Findings on the appropriateness and inclusivity of concepts in data
Survey response options
Participants discussed the concepts and definitions used to collect and categorise data, within a range of contexts. Concerns were raised around the problematic wording of survey questions and response options, which could lead to community members feeling excluded from data collection and potentially yield inaccurate results as respondents must select a response option which they do not identify with. An example was provided of surveys where gender questions only have male and female options.
“So right from the outset, we’re not including people within our surveys. They’re made to feel like they don’t fall into one of those two categories. You’re immediately creating a not very inclusive environment and making people feel like they don’t belong as part of that survey.” (Welsh Government participant)
A Northern Ireland Executive participant mentioned that some individuals cannot accurately reflect their nationality in labour market statistics, for example, because they are unable to identify as dual nationals. “You can be British or Irish, or both. You’re not allowed to answer both.” It was noted that concepts, or the differences between concepts, may change over time, for example with sex and gender, or race and ethnicity. As people’s understanding changes this could lead to inaccurate data collection, or inaccurate reporting by respondents, it could also prevent people from responding in a way that reflects their identity or become a barrier to them responding at all.
A research funding organisation participant noted that recognition of the fluid nature of identities is not currently at the forefront of survey categorisations, and there was said to be a significant time lag between being able to “identify yourself in the way that works for you, versus what is currently being rolled out.” Local government participants also highlighted issues relating to the harmonisation of data definitions at a local level, and the importance of local data reflecting the evolution of definitions to represent changes in society. Academic participants reported the need to balance keeping up with society in terms of definitions and classifications with the potential loss of comparability over time if classifications and concepts are frequently revised.
“You want to respect people’s experience while on the other hand you need some level of consistency.” (Academic participant)
Harmonisation and comparability
There were calls among academic participants for guidance to be put in place around the revision of harmonised definitions so this it is made explicit.
“There is a challenge, do we stick with the standard definitions for some foreseeable future or do we kind of have some kind of properly thought through revisions protocol process, which says, we’ll stick with these for a period and then here will be the way in which we may update them or not.” (Academic participant)
Participants from London boroughs also stressed the importance of using tailored definitions in local surveys, so that the diversity and distinctiveness of each London borough can be better reflected in the data. It was mentioned that specific populations could not be identified for policy purposes.
“We were unable to identify South-Eastern Asian communities within London and… in certain parts of North London they’re very concerned that they’re unable to identify some of the Cypriot communities.” (London borough participant)
Academic participants discussed problems with the comparability of ethnicity across the Devolved Administration Censuses.
“The census varies between England, Wales, Scotland, Northern Ireland, say, for ethnicity, with the smaller number of categories in Northern Ireland, meaning that if you want to do a UK analysis, you end up having to reduce the groups down to the categories you used in Northern Ireland across the UK [to be] compatible.” (Academic participant)
Classifications and categorisations
Participants discussed the evolving classifications that are used to capture ethnicity in society, reflected that the 2001 Census categories for ethnicities are still widely used in data collection and that these do not adequately reflect the characteristics of under-represented groups today.
“Ethnicity categorisation, we’re still using the 2001 census. They [the health system] haven’t updated it [ethnicity categorisation] in over a decade, rendering entire ethnic groups invisible… because they said they couldn’t afford it. Well, that’s a matter of prioritisation.” (Central government participant)
The use of broad meta-categories for ethnic grouping, such as “BAME” (Black, Asian and Minority Ethnic), was also criticised as a term with which nobody identifies. The term was also seen to marginalise certain groups, making them “invisible” in statistics and aggregating groups which have very different characteristics, thus providing little meaningful insight. It was highlighted that such categorisation results in ethnic groups being “lumped together” irrespective of their differences.
“The attitudes could be very different depending on whether you are Muslim, Jew, Sikh, Hindu or Christian…. the way ethnicities currently are grouped together… people do not see themselves in the data and that is a vital point of being inclusive.” (Learned society participant)
Concerns were also raised about the perceived inconsistent use of the “other” category within data collection processes.
“An APS [Annual Population Survey] ‘other’ is quite wide when you’re looking at ethnicity, whereas in other cases, in other datasets, ‘other’ is just everything not listed above and usually it’s a very long list.” (Combined authority participant)
Within ethnicity data in particular, it was noted that there has been a substantial increase in individuals selecting “other” or “mixed category” within surveys.
