Response Categories

The four WG-SS response categories − “cannot do at all,” “with a lot of difficulty,” “with some difficulty,” and “with no difficulty” − must be used. Substituting yes/no response categories undermines the validity of the questions. First, human functioning is not a dichotomous variable. A range of difficulties (or disabilities) exists. Collecting data using yes/no response categories fails to allow for analysis of the continuum of functioning. Second, some respondents who answer “yes” may only have minor difficulties that do not rise to the level generally considered to be disabling and thus may generate false positives. Third, where people draw a line between “yes” and “no” may vary based on their cultural and socio-economic backgrounds. This is less true for the four category responses because the gradations between response categories are smaller.

Box 15. Question wording

It is recommended that special attention should be given when designing census or survey questions to measure disability. The wording and the construct of questions greatly affect the precision in identifying persons with disabilities. Each domain should be asked through a separate question. The language used should be clear, unambiguous and simple. Negative terms should always be avoided. The questions on disability should be addressed to each single household member and general questions on the presence of persons with disabilities in the household should be avoided. If necessary, a proxy respondent can be used to report for the family member who is incapacitated. What is important is to account for each family member individually rather than ask a blanket question. Scaled response categories can also improve the reporting of disability.

Principles and Recommendations for Population and Housing Censuses, Revision 3, (ST/ESA/STAT/SER.M/67/Rev.3), 2015, para 4.207.

 

However, a number of countries have opted for the “yes/no” approach rather than the four response categories. This is because they believe this approach takes less time to administer. Cognitive testing of the WG questions, however, suggests this is not the case, because respondents take longer to figure out if they should answer yes or no if they have more mild difficulties.