Disability Data Measures project

• The Disability Royal Commission emphasised a need for a consistent way to collect data about disability across surveys and mainstream services (Recommendation 12.5).
• The Disability Data Measures project started looking into how to achieve this consistency in Australian surveys.
• Led by the Australian Bureau of Statistics with support from the Department of Health, Disability and Ageing, the project aimed to find a simple, easy-to-use question or set of questions that could collect data comparable to existing disability measures.
• Findings showed that using a well-established conceptual approach and minimum question set can produce detailed and nationally consistent disability data.
• Further consultation, refining of questions and testing are needed to guide the future implementation of a consistent approach to collecting disability data.

The Australian Bureau of Statistics (ABS) led the Disability Data Measures (DDM) project with support from the Department of Health, Disability and Ageing. The DDM project addresses Recommendation 12.5 of the Disability Royal Commission (DRC):

'Recommendation 12.5

The Australian Government and state and territory governments, through the Disability Reform Ministerial Council, should address the lack of consistent disability data by developing a nationally consistent approach to collecting disability information.

By December 2024, the Australian Government and state and territory governments should agree to a core set of questions to identify disability status to be used across all mainstream services and population surveys. This should be led by the Australian Bureau of Statistics and the Australian Institute for Health and Welfare. The questions should be co-designed with people with disability and their representative organisations, and with First Nations subject matter experts.'

The project aimed to find and suggest a consistent way to collect disability information in population surveys across Australia.

The DDM project partially addressed Recommendation 12.5 by focusing on disability measures for population surveys only, not for mainstream services. Recommendations 12.6 (addressing the lack of disability data) and 12.7 (a strategy to extend disability data collection) from the DRC were considered where possible but were mostly outside the project’s scope. Refer to Appendix - Disability Royal Commission recommendations for more information.

The scope of the DDM project included:

• conducting a review of current disability question sets including disability identification and disability status in the ABS Survey of Disability, Ageing and Carers (SDAC) and the ABS Short Disability Module (SDM)
• consulting with the disability community (for example, Disability Representative Organisations and people with disability), researchers and relevant Commonwealth, state and territory agencies on contemporary best practice for collecting disability information and any recommendations for improvements and better alignment across data sources
• undertaking testing with people with disability and those without disability of question sets proposed for data collection, to understand alignment with existing data sources and impacts of any recommended changes
• developing a proposed short standard set of questions for the collection and dissemination of data on disability status in population surveys.

This paper presents findings from each component of the project.

The ABS conducted a review of question sets between July and November 2024. This aimed to identify different approaches taken to measuring disability, to inform the scope for testing disability question sets. 

A range of question sets were identified within Australia and internationally. These can be categorised as being lengthy and complex (for example, SDAC), of medium length and complexity (for example, the SDM) or short (1-3 questions). Different question sets provide differing levels of information about disability according to:

• their intended use (for example, to determine prevalence of disability; for administrative purposes)
• the underlying concepts used (for example, functioning or social model)
• operational survey requirements (for example, the need to minimise respondent burden or to minimise complexity of data management).

The review found that many of the question sets used in Australia and internationally are guided by the ICF. The ICF is a framework for describing and organising information on individuals’ functioning and disability and is the conceptual basis for both the SDAC and the SDM. Concepts and wording used in the SDAC and SDM are included in many of the single or shorter question sets used in Australia. For example, many question sets refer to similar or identical lists of conditions to those used in the SDM. 

More comprehensive measures of disability require more questions. This can increase respondent burden and may require more involved data collection, processing and coding. However, simple measures may need to combine several different concepts into a single question, which can also increase respondent burden.

An example of a relatively short question set is the Washington Group Short Set on Functioning which uses 6 questions, underpinned by the ICF, to determine whether a person has disability and the level of difficulty a person has with specific tasks. The Australian Public Service Employee Census identifies disability using a single question, underpinned by ICF concepts.

The information-gathering phase of the project included consultation with disability stakeholders, to understand stakeholders’ current data requirements and their use of disability data, to inform recommendations for testing of disability question sets. 

