Have you ever wondered what higher education provision is on offer in your area? Have you ever wondered whether there was a measure of demand for these courses?
And do you ever wonder what research is going on near you?
To be clear, if it is your job to manage such within your provider, you probably do. If you work on developing a vocational course portfolio, or relationships with employers, or research collaborations, this is information that you can see by looking at your institutional dashboards, or what is pinned to your office wall.
But the Skills White Paper promised a full-on market monitoring function – something aimed as much at the world around Skills England and local skills partnerships as the higher education sector. And this morning the Office for Students published a “prototype” of this dashboard, for sector feedback.
What OfS has released?
There’s already open data available on student recruitment by provider (UCAS), students on higher education courses by provider (HESA), student loans by provider (SLC) and student outcomes (OfS B3, Graduate Outcomes, LEO). OfS’ new dashboard also adds in information about research capabilities, which it derives from commercial data on research funding and research outputs published by Elsevier.
I’m guessing that the inclusion of the latter (along with some TRAC-related information that is merged into the research data) is why OfS has departed from Official Statistics best practice by choosing not to publish the underlying data alongside the dashboard. We are promised that this will arrive alongside future iterations of the tool – which will be determined by feedback from users.
So what we do get is what amounts to a series of tabs (one per data source) summarising information either by provider, or in some case by local area (ILTS1 region, ILTS2 region, LSIP region). The default is to allow the user to view changes in key variables over time – there are some maps linked to student uptake which largely duplicate what HESA has already published.
What’s missing?
While any release of data is welcome, it feels like OfS is not getting the maximum value from these sources. This is because of a curious design choice.
For me, the value of data from multiple sources is the ability to cross-tabulate them in order to answer questions of strategic value. For instance, do areas that are research active in a particular subject also offer good teaching in that subject? Which universities recruit locally and which are more focused on bringing students into a local area? Where are (as the British Academy has investigated) subject cold spots, and which providers have the potential to fill them?
The OfS dashboard keeps each source of data separate – it is best understood as a set of complementary dashboards rather than a single entity. It is a way of allowing people to explore each of these datasets, but it doesn’t allow for the kind of analysis that combining them should be able to bring.
In part, this is understandable – different data sources use different standards (especially geographically and in defining subject areas) which makes it complicated to compare like with like. League table compilers famously struggle to include research data (which generally maps to either cost centres or REF units of assessment) while also using teaching data (which these days pretty much standardised to CAH): there is not a reliable mapping between these standards.
But it is not impossible – especially when you are an arms-length body with the ability to see and use all kinds of data not routinely available to mortals like ourselves, and even specify how it is organised. My suspicion is that the choice and nature of this release is defensive, in that it does not require data other than what is routinely published (or purchasable) – making this choice unfortunately limits the ability to provide new insight. OfS should use this exercise as an opportunity to consider what data is currently published and how it is made available.
We also do not get the promised information on research infrastructure – the research section of the dashboard is sub-optimal for reasons that I will get to – but it is a shame not to look at what kit is available where given that Jisc runs the marvellous Equipment Data Service that holds this data.
Why is OfS trying to do this now?
The impetus for this market monitoring dashboard comes from the Skills White Paper. The plan was to “provide a clear, single picture of higher education supply and demand” in order to “highlight any gaps between demand, supply and local need.” It is very much of a piece with the work of Skills England (there’s a local skills dashboard, for example) that aims to inform the work of local skills improvement partnerships (LSIPs) in understanding what provision needs to be commissioned to meet local industry needs.
Situating skills and training decisions in local areas is one of the key themes in the government’s devolution agenda – and Labour’s decision to actively include higher education (as a skills provider as well as an employment sector) means that the sector is much more engaged in this work. The aim of the dashboard, therefore, is to offer an approachable single source of truth to inform these local discussions.
Data like this also has the potential to be used in national planning (although opinions are variable as to whether we have any chance of doing that). There’s been rumblings about a retooling of the OfS strategic priorities grant (SBG) to focus on national priorities rather than “high cost” subjects, and you’d imagine that data like this would add a local or regional aspect to decision making. It is possible that OfS could directly address cold spots on a subject-by-subject basis via the use of financial incentives – and to do so it would need to be certain of both the scope of current supply and the nature of likely demand.