“That data is essentially useless at that point, if we’re looking at particular answers to certain questions.” (Research funding organisation participant)
Overall, “other” categories were described as problematic as they provide no understanding of an individual’s identity, especially if they are from mixed ethnicity backgrounds.
Although issues were highlighted with ethnicity definitions and the need for classifications to keep pace with society, some participants promoted the potential for granularity with ethnicity classifications, and felt that this could be a good approach for disability to follow.
“[For disability] we probably need to move closer towards our ethnicity classifications, where you have…multiple different levels of granularity.” (Central government participant)
Additionally, some participants highlighted a lack of standardised recording of disability across the UK, with the vast number of different definitions and classifications causing significant issues for data analysis. A central government participant said that mixed usage of Government Statistical Service (GSS) harmonised standards could result in fragmented knowledge of what disability data is available which can cause difficulties when you’re trying to compare data between datasets.
Welsh Government participants discussed how the social model of disability is used in Wales, which considers the personal barriers that people face. This therefore creates a need to capture information in a slightly different way, that would be inconsistent with what’s happening across the rest of the UK.
Participants from the academic and government groups discussed how disability identification can depend on the context within which the data are collected.
“[Disability can be] where someone identifies as disabled, where someone would fit the Equality Act definition of being a disabled person…anything around medical histories…and barriers facing day-to-day life as well.” (Research funding organisation participant)
Social class and disadvantage were also identified as concepts causing issues for analysts, due a lack of consistency or harmonised definitions. Highest qualification of parents, self-professed socio-economic status
were highlighted by a central government participant as examples of how class has been captured within data.
“[It would be good to] come up with some easy way to ask about socio-economic class, that could be even asked as a census question and become more of a protected characteristics type question.”(Scottish Government participants)
For children and young people, disadvantage was said to be defined differently depending on their age, with free school meals being a measure for younger pupils, and indices of deprivation being used for older young people and university students. However, a research funding organisation participant noted that area-based measures are actually quite a poor proxy for identifying poor children or children from groups that are disadvantaged.
Local government participants reported that this was particularly problematic when trying to undertake analyses for specific areas.
Solutions for addressing conceptual issues
Participants recommended allowing individuals to self-identify and categorise themselves for personal characteristics, such as ethnicity and gender, and providing as many response options for these as is feasible. This could be done by providing free text boxes where “other” categories have been selected and undertaking analysis on these responses. This would inform researchers of how people are self-identifying and form an evidence base for revising classifications moving forward.
There was a suggestion that “intersex” should be regularly included as a response option under sex.
“It is a legitimate category that people want to identify into, and it is almost always missed off as well.” (Research funding organisation participant)
Providing a clear steer and guidance from government regarding the correct categorisations to use in surveys, to maintain a consistent picture across UK data collection and ensure that it is not entirely the responsibility of research organisations to “get this right.”
Several participants suggested the need to improve the consistency and harmonisation of definitions throughout UK data sources and between government departments to ensure that all parts of the UK are sufficiently represented within data and to bridge the gaps caused by the lack of harmonisation. It was stressed that the Government Statistical Service harmonised standards should be used more frequently and more consistently, to better bring together definitions between different datasets. However, ensuring categorisations are appropriate and up-to-date with societal definitions was also said to be needed.
“The more you can bring that data together the more you’ll get a more holistic picture.” (Learned society participant)
“Pushing at the next boundary once you’ve got the most recent guidance out, rather than letting a long time-lag develop.” (Research funding organisation participant)
Irrespective of the described issues arising from lack of comparability over time with changing definitions, a research funding organisation participant called for the need to address out-of-date categorisations, as actually not acting could cause more harm than acting at this point.
Suggestions were made for how to keep up with societal changes.
“Horizon scanning, reading the news, trying to be aware of the emerging issues and thinking around whether this is going to be something which is impacting your local community.” (Non-metropolitan local authority participant).
Reviews of current classifications were also encouraged, particularly considering bigger societal events which may require changes, such as Brexit
. Some participants suggested the need for a full review of ethnicity classifications in the UK, in conjunction with categories for religion.
“[Need to create a] combination of almost a matrix of ethnicity and religion to identify the key groups or communities that need to be separately identified for statistical purposes to inform policy.” (Learned society participant)
Certain participants suggested that this would help to make visible the communities that have been hidden behind broad categories until now.
Back to top