Harmonising decentralised data collections
Harmonised standards and guidance help us improve how we understand and compare statistics. Harmonised standards include:
- survey questions
- suggested presentations
- information for data users
Producers of statistics can use these harmonised standards to align with others to increase the usefulness of their statistics.
My colleague Darren Stillwell has recently written a blog on the importance of data harmonisation. The Race Disparity Unit uses harmonised ethnicity classification in the collection and publication of statistical and other data.
Challenges of harmonisation
There are many benefits to harmonisation. But there are also many challenges, which Darren has written about previously. For example:
- the potential cost of introducing harmonised standards where they do not currently exist
- the data requirements for departments or users – these requirements might include data for specific ethnic groups which are not collected in harmonised standards
- the need to keep consistent data over time – this might be difficult if there is a change to a harmonised standard
Another challenge of harmonisation is where many similar organisations are collecting and publishing data. Some departments need to make sure that harmonised standards are followed by everyone involved in the project. This includes departments with agencies or local bodies which provide a service and collect data about users, even if they are operating at arm’s length.
For example, each fire and rescue authority in the UK collects data on the ethnicity of their workforce.
Departments also need to make sure that harmonised standards are being used when they outsource work to the private sector. We recently became aware of an example of this. The UK Health Security Agency (UKHSA) has outsourced coronavirus (COVID-19) polymerase chain reaction (PCR) testing to over 1,100 companies as of late December 2021. It also has its own national testing service. UKHSA asked the companies to collect the ethnicity of the people whose samples they are testing, but it did not tell the companies which classifications to use. This meant some companies used non-standard classifications, such as:
- Mediterranean or Hispanic
- African or Caribbean
- Indian or Asian
- Chinese, Japanese, East Asian or South-East Asian
- Arabic or North-African
We spoke to UKHSA about this problem. They told us that they will contact the private laboratories to resolve it. We welcome this, as otherwise it would not be possible to produce coherent statistics about PCR testing by ethnicity. It is much better to collect data using harmonised standards than to try to recode data after it has already been gathered.
If data is collected on a comparable basis, then it will lead to better quality analysis and better quality decision-making.
Future work on harmonisation
We will be working to consider what changes might be made to the ethnicity harmonised standard following the 2021 Census. We are looking forward to working with ONS, and the network of harmonisation champions across government and the devolved administrations on this.
The ONS has recently set out Government Statistical Service (GSS) harmonisation priorities in their workplan. They have identified that the harmonisation of ethnicity classifications is a priority area, along with disability, mental health, and sex and gender identity.
It will be important that any new standard developed by the GSS about health data is consistent with any new standard used by the NHS. The NHS still uses the classification from the 2001 Census. This was recommended by the Minister of State for Equalities in her final report on progress to address COVID-19 health inequalities
This year, we are planning to work with English local authorities and other public bodies, like fire and rescue services, to look at the ways they collect ethnicity data. We will be considering whether we could promote the harmonised standard. We will need to include companies that collect data and provide public services on behalf of government as part of this work.
In the meantime, I would encourage departments to review any outsourced data collection to make sure that it is fully harmonised. This includes all topics, not just ethnicity.