In our previous blog, we introduced our project to refresh the classification of the national charity registers into ICNPTSO (activity area) categories, using keyword searching, and where possible applying machine learning techniques. To our knowledge, this was last attempted at scale over a decade ago.
We also, however, acknowledged some of the challenges UK researchers face when using the ICNPTSO categories. These categories are broad groupings, originally designed to help enable international comparisons of non-profit organisations, particularly as part of national accounting processes. There are large gaps in a UK context, with notable omissions such as food banks, drug addiction services, or domestic violence refuges.
As part of our project, therefore, we wanted to develop and publish a new, additional classification system, alongside the ICNPTSO. We refer to these classifications informally as ‘tags’ and have aimed to make them more UK specific, more fine grained, and to allow charities to be associated with several tags at once.
This UK Charity Classification System (suggestions for a catchier name than UKCCS on the back of a postcard please), was developed iteratively over the early stages of the project. Each week we revisited the list of tags to make any changes in light of that week’s classification work.
Our classifications are derived from the activities that charities write about in their annual returns. Where possible we have tried to hone in on what it is that makes an organisational ‘charitable’. In other words, what is the charity’s mission?
Some define this in terms of a desired outcome, such as a less polluted world. Others focus on a particular group that they seek to help, such as military veterans. And others still focus on the specific activity or facility, for example offering benefits advice. Whilst trying to avoid overlap, we have included all of these types in the tagging framework to at least some degree.
The classifications included so far can be viewed on our Airtable list of charity tags.
How successful have we been?
It is with some trepidation that we release our draft tagging framework into the world. Why? Because it is almost certainly ‘wrong’, or at least ‘not quite right’. Because of course there is no right set of categories, no right level of specificity, and no right use of terminology. Instead, we are in the businesses of trade-offs, sacrificing some potential benefits for others.
Furthermore, the tagging system remains a work in progress. As we start to develop keyword search rules, we are likely to continue to consolidate the list. There is little point, however, in asking for feedback at the very end of the development process.
Harder still, we are aware that as authors of the tagging framework, we are implicitly helping to make some charity groupings more visible, whilst submerging or obscuring others. Some authors argue that this can help to reinforce existing power structures in society. They question who benefits from these classifications?
For our part, we hope the framework will be useful primarily for research, including by analysts from within the sector themselves. It is inherently a somewhat conservative approach, as we hope to capture the fields that charities and researchers are already using to frame their activities, rather than develop categories that are deliberately novel or challenging.
How many tags?
The tagging framework is currently over 260 tags long, though it may become shorter. Its length may make it unwieldy and difficult to apply without a high level of familiarity. More nuance also arguably allows more room for ambiguity and disagreement over which tag fits best.
On the other hand, it is inevitably not long enough. Undoubtedly some users will be frustrated that their area of interest has not been included. Sometimes a combination of tags will be needed to capture a particular area of activity.
We have tried, as far as this is possible, to be led by the data itself. When in doubt, we entered key terms into the existing keyword charity search tool, always aiming for a minimum of 100-200 associated charities. Users are, of course, able to use this tool themselves to create their own searches for bespoke groups.
Despite the challenges, or perhaps because of them, we would welcome your feedback on the categories! We are keen to break out of our own ‘group think’ and understand how meaningful these categories are to a wider audience.
- Are there too many tags, or not enough?
- Have we missed any major areas, or are any of the tags superfluous?
- Is our hierarchy of categories and subcategories the right one?
Please send us your thoughts via our feedback form.
As noted above, the classifications can be viewed on our Airtable list of charity tags.
Chris Damm (Sheffield Hallam University), Oliver Chan (National Council for Voluntary Organisations); and David Kane (independent Data Scientist).