The ABS circulated a consultation paper in August 2024 to disability stakeholders and held virtual consultation sessions for stakeholders to provide feedback on the content of the paper. The ABS also invited stakeholders to provide written feedback. Plain language and Easy Read versions of the paper were included.

The consultation paper included the following questions for stakeholders to consider:

1. How are disability data used in your organisation?
2. Do you collect any disability data and if so, what tools do you use?
3. What data outputs (items) are produced from your collection?
4. What level of detail do you require to support your work? For example, do you need to know about whether a person has disability only or do you need more information about the type of disability they may have, and/or the level of help they require with everyday activities?
5. What are your greatest challenges working with disability data? For example, implementation in your collections, data quality issues etc.
6. Do you have any suggestions for improving information captured for cohorts that are currently under-represented?

Below is a summary of feedback for each of the six questions in the consultation paper.

Question 1. How are disability data used in your organisation?

All stakeholders recognised people with disability as a priority population group. In addition to whether a person has disability, stakeholders use data on disability status, type of disability and type of assistance for policy and strategy development, estimating service level needs, prioritising funding allocations, meeting reporting obligations and assessing program outcomes for people with disability. 

Some stakeholders outlined analytical studies which used disability data and associated modelling methods, while others mainly used data to build evidence about the lived experiences and outcomes for people with disability to assist with advocacy work, public education, community profiles and funding submissions. 

Many stakeholders reported that they frequently use ABS data (for example, SDAC, the Census and the Person Level Integrated Data Asset) as well as data from the National Disability Insurance Agency, the Australian Government Department of Social Services and the Australian Institute of Health and Welfare (AIHW). 

Question 2. Do you collect any disability data and if so, what tools do you use? and Question 3. What data outputs are produced from your collection?

Stakeholders provided feedback on data measures, questions and tools they use to collect their own disability data.

For example, the Australia's Disability Strategy Survey provides data on community attitudes towards people with disability as part of Australia’s Disability Strategy 2021–2031. The survey includes a subset of questions from the ABS SDM, as well as separate questions on type of disability/long-term condition (for example, intellectual, neurological and physical) and self-perceived severity of disability.

Several areas of the AIHW collect and report data on disability related to housing and homelessness, prisoner health and aged care. Much of this administrative data is sourced through service use and eligibility criteria on client forms such as those used in hospital admissions and school enrolments. 

Other question sets stakeholders use include a modified Washington Group set and the Parkville Disability Identifier (which allows consumers and carers to inform clinicians about their individual disability and support needs).

Question 4. What level of detail do you require to support your work?

This question elicited the most discussion with stakeholders. The table below summarises the key themes to emerge, listed in order of relative priority.

Question 5. What are your greatest challenges working with disability data?

In addition to Recommendation 12.5, the DRC recommended further improvements to disability data in Australia (Recommendations 12.6 and 12.7) such as improving data to allow for greater intersectional analysis. While directly addressing these recommendations is out of scope for the DDM project, Questions 5 and 6 of the consultation paper were included to seek stakeholders’ input with a view to potentially informing future data development work such as improving data collection for priority populations.

Challenges working with disability data reported by stakeholders included:

• Access to data, datasets not being available or integrated in a single platform, or being available in separate systems with different access arrangements (including costs).
• Datasets having different definitions and concepts for measuring disability (for example, functioning vs social vs health/medical domains). This can also be due to the data collection being shaped by the core information needed.
• Type of disability and how particular conditions fit into disability types.
• Challenges related to lack of detail - datasets often have sample sizes that do not allow for detailed breakdowns such as small areas (for example, data only being available at the national level) or for populations of interest (for example, First Nations peoples, CALD, LGBTIQA+ communities). There is a trade-off between depth of detail in a comprehensive survey such as SDAC (which provides limited data for small populations) versus limited detail in Census (data on need for assistance with core activities only) but which can provide data for small populations.
• The need to better understand diversity and intersectionality, making detailed data for a wide range of population and socio-economic characteristics highly desirable. For example, disability amongst First Nations peoples in aged care settings; the intersection between disability and domains such as housing, justice, early education, regionality and remoteness, gender, sexual orientation, culture, socio-economic resources, children and youth, and hospitalisations.
• Inconsistencies in measuring disability across the life course or in different settings can make it difficult to interpret data; for example, across the transition between schooling, higher education and employment, or between disability and aged care services.
• Data on intellectual disability was noted as a particular challenge, as this is a specific type of disability separate to cognitive impairment, which is lost in the broader psychosocial disability group in existing data collections.

