Chapter 6. Designing the Instruments
We now turn to the description of the development and piloting of a disability survey for Ireland. Carrying out a disability prevalence survey typically involves two stages: the first “screens” the population or large sample to identify a sub-set affected by disability, which the second stage then investigates in depth. We first provide in this section a brief background to disability measurement internationally and to the advantages and disadvantages of Census versus survey approaches. We then focus in the next section on the “screening” stage, before moving on in the following section to the in-depth survey element of the Irish pilot exercise.
6.2 Disability Indicators
Gudex and La Fortune (2000) provide a concise history of disability measurement by describing disability indicators in use since the early 1960’s and their purpose. One of the early disability indicators measuring activity limitation was the ‘ADL’: a scale of activities of daily living developed by Katz et al (1963) aimed at assessing people with severe disabilities. The ADL measured independence in carrying out such activities as bathing, dressing and eating. In 1969, Later, Lawton & Brody devised a more encompassing scale of activities termed ‘IADL’, Instrumental Activities of Daily Living, which included among the list of activities for example the management of finances and housework.
Generic health indices (SF-36, EuroQol-5Dand the McMaster’s Health Utility Index (HUI-3) are a more recent phenomenon and serve to summarise health and activity limitations using a scoring method. According to Gudex and La Fortune these type of measurements are being used increasingly in national surveys to assess the prevalence of disability.
In the 1980’s, the OECD developed 25 questions on Activities of Daily Living (ADL’s)in order to screen respondents with disabilities. Attempts to include all 25 questions on most social survey questionnaires proved difficult. Canada provides an example of this. In 1986 HALS used 25 questions in its disability screen. The impossibility of including these on its census screen prompted the development of a disability filter question known as a ‘global disability indicator’, Langlois (2001:3). Its purpose was the identification of the target population through a set of reduced questions.
According to Langlois (2001) global disability indicators serve two purposes 1) descriptive and 2) screening:
Descriptive: to provide general information a global measure on the functional status of a given population through inclusion in social surveys.
Screening: as a screening mechanism to identify target population for a follow up study. Here it is important that the indicator be as comprehensive as possible so as not to exclude any member of the target population. It should be overly broad and exclusions can be made later to eliminate respondents who do not possess the characteristics of interest.
The number of countries which now include disability questions on censuses has increased from 20 in 1970 to around 80 in 1990 ABS (2003). In the past measurement of disability through this method covered only those with severe impairments. Key points which merit consideration when attempting to choose between the collection of disability data in a census versus a sample survey include the following:
Advantages of Census
- Data can be tabulated for small local areas.
- Prevalence rates can be calculated for small geographical areas because data are also gathered for the population at risk.
- Household data (living arrangements, employment and education) can be tabulated for persons with and without disabilities to measure different levels of participation for example at home, school or work.
- Descriptive statistics are not subject to sampling errors
- Can provide a useful sampling frame for follow up surveys on people with disabilities who are otherwise difficult to find.
Disadvantages of Census
- Limited range of questions are possible to capture adequately the situation of people with disabilities.
- It is difficult to measure disability with a binary classification yes/no. Rather it is recommended to measure disability on a continuum which would consider several issues such as intensity, duration and environment. This is particularly difficult when the number of questions is restricted, as they generally are in the limited space available on a national census form, Mbogoni and Me (2000)
- The use of short questionnaires tends to underestimate prevalence among special population groups such as children and elderly people.
- Data collection is infrequent usually every 4 years
- Potentially subject to high non-response rates and under enumeration because of the complexity and sensitivity of the question.
- Censuses have an extensive use of proxy respondents.
Advantages of Sample Surveys
- Flexibility in the depth and range of topics that can be covered.
- Easy to initiate given the availability of a sampling frame
- There can be greater control over the conditions of observation and the interview because of limited coverage geographically and a smaller number of interviews to be completed
Disadvantages of Sample Surveys
- Sample size is limited unless the survey is very large – the size of the population with disabilities identified is likely to be especially small. This means that although prevalence rates can be measured with relative ease the generally low incidence rates mean that the absolute number of cases identified in any sample as experiencing a disability or difficulty is relatively small and so analysis of the characteristics of the sub-population in question becomes problematic from a statistical perspective.
- A special case of the above is the limited ability to analyse prevalence rates at a regional or county level because of a limited sample size and subsequent sampling errors associated with spatial disaggregation of the data.