New data analysis from David Kernohan shows that, contrary to popular belief, more staff may not be the most important way to keep students engaged and satisfied with their experience
This article is more than 2 years old

There’s a broadly-drawn equation that seems to underpin a lot of sector strategic thinking.
Common sense suggests that as a provider you improve the student experience by investing in academic staff.
A provider – indeed a subject area at a provider, as we know there is a subject effect on student satisfaction – offering a student experience involving packed lecture theatres and sporadic supervision will be perceived as less satisfying than smaller classes, seminars, tutorials, and individual support.
Everyone believes this. But it doesn’t appear to be true.
Plotting student:staff ratio against NSS fails to show even a non-significant relationship between satisfaction and staff numbers. Looking primarily at NSS question 15 (which relates to the ease of contacting teaching staff and seems most likely to see an impact from staff student ratios) there are no clear relationships between our two variables in any subject area. Clearly this is a data combination that is indicative at best (see the caveats below) but the absence of any visible relationship at all is remarkable.
Neither does there appear to be any relationship between the size of a department (in student numbers) and satisfaction – you would expect, perhaps, to see large departments with many hundreds of students making it harder for students to contact staff in a meaningful way.
You can try for yourself, across all NSS questions and subject areas, using this dashboard:
And it is also possible to see a similar lack of impact at a provider level, by subject area:
Do please have a play and comment if you spot anything that looks interesting.
Health warning – the version of this chart up first thing this morning had a very small number of errors. This new version (as of just after 9am) is correct as far as I can currently see.
The relationship between cost and quality is well understood – generally if you want to decrease costs you also need to decrease quality. Student satisfaction feels like the most appropriate way to measure the quality of the direct student experience (alternatives could include student continuation, or graduate progression – though these are also influenced by exogenous factors like the cost of living and the state of local or regional graduate employment).
Given the above findings it does feel like there must be ways to improve the student experience without increasing costs – teaching quality enhancement has long been neglected as a policy priority in England (the carrots of the 2000s have become the sticks of the current model of quality assurance), and where a subject area in a provider appears to be providing a good experience despite a high student-staff ratio it feels like it would make sense to investigate what is going on. You could imagine a kind of “reverse investigation” where the regulator or some other body could collect and share good practice in outstanding departments.
These charts – as is typical with NSS data – do a lot to dismantle institutional hierarchies. There are a fair number of prestigious courses at prestigious providers that have a surprisingly high SSR, whereas we see a low SSR at providers we would not necessarily expect – the University of Chester law school is more similar to Law at Oxford (in terms of size, SSR, and Q15 NSS positivity) than any other. SSR is no respecter of prestige or league tables either.
I showed this data to Shitij Kapur at King’s College London – he told me:
Having been in five English-speaking institutions in four continents over my career I have been struck by the difference in SSRs across nations while delivering seemingly similar student experience and outcome. I always thought that reason I could not draw simple links between SSR and outcome was because these national systems were very different and had different tariffs and different regulations. Therefore it was very striking to see the wide variations in SSR across UK institutions, when they are all working within a single regulatory and tariff framework and trying to be their efficient best. It is remarkable how even within a discipline, and within a mission group, there can be nearly a 200 per cent variation in SSR with no discernible effect on NSS-measured student experience. These simple graphs raise difficult questions. Is this variation by intentional design? Or is it an accident of history and tradition of the different institutions?
SSR and NSS are both blunt measures and perhaps there are complexities in these measures themselves that obscure the relationship between staff effort and student experience. But, given that staff time is probably the most precious asset in any university – it behoves us to understand this relationship much better. In fact, as Gen-AI technologies may begin to replace some of the routine and rote aspects of university work, how and where we deploy our precious staff time will have to be a much more deliberate decision.
This analysis hinges on a “best fit” mapping between Common Academic Hierarchy (CAH) subject codes and HESA Cost Centres – for details of how this has been done please see the end of the article. CAH (and the HECOS codes that underpin it) relate to courses – actually modules if we are being specific – whereas cost centres refer largely to departments, faculties, and school. It’s not completely safe to assume that courses are taught in the most obvious school, but for these purposes there are a fair number of confident assumptions that can be made – for instance law courses are likely to be taught (in the main) in your law school.
Staff are only mapped to subject areas via cost centres, and as we get student numbers by cost centres too it’s fair to use these to construct a student:staff ratio (the common complaint is that this includes research only staff – using custom HESA data these have been removed). You may wish to argue that it would make more sense to look at undergraduate students only – for me it is arguable that the expansion in PGT makes it more important to include postgraduates too (and one day we may even get a PGT NSS!), though clearly NSS is (primarily) undergraduate only). If undergraduate students are struggling to get access to staff, it is very likely that this is happening for PGT too – and a large UG and PGT student population would have much the same impact as a similar sized population made up solely of undergraduates.
Some of the larger and more unlikely SSRs will stem from partnership and franchise arrangements (all students registered to a provider have their FTEs recorded in this dataset, though only staff employed by a provider are shown). I’d like to be able to finesse this issue further, but we’re still waiting for detailed data on franchise and partnership arrangements.
The staff and student FTE data is from 2021-22 – the NSS data from 2023 (the staff and student FTE refers to the second year of study for those who completed the 2023 NSS, which felt like a good compromise). This can be updated when we get 2022-23 student data in mid-April to see if there is any difference.
The analysis uses CAH level 3 subjects as an appropriate level to map to cost centres. As above, this is a best fit mapping only and does not take into account individual provider circumstances (unlike, for example, the Guardian University Guide mapping). This is probably fair in that everyone is disadvantaged by this choice. The full mapping is available as a google sheet – feel free to reuse it if it is helpful, or critique it if not. In short:

Jim Dickinson | Analysis | 4/09/26
Ola Thomson | Analysis | 27/08/26

Wendy Ashall | Analysis | 26/08/26

Jim Dickinson | Analysis | 25/08/26

David Kernohan | Analysis | 12/08/26