Question 6. Do you have any suggestions for improving information captured for cohorts that are currently under-represented?

Stakeholders provided feedback about preferred language and terminology, stressing the importance of appropriate wording of questions for people with disability. Providing easy to understand information on the reasons for collecting these data, their use, and providing means for results to be accessed by people with disability was also considered a priority. Stakeholders advised that terminology such as ‘harder to reach’ or ‘vulnerable’ cohorts of people with disability should be avoided. 

Additionally, stakeholders recommended seeking feedback on preferred identity language from the perspective of people with disability (for example, person-first and identity-first language). Allowing for and respecting how people with disability choose to identify was considered to be an important aspect of collection and reporting on disability.

Stakeholders noted that people may or may not disclose their disability based on how safe they feel it is to answer the questions. They may also have a different understanding of the intent of the questions, so the language used is very important.

Inconsistency in terminology was raised. Terms such as disability status, level or severity are often used interchangeably, and similarly, a range of words are often used to describe type of disability (for example, type, category, cause, group). It would be valuable to have consistent and agreed terminology.

The scope of the testing phase of the project was based on the review of disability questions and feedback from consultations.

Alignment with disability data in existing collections such as the ABS SDM and SDAC was a consistent theme amongst stakeholders. As the ICF underpins both the SDAC and the SDM, most of the questions in scope for testing were based on ICF concepts. 

However, feedback from stakeholders also noted the importance of self-identification of disability so this concept was included in the testing.

The final parameters for the scope of testing of question sets explored three key measures:

1. Whether a person has disability

   The ABS tested a single question to ask respondents directly whether they have disability, to allow self-identification. To understand how this question measures disability, this was tested in conjunction with the SDM-based measure of ‘whether has disability’ (in 2. below) for comparison purposes.

2. Whether a person has disability and disability group

   Questions which measure whether a person has disability and disability group were included in testing. Improvements to existing language and terminology used to describe the types of conditions presented to respondents were also part of the testing.

3. Disability status

   Questions that measure disability status (that is, the extent to which a person is restricted in everyday activities) which give an indication of the levels of support people with disability may require were tested.

The aims were to test:

• If a single question could be used to measure whether a person has disability, and if the results would be comparable to those from existing SDM questions.
• A shorter set of updated SDM questions to measure 'whether has disability' and 'disability group/type'. The language and terms were revised to ensure they would be contemporary, appropriate and clearly understood. In this paper this set of questions is referred to as a ‘minimum question set’.
• Whether some conditions included in the existing SDM that are assumed to restrict people in everyday activities are considered restrictive by respondents.
• A revised/reduced set of questions to measure disability status.
• Suitability of the set of disability questions for individuals to answer about themselves and answer as a proxy (that is, a person who answers survey questions on behalf of another person).

Testing of questions among people with different demographic characteristics was an additional testing priority.

Cognitive interviews were conducted over five rounds from February to May 2025. ABS staff who are trained in cognitive interviewing techniques and methods recruited participants, developed protocols, conducted the interviews and analysed and summarised the results. Participants included people with and without disability with a broad cross-section of demographics, such as age, gender, country of birth and First Nations peoples.

Cognitive interviewing is a qualitative method of evaluation that is used to determine whether survey questions are working in the way they were intended (and if not, the reasons for this). It involves an in-depth interview conducted by trained and experienced interviewers to identify where respondents may have difficulties with questions or interpret questions differently than what is intended. The respondent is asked to ‘think aloud’ as they are interviewed, and the interviewer uses probes where necessary. This provides further information about an answer, to determine how the respondent arrived at an answer, or to investigate the respondent’s interpretation of a question or term. These techniques assess respondent understanding of the questions and concepts, and how easy or difficult it is to answer the questions. Cognitive interviews are conducted in person or by video interview to enable behaviours and reactions to be observed. 

