory problems or periods of confusion; it specifies that occasional forgetfulness, such as not remembering where you put your keys, is excluded. This example helped to distinguish between real cognition difficulties and common memory lapses. Previous versions of the question included examples such as Alzheimer's and dementia, but respondents tended to focus on these examples and reported that they did not have a memory problem. The version of the question that was retained worked well in qualitative testing.

Other screeners

Like the “Other” filters (F5 and F6), the “Other” screeners (S32 and S33) are a catch-all category. Respondents are asked either filter 5 and 6 (if they said “No” to all previous filters), or the “Other” screeners, but not both.

For the Other screeners, the reference to six months’ duration is included to ensure that respondents do not report a recent injury from which they will recover in a short time.

The Other filter and screeners are used to define the Unknown type of disability. If respondents do not report any of the 10 disability types, but report that their activities are limited by another health problem or condition (to either the “Other” filter or screener), they are identified as having a disability of an unknown type.

4.4. Reliability of the DSQ

After the qualitative testing to ensure that the questions were valid, the questions were tested for reliability to see the extent to which they give consistent results.

The DSQ were used as a supplement to the Labour Force Survey (LFS) for September and October 2011, ^{Note 10} and in the Canadian Community Health Survey (CCHS) for the same months. This made it possible to examine disability rates by type of disability which take into account activity limitations and allowed a comparison of results for the same population in different contexts: an employment survey and a health survey. The results were analyzed recognizing that CCHS tends to yield higher disability rates,^{Note 11} regardless of the questions used.

The total number of respondents to the September and October LFS supplements was 13,500. The DSQ module was asked after the LFS interview was completed. For the CCHS, the number of respondents in September and October was 21,000. The DSQ were asked part way through the CCHS questionnaire, after questions on general health, mental health, self-reported height and weight, chronic conditions, pain and discomfort, health care utilization, flu shots, fruit and vegetable consumption, physical activities, neighbourhood environment, and sedentary activities. Because the CCHS already included questions on pain (from the HUI3 module), the DSQ pain screeners were not asked. Therefore, analysis of the pain questions from the CCHS is based on the HUI3 questions, not those from the DSQ. The Pain questions in the CCHS were asked well before the DSQ module and did not replace the pain questions in the DSQ module.  

For each survey, weights were produced to allow inference to the entire population aged 15 or older. The results from the LFS and CCHS quantitative tests in this report are all weighted, thereby taking differences in survey designs and respondents' characteristics into account. 

Two reports^{Note 12} with in-depth analyses of the results of these quantitative tests were prepared. Only summary results and main conclusions of these analyses are presented in this document.

4.4.1. Analysis of the quantitative tests

4.4.1.1. Filter-in rates

The purpose of the filter questions^{Note 13} is to ensure that people not likely to have a disability are filtered out of the DSQ in order to save interview time. These questions are not intended to measure disability. The LFS and CCHS tests were the first occasions to calculate filter-in rates for the DSQ. Chart 1 shows the overall filter-in rates for the LFS and CCHS, and filter-in rates by filter question. Filters 5 and 6 (the “Other” filters) were grouped together because it takes both to determine if the filter is positive.

As expected, overall filter-in rates and rates for each filter were higher for the CCHS than for the LFS (except for the catch-all category “Other” where rates are similar). Experience with the Census filter questions when used in general population surveys has shown that filter-in rates on a health survey tend to be higher than those of other surveys. The LFS filtered-out two-thirds of its respondents, while the CCHS filtered-out half of them. The main consequence is that the average CCHS interview time exceeded that of the LFS.

In Chart 2, the filter-in rates for each filter question in the LFS and the CCHS, by answer category (“Sometimes,” “Often,” and “Always”), show what the filter-in rate would have been if only “Often” and “Always” had been considered positive answers.  