Parallel to this, the White Paper also committed the Department for Science, Innovation and Technology (no flowers by request) and UKRI to examining “the provision of research activity which delivers against the Industrial Strategy, missions and sovereign capability”, and “assessing changes in capability against our needs.” Given the White Papers interest in using research funding to drive teaching provision, it makes sense to bring these sources of data together.
The dashboard era
I’m hardly one to complain about dashboards as a data analysis and exploration tool – after all, I’ve built one or two in the past. But for me a well-designed dashboard needs to use visualisation and data design to allow the reader to derive an insight that is not available directly from the raw data.
There has been a trend, both in higher education data and in public data more generally, to release data dashboards for public facing information. The canonical examples of excellent practice here are the crime maps that display information to the public about the location, nature, and outcomes of reported crimes.
So when I look at an OfS dashboard, my usual approach is to ask
- What insight is this dashboard trying to provide?
- What data is it building on?
- Does the design use what is in the data to offer the insight that is promised?
- Can I use the underlying data to build something that does?
For me, on question 3 the answer is “no, not yet”. That we don’t get the underlying data so we can fix that is annoying, bits that are causing the hold up (the research stuff) is neither interesting nor useful compared to what can be derived from what is already available. I know everyone and their chosen AI life companion is knocking up dashboards and interactives at this point, but the fact that we can chuck all kinds of data up on the same screen doesn’t always mean that we should.
That said, I thought I’d have a go at something providers might find useful. Fire up your big monitor and let me know what you think.
So this lets you examine a bunch of different datasets by provider and subject. The two yellow filters affect all but one of the visualisations on the dashboard – the exception is the very first one which shows home and current graduate regions using the LEO data set: that’s only available by provider.
The two on the row below are based on UCAS data and work (like the rest of them) by provider and subject. The usual gender, age, and domicile filters are available. On the left we have acceptances, offers and applications as a time series, on the right we have year on year change in that subject area across the sector (there is a flying filter letting you choose your comparator year; this popular chart shows you who is fishing deeply in the pool of available undergraduates, by subject).
Under this is a “cold spot” map by your chosen subject area – you can filter by mode and level and this shows the most recent year of data. There’s a highlighter, or you can mouse over any of the previous charts to highlight the provider you have chosen in the filter.
Right at the bottom we get graduate destinations by provider and subject – there’s filters for the level and mode of the qualification, and the interim study marker.
Data sources
For recruitment we turn, of course, to the UCAS End of Cycle provider data – the most recent release is the 2025 cycle. Though OfS supplements this with student loan take up data from SLC it only seems to do this to provide another perspective on student headcount by subject area, so I’m happy to ignore it at this point.
OfS publishes a size and shape dashboard covering student numbers by provider, and it has some nice features offering a bit more fidelity on entrant status, student domicile and level of study. But HESA DT051 Table 49 offers a UK wide perspective, and I’m not really sure why OfS wants to separate out home funded students given that we as a nation benefit greatly from the skills of international students too.
For student outcomes we could use the B3 release (with the usual caveats about the way indicators are constructed) for completion and continuation – but it is really progression that interests us. For progression there is probably more of interest in graduate outcomes – which allows us to look at what graduates end up doing by provider and subject – and the regional movement stuff available in LEO.
Arguably we should really be looking at course level and at the demand for the skills each course covers. There’s a kind of version of that in this article, but the perils of using the Unistats dataset would make it indicative at best. And arguably we should be able to allocate students, using a course based approach, to individual campuses rather than the main campus in order to build better maps. You can’t get data on student numbers by campus (you can get them by franchise provider which isn’t quite the same thing).
For research I don’t have access to the citation measures from Elsevier, but honestly citation-derived metrics are a hugely limited and deeply problematic proxy for research quality (and I’m still idealistic enough to sign up to the Declaration on Research Assessment) so I don’t mind that much. OfS also uses funding inputs, and staff FTE, but these are only available by cost centre – and given that all the other data I am using defaults to CAH and there are no reliable ways to join the two that also causes a problem.
Fundamentally, while research power is an important thing to think about on a local and regional level (though arguably not in the sense of cold spots as high level research is often commissioned commercially nationally or internationally) I’m not convinced I’m going to easily add much value by adding research data to the student-related stuff at this point. So I’ll deal with it separately.