Testing of proposed question sets was conducted as cognitive interviews to:

• determine whether concepts were well and consistently understood
• determine whether questions were clear and measured what they were intended to measure
• assess participant sensitivity to language and terminology used
• identify and refine questions that were cognitively burdensome.

Questions evolved over the course of testing, based on results from each successive round of testing. Changes to the questions were made between rounds, as well as mid-round if it became clear that all participants were misunderstanding a word or phrase. The order of questions was also adjusted to minimise any context effects on the results of testing.

Testing did not include usability testing or testing for different modes of collection. How a question is laid out in a survey, whether online or on paper, is important to how a person interprets and then answers the question. Usability testing might include whether prompts are needed or how extra information is displayed. This testing was not conducted but should be, depending on the final mode of collection.

Participants were reimbursed for out-of-pocket expenses for their participation.

The outcomes from the five rounds of testing are summarised below for each key measure identified in scope for testing:

• whether has disability
• disability group
• disability status.

Single question

Different versions of a single question to measure ‘whether has disability’ were tested. The single question was initially based on questions identified in the review of disability questions. It was then revised using the findings from each round of cognitive interviewing.

Questions that used direct self-identification and questions that used ICF concepts as a conceptual basis were tested. Different versions of these single questions explored respondents' understanding and reactions to different terms such as ‘disability’ and ‘long-term’. 

All respondents indicated that a single question was easy to answer, and no negative reactions were observed. However, versions of the single question that directly asked about ‘disability’ were sometimes interpreted differently to ICF-based measures of ‘whether has disability’. The direct question about self-identified disability generally resulted in fewer people being found to have disability compared to responses to the SDM-based measure; that is, some respondents reported having restricting conditions in the SDM-based questions but did not consider themselves to have disability using the self-identification question. 

Alternative single questions that measure ‘whether has disability’ based on ICF concepts were also tested. Results indicated that a single question that measures disability based on the ICF (that is, using wording such as ‘a condition, limitation or impairment that restricts everyday activities’) gave comparable results to the existing SDM-based measure of ‘whether has disability’.

Testing confirmed that terms such as ‘long-term’, ‘ongoing’ or ‘permanent’ were well-understood by respondents when answering the different versions of the question, even if a definition about the duration of a condition (such as ‘lasted or is expected to last for 6 months or more’) was not provided.

While single questions to measure ‘whether has disability’ were received positively by respondents, the single self-identification question gave inconsistent results. The single ICF-based question gave more comparable results to the SDM-based measure, but further testing is needed to determine if a single question can produce results that are conceptually consistent with existing sources.

A measure of ‘whether has disability’ was tested using different versions of multiple questions based on the SDM. These questions consider a person to have disability if they report having a condition that has lasted or is expected to last for six months or more that restricts them in everyday activities.

These questions present two lists of conditions, limitations and restrictions. Respondents who select having a condition in the first list (refer to example question 1 below) are then asked if they are restricted in everyday activities by that condition. However, respondents who select having a condition in the second list (refer to example question 4) are assumed to be restricted by their condition, and no additional question is asked. A question (refer to example question 5) that asked if participants were restricted by conditions presented in the second list of conditions was therefore tested.

Testing showed that a minimum question set based on the SDM (refer to example questions 1 to 4 in 'Example of a minimum question set tested' below) can be used to effectively measure ‘Whether has disability’. These questions can also provide a measure of ‘Disability group’ discussed in the next section. 

Testing of example question 5 showed that some people did not consider conditions in example question 4 to restrict them in everyday activities. 

However, testing also found that some people who say they are not restricted by a condition will sometimes go on to talk about having difficulties or requiring help or supervision with core activities when asked further questions. This suggests people may interpret concepts such as ‘restricted in everyday activities’ differently. This may be due to people with a condition making adjustments to their life such that they may not consider themselves restricted in activities. For example, because of the help they receive (from people, aids or technology). 