Most of the difference in the filter-in rates between the two surveys is attributable to the softer answer (“Sometimes”); percentages reporting “Often” or “Always” having difficulty were very similar in the LFS and CCHS. The exception was the third filter on Cognitive disabilities, for which the percentage of “Often” responses on the CCHS was almost double that for the LFS.

Because these are filter questions and not screeners, a difference in filter-in rates affects only average interview time. What is most important to assess reliability is a comparison of disability rates, using the screener questions on both surveys.

4.4.1.2. Disability rates

According to the social model of disability, an individual has a disability if daily activities are limited because of a long-term condition or health problem. The DSQ include questions on the frequency of daily activity limitations for each disability type, with the answer categories “Never,” “Rarely,” “Sometimes,” “Often” or “Always.” Because “Rarely” was a new answer category compared with previous disability instruments, it was not clear if it should be included in the definition of disability. The reason for introducing this category was to set apart people with very minor problems and ensure that “Sometimes” would be reserved for more significant issues.  

Several definitions of disability were compared with the data from the quantitative tests. The only constant in all definitions is in developmental disability, where receiving a diagnosis alone is enough to be considered as having a disability.

Because the pain screeners on the CCHS came from the HUI3, not the DSQ, an attempt was made for this analysis to mimic the DSQ definitions of a pain disability using the HUI3 questions. The HUI3 pain definitions used for this comparison were based on two questions, one about intensity of the pain (“Mild,” “Moderate” or “Severe”), and the other about the number of activities that were limited by the pain (“None,” “Some,” “Several” or “Most”). Since the HUI3 pain questions were asked before the DSQ, for this comparison, if respondents said “No” to all DSQ filters, their disability status (and that for each type) would be set to “No,” even if the HUI3 pain questions indicated otherwise. This was done for consistency with other disability types, which are set to “No” when all filters are negative.

Disability Definition #1: Include all “Rarely” answers

For this definition, if daily activities were at least “Rarely” limited, this counted as a disability (except for Developmental). Table 2A illustrates this first definition. For the CCHS, the pain definition comprised people for whom “Some,” “Several” or “Most” activities were limited.

Disability Definition #2: Exclude all “Rarely” answers

For this definition, if daily activities were “Sometimes,” “Often” or “Always” limited, this counted as a disability (except for Developmental). Table 2B illustrates this second definition. For the CCHS, the pain definition comprised people for whom “Several” or “Most” activities were limited.

Disability Definition #3: Include some “Rarely” answers

This definition falls between definitions 1 and 2. If respondents reported that their activities were “Rarely” limited, the intensity of difficulty was taken into account; if it was “A lot of difficulty” or “Cannot do it at all,” this counted as a disability. If it was “No difficulty” or “Some difficulty,” it was not considered a disability. Table 2C illustrates this third definition. Again, the definition for Developmental remained the same. To mimic this definition for the pain questions in the CCHS, respondents for whom “Several” or “Most” activities were limited, and those for whom “ Some” activities were limited and who also reported that the pain was “Moderate” or “Severe,” were included.

This third definition could not be used for the Other type, because no question was asked about the intensity of difficulty for that type. Therefore, only respondents who reported that their activities were “Sometimes,” “Often” or “Always” limited were included (as in definition 2).

Comparisons of the 3 definitions

Table 3 shows disability rates for each of the 10 disability types according to the three definitions, based on the LFS and CCHS quantitative test data. Except for the pain questions, which differed on the two surveys, both used the DSQ and data were collected in September and October of 2011.

Table 3 also shows disability rates from the 2006 PALS (15 or older rates excluding the territories and collective dwellings ^{Note 14} ). When the DSQ quantitative tests were conducted, the 2006 PALS was the most recent benchmark survey to measure disability at Statistics Canada. The 2006 PALS results are not directly comparable to those of the LFS and CCHS: they used a different set of questions; the PALS was administered to a pre-filtered population; and data were collected in 2006. The PALS rates are presented to help understand the impact of the DSQ. The PALS rates were however age-standardized to account for aging of the population.

