The answer to the first question posed in this blog is easy. No, the WG-SS does not identify all people with disabilities. First, there is no gold standard to identify all people with disabilities for all purposes. In a previous blog we discussed, for example, the difference between identifying people with disabilities for monitoring the SDGs versus determining eligibility for specific programs.
But even for the purpose that the WG-SS were developed – to identify people at risk of exclusion because of activity limitations – we know the WG-SS does not capture everyone. To identify all people would require many more than six questions. However, we are limited in the number of questions we can ask, especially on a census where space is very tight.
Space is also an issue on surveys. To disaggregate the SDGs we would like to see all household surveys used for SDG indicator production have the WG-SS on them. The more questions we ask to be put on a survey, the less chance that they will be included because of cost and time constraints. So we must consider diminishing returns. How much extra value will be added for the purpose the WG-SS is intended by adding additional questions?
The Washington Group Extended Set of questions was designed to capture those persons at risk for exclusion that might be missed by the WG-SS. In the table below using data from the US, you can see how adding questions from the WG-ES on upper body difficulties and psycho-social difficulties, two domains omitted from the short set, affects prevalence. Adding these domains does increase the number of people identified, but at the cost of additional questions. For example, in the US, the WG-SS yields a 9.5% prevalence rate, which increases only slightly to 10,0% if the two upper body questions are added from the extended set. It is believed, however, that more people with psycho-social disabilities will be missed by the WG-SS. Some are identified through the communication, cognition, or self-care question, but not all. In the US, 3.6% of the population is identified as having a disability based on the anxiety and depression questions, but 2.1% of the population so identified were also identified by at least one of the WG-SS questions.
That means that using only the WG-SS will miss 1.5% of the population that has a psychosocial disability. Of course, only using the WG-SS will mean that the remaining 2.1% of the population that have psychosocial disabilities will not be identified as having that type of disability, but most likely by the cognitive or communication domain.
| Questions | Number of Questions | Prevalence |
|---|---|---|
| WG-SS | 6 | 9.5 |
| WG-SS plus Upper Body | 8 | 10.0 |
| WG-SS plus Upper Body plus Anxiety and Depression | 12 | 11.9 |
Source: NHIS/NCHS (Download a powerpoint presentation on these results)
Clearly if room on a survey or census allows all 12 of these questions to be asked, more people with disabilities will be identified, and the resulting analysis will be richer. But how important is it to do so?
To a large extent, it depends on the purpose of using the questions. Recall the main purpose of the WG-SS is to be able to disaggregate indicators by disability status to see if outcomes for people with disabilities are different from those without disabilities – in other words, the disability gap. Statistically speaking, it is not necessary to identify all such people – only the large majority, in order to pick up associations between disability and various outcomes in the data. Leaving out 2.4% of people not identified by the WG-SS but with upper body or psycho-social difficulties will have a small impact on measuring the disability gap. And it is true, that if that 2.4% also faces barriers, then there will be a negligible underestimation in that gap. This is because even if the people with disabilities missed by the WG-SS faced similar barriers and so included in the non-disabled population, there numbers would be so swamped by the people without disabilities, as to make any underestimation of the gap barely register. And importantly, leaving out that 2.4% will not have a significant impact on measuring trends in the disability gap. Because whatever that initial underestimation would be, it is most likely going to be constant across time so that the changes in the measured disability gap will not be affected.
Nevertheless, if a country is not willing to live with the small underestimation of the disability prevalence and the possible impact on the disability gap – and feels it is very important to examine the population of people with anxiety and depression difficulties – the questions from the WG-ES are there to use. Only they come at a cost of space on the survey. Whether that is worth it depends on the priorities of the data gatherers and the resources they have at their disposal.