The assumption that conditions in example question 4 restrict everyday activities needs further investigation. However, it is useful to note that answers to all these questions will depend on respondents’ perspectives at the time of their interview, cultural influences, sensitivity aspects, the episodic nature of some conditions and individual differences in the interpretation of ‘restricted in everyday activities’.

Additionally, updates to terminology of the conditions were made and tested to align with best practice language. Respondents provided a range of feedback on terminology, such as the term ‘Mental health condition’ being preferred over ‘Mental illness’. Terminology updates improved clarity and usability of the questions.

Testing included measuring ‘whether has disability’ by self-report or proxy. Proxy respondents were less confident answering about subjective experiences on someone else’s behalf and more confident answering about conditions with observable associated behaviours. Overall, outcomes were broadly similar for self-report and proxy responses, however fewer people were identified as having disability when questions were answered by proxy. Additionally, responses provided by proxy identified fewer conditions than by self-report and were less likely to include ‘other conditions’ and ‘taking medication for other conditions’ than when self-reported.

Testing showed that asking a set of questions about specific conditions and restrictions to derive whether someone has disability takes longer to answer and is more cognitively burdensome for some respondents than answering a single question to measure whether has disability. However, a minimum question set provides richer data than a single question and is conceptually comparable to existing measures. A minimum question set is likely to provide comparable data across collections. Further testing is required to confirm this.

The ABS tested measuring ‘Disability group’ (for example, physical or sensory disability) using different versions of a minimum question set based on the SDM (refer to example questions 1 to 6 in 'Example of a minimum question set tested' above). Disability groups can be derived by combining conditions listed in the minimum question set. For example, the sensory disability group comprises ‘Sight problems not corrected by glasses or contact lenses’, ‘Hearing problems’ and ‘Speech problems’.

Testing showed that a minimum question set can be used to effectively derive ‘Disability group’, in addition to ‘Whether has disability’.

Testing included swapping the order of the conditions lists (refer to example questions 1 and 4) and testing a single combined list. 

Testing showed that asking a set of questions about specific conditions and restrictions to derive disability group presents some cognitive burden for respondents. It involves reading relatively long lists and working out which categories apply to the respondent. Some respondents found the condition lists were not easy to use and select from and therefore selected ‘Other’ response categories. Some respondents commented that it was difficult to put their conditions ‘into a box’. These findings led to the testing of a single question to measure ‘Disability group’, discussed below.

Testing compared results for disability group when questions about whether participants were restricted by conditions in the second list (refer to example questions 5 and 6) were asked or not asked. When these questions were included, fewer people reported disability in each disability group, as some people did not report that they were restricted by those conditions that were assumed to be restricting.

A condition list that combined the first and second lists was tested (refer to example questions 1 and 4). Additionally, updated terminology of the conditions, including more inclusive language, was tested. For example, ‘Nervous or emotional condition’ and ‘Mental illness’ were combined into ‘Mental health condition (e.g. anxiety, depression)’. Terminology updates and list restructuring (for example, combining or rewording items) improved clarity and usability.

As it is longer, a combined list may increase cognitive burden, however two separate lists also present some cognitive burden. Further testing and refinement of a single condition list is needed, in the context of best practice for survey design and accessibility across modes. 

The ABS also tested two versions of a single question to measure ‘Disability group’. Testing showed that a single question was difficult to answer when no definitions or examples of conditions for each group were provided. For example, some respondents were unclear what was meant by ‘psychosocial’ disability or how to decide between ‘physical’ and ‘sensory’ disability. 

A single question that included examples of what each category included was easier for respondents to answer. Further testing would be required to understand the statistical impact of measuring disability group using a single question compared with disability group derived by combining conditions listed in the minimum question set. 

During consultation many stakeholders noted disability status as a data need. If disability status is required, additional questions could be added to a minimum question set (refer to example questions 7 to 11 in 'Example of a question set tested to measure disability status' below). The ABS tested questions based on the existing SDM ‘Disability status’ questions. Stakeholders also noted that education and employment restrictions (part of the disability status outputs of the SDM) were lower priority for inclusion in a minimum question set.