For both the LFS and CCHS, a large difference between the overall disability rates based on the first and second definitions is apparent: about 7 and 10 percentage points, respectively. Examination of the data revealed that a substantial number of respondents indicated that they had “Some” difficulty with a particular task, which “Rarely” limited their daily activities; it is unlikely that these people actually had a disability. For this reason, from the outset, the second definition (which excludes “Rarely” answers) was preferred over the first one.

However, some respondents who reported being “Rarely” limited, also reported having “A lot of difficulty” or “ Cannot do at all.” These people likely had cyclical or episodic disability which, when experienced, had a strong impact. In the LFS data, this group is proportionally larger among those with mental health-related disabilities, pain disabilities, memory disabilities, and learning disabilities. Many underlying conditions for pain disabilities and mental health-related disabilities are cyclical—for example, migraines, depression, anxiety, etc. As well, some people may always have an underlying condition that limits them only in certain environments. For example, people with learning disabilities who are no longer in school or other learning-focused situations may only be limited in certain situations. These people are included among the population with disabilities, which is why the third definition is proposed. In some DSQ documentation and in this report, the third definition is called the RACXR definition (R for “Rarely,” A for “A lot,” C for “Cannot,” and XR for exclude “Rarely” for the type “Other”).

Based on the third definition, the disability rate was 22.8% for the LFS, and 27.3% for the CCHS. Disability rates by type for this definition were similar for the LFS and CCHS, except for pain because of the use of different questions.

The results from the LFS and CCHS tests were compared with those of the 2006 PALS to determine if the differences were as expected. It was anticipated that DSQ rates would be lower for some disability types because of the requirement for activity limitations, while for other types, DSQ rates would be higher because the population was not pre-filtered by the Census questions. 

The rates of most disability types are close (Chart 3). Flexibility and Dexterity for the DSQ were combined into Agility to allow comparison with the PALS.

Mobility

The prevalence rates of mobility disability in the LFS and CCHS (8.3% and 8.0%, respectively) are much lower than in the PALS (11.5%), mainly because the PALS did not require activity limitations to identify mobility disability. If the requirement for activity limitations is removed from the DSQ definition, the prevalence of having at least some difficulty with a mobility task is 12.1% and 12.9%, respectively, for the LFS and CCHS.

Pain

The prevalence rate for pain from the DSQ (only the LFS used the DSQ for pain) is higher than that from the PALS. This may be surprising because the questions are very similar, and the PALS required activity limitations to identify pain disabilities. The main reason for the difference is that the PALS was administered to a pre-filtered population based on the 2006 Census filter questions. The Census questions are not effective in filtering-in people with less visible disabilities. Consequently, people whose pain affects only their cognitive abilities (difficulty concentrating, for example) may not have been covered by the PALS. This could also be the case for people who experience migraines.

Learning

As anticipated, the learning disability rate from the DSQ is higher than that from the PALS. The PALS pre-filtered population tended to exclude people with non-physical disabilities. The PALS disability definition for learning did not require activity limitations; if it had, the PALS rate would have been even lower.

Mental health-related

The DSQ rate for mental health-related disability is much higher than the PALS rate. Although the PALS required activity limitations for this disability type, it relied on a pre-filtered population, and the filters are inefficient in capturing people with non-physical disabilities.

4.4.2. The DSQ on the Canadian Survey on Disability

Based on the analysis of the quantitative tests results, the DSQ module was considered ready for use in the 2012 Canadian Survey on Disability (CSD).

The CSD underwent qualitative testing that included the DSQ. To ensure that the choice of the DSQ over the PALS screeners was the right one, the PALS screeners were also tested for some interviews. The DSQ flow was superior; the DSQ were clearer and easier to understand, and they raised few queries. Only minor wording changes were made to the DSQ after this qualitative testing.