Overall, testing showed that the existing SDM-based questions were well understood and could be used for a measure of disability status if required. Minor wording and sequencing improvements were tested. These changes would reduce respondent burden and improve efficiency if collecting disability status.

While education and employment restrictions were lower priority for stakeholders, the ABS still tested different versions of separate questions. The aim was to reduce respondent burden by reducing the number of questions required to derive disability status (without education and employment restrictions outputs) and to provide separate optional questions to measure education and employment restrictions if required (refer to 'Example of a single question tested to measure education and employment restrictions' below).

These questions were asked as ‘Yes/No’ questions, with lists of examples provided to assist respondents’ understanding. Testing showed that these questions were not simple to answer. Examples of education and employment restrictions were also updated to move towards more contemporary concepts and language. More testing and refinement is needed if these are required. 

Findings for proxy reporting of disability status showed there was little difference for proxy responses compared to respondents who answered for themselves.

To ensure consistency, new disability measures should align with the ICF. The ICF framework considers an individual's functioning or disability as multidimensional concepts that reflect an interaction between conditions, environmental factors, and personal factors. The ICF is used as the conceptual underpinning for many existing measures of disability, including the SDAC, SDM and the Washington Group Short Set on Functioning. 

To align with the ICF, disability measures should refer to concepts such as conditions, limitations or impairments that restrict everyday activities. Findings from the review of disability questions and consultation indicated these concepts are widely used as the basis for questions on disability and used by stakeholders to categorise their data requirements (for example, disability group, level of assistance required).

Questions using ICF concepts can be updated to use contemporary examples to better align with best practice without significant impacts to data. 

The ABS’ testing indicated that using alternative concepts, such as a self-identification question, could produce unknown and inconsistent results which may reduce national comparability. Additionally, some people with restricting conditions may not identify as having disability due to factors such as individuals’ experiences and perspectives of disability as well as historical and cultural factors, leading to under-reporting.

A minimum question set should be considered for use first, particularly when more detailed data are needed (including ‘whether has disability’ and ‘disability group’).

The testing showed that in most instances giving people more opportunities and a framework to describe their experiences, such as that provided by a minimum question set, will produce richer results. Additionally, as the minimum question sets tested were underpinned by ICF concepts and based on the SDM they produce comparable results with existing data measures. However, these questions require more effort from respondents and take longer to complete than a single question.

While testing indicates that a single ICF-based question gives comparable results to the existing SDM-based measure of ‘whether has disability’, testing on a larger scale would be required to determine the statistical impact of using a single question over a minimum question set.  

Using contemporary and inclusive language is important as it can reduce respondent burden and make questions easier to understand. 

Existing disability questions that use ICF concepts often refer to specific conditions, limitations or restrictions. Testing found that updating existing examples of specific conditions with contemporary language improved the clarity and usability of questions.

A minimum question set can be used to identify if someone has disability and the type of disability they have (Finding 2). If more information is needed, questions could be added to a minimum question set to meet user requirements.

For example, if data are required about disability status, questions based on the SDM could be used. Testing found that these worked effectively. If data on restrictions in other domains (such as education and employment) are required, further questions could be asked.

This project is an initial step towards addressing some of the DRC data-related recommendations, specifically Recommendation 12.5. Further refinement, testing and consultation with disability stakeholders, including people with disability and their representatives, are needed to implement a nationally consistent way to collect disability data.

Further consultation may focus on:

• clarifying the conceptual underpinnings used in disability measures
• refining the language and terms used in questions
• requirements for future testing of refined disability questions.

Based on outcomes from the consultation, further testing of disability questions would be required. This may include: 

• additional testing of language and terms
• further testing of question topics such as specific restrictions, disability type and disability status
• testing of questions for usability across various modes, such as digital or phone formats
• large-scale testing (for example, field tests) of questions to better understand quantitative and statistical impacts.

If you would like further information about this project, you can contact us by emailing disability.statistics@abs.gov.au.

The ABS would like to thank all participants in the cognitive testing phase of the project. The ABS would also like to thank disability stakeholders for their participation in the consultation phase of the project. The ABS greatly values their perspectives, expertise and contributions and thanks them for their time and input.