Including the DSQ on the CSD was an improvement over the PALS screeners, but the CSD suffered from the same drawback: it used a population that was pre-filtered based on the Census filter questions that were put on the National Household Survey (NHS).^{Note 15} Consequently, people with non-physical disabilities would continue to be underrepresented. For this reason, although the DSQ would be used on the CSD, the DSQ filter questions would be “turned off”; that is, everyone would be asked the screeners regardless of their responses to the DSQ filters. Given that the population was already pre-filtered, it was not necessary to filter it again.

During collection, the CSD used the most inclusive definition of disability to identify people who would go through the entire survey questionnaire. Hence, respondents who reported being at least “Rarely” limited in their daily activities (definition 1 described previously) were flagged to continue with the full CSD interview. During data processing, further analysis using the different definitions of disability facilitated a decision about whether people who are “Rarely” limited should be considered persons with a disability. Based on the CSD data and feedback from interviewers,^{Note 16} the third definition (RACXR) was chosen—people who were “Rarely” limited were considered to have a disability only if they also said they had “A lot of difficulty” or “Cannot do at all.”  

Chart 4, which displays disability rates from the 2011 LFS and the 2012 CSD based on the third definition, shows the impact of administering the DSQ to a pre-filtered population using the NHS filter questions, as well as the impact of doing a post-censal survey 18 months after the NHS. It is likely that some persons who had a disability at the time of the NHS no longer had it when they were interviewed for the CSD, due to surgery, medication or other reasons.

The CCHS results are not shown^{Note 17} because the pain questions were from the HUI3 and not the DSQ.    

The consequence of working with a population that was pre-filtered by the NHS filter questions is substantial. These questions tend to exclude people with very mild disabilities of any type, and those with non-physical disabilities. Hence, for all disability types, the CSD rates are lower than the LFS rates. The disability types most affected by pre-filtering (highest relative differences between LFS and CSD) are pain, mental health-related, and learning.

This comparison demonstrates the need for better filter questions on the next NHS, a topic that will be discussed in Section 7. 

4.5. Official definition of disability using the long DSQ^{Note 18}

Analysis of the entire CSD content made it possible to select the proper definition of disability for use with the DSQ. As mentioned previously, for each disability type, the DSQ include at least one question on the associated level of difficulty (“No difficulty,” “Some difficulty,” “A lot of difficulty,” or “Cannot do”) and a question on the frequency of the limitation of daily activities (“Never,” “Rarely,” “Sometimes,” “Often,” or “Always”). To consider someone as having a disability for a given type, the frequency of limitation of daily activities must be “Sometimes,” “Often” or “Always,” or it must be “Rarely” combined with a difficulty level of “A lot” or “Cannot do” (Table 4). The exception is Developmental disability, for which a diagnosis is sufficient to be considered as having a disability. Also, for the Other type, disability requires being limited “Sometimes,” “Often” or “Always.” This corresponds to the third definition (Section 4.4.1.2).

4.6. Length of the DSQ

One of the goals was to have a DSQ module that would take an average of 2 minutes to administer. The "Audit Trail" files from the LFS and CCHS quantitative tests showed that the average time to administer the DSQ on the LFS was around 3 minutes, and on the CCHS, closer to 4 minutes. As noted, the CCHS had a much higher filter-in rate because respondents were more likely to report difficulties to the DSQ filter questions, which increased the average interview time.

Three to four minutes may be too long for inclusion on general population surveys, which prefer to devote more interview time to specific subject matter. As a result, a shorter version of the DSQ was developed. The short DSQ version averages less than two minutes per respondent and identifies five broad disability types: Seeing, Hearing, Physical (combining Mobility, Flexibility, Dexterity, and Pain), Cognitive (combining Learning, Developmental, and Memory), and Mental health-related, plus an Unknown category. Because the short version combines some disability types, it does not allow calculation of a severity score.

The short DSQ version is described in Section 6.

• Previous
• Next