A year ago I wrote about what was happening to the staff:student ratio, prompted by a sabbatical officer whose experience of “personal” tutoring in groups of 35 turned out to be visible, in outline at least, in the Guardian league table data for his subject.

The problem back then was that the Guardian had just changed how it calculates the ratio, so the 2023/24 figures couldn’t really be compared with what came before. The 2027 edition of the guide, out this week, uses 2024/25 data on broadly the same new method – which means for the first time we can look at a year-on-year change without the numbers lying to us quite so much.

Because a fair bit of what follows is about looking things up, I’ve rebuilt the interactive charts so you can do that yourself. The tool below lets you look at every subject at a provider, every provider in a subject, or one provider and one subject over time.

One health warning before we start. The Guardian’s new method – it now uses Data Futures student returns and discounts placement activity much more heavily than it used to – pushes ratios down compared with the old one. So anywhere a ratio has gone up across the change in method, that rise is real and probably understated. Anywhere it has gone down, we can’t be sure. That asymmetry matters for everything below.

Flat on average

The headline is that nothing much happened. Across the 123 providers that appear in both years, the average institution-wide ratio went from 15.2:1 to 15.5:1. Fifty-nine providers went up, 61 went down, three stayed the same, and the median change was zero.

If you stopped there you’d conclude that the story I told last year has petered out. It hasn’t – it has just moved from the middle of the distribution to the top of it.

Twenty-eight providers saw their overall ratio rise by a full point or more in a single year, and 15 rose by two points or more. Only 15 fell by a point or more, and just four fell by two or more. The number of providers whose overall ratio is 20:1 or worse went from six to nine. The average is flat because the middle is flat – the tail is stretching.

The biggest single-year rises are eye-catching. East London went from 18.4:1 to 25.0:1. Greater Manchester went from 13.4:1 to 17.9:1. Bucks New University and the University of the West of Scotland each rose by 3.4, Glasgow Caledonian by 3.3 to 26.1:1 – now the highest overall ratio in the tables – and Trinity Saint David by 3.1. Sunderland, Huddersfield, London South Bank, Middlesex, Surrey and Kent all rose by between 2.3 and 2.7 points in a single year.

Remember that these are rises measured on a method that tends to push the number down. Whatever is going on at those providers, it isn’t noise.

The other direction

The falls are interesting too, though for a different reason. London Met dropped from 19.1:1 to 15.1:1, De Montfort from 18.2:1 to 15.2:1, and Wrexham from 23.0:1 to 20.2:1. Over the longer run, Bedfordshire has gone from 24.5:1 in 2021/22 to 15.7:1, Brunel from 18.2:1 to 11.4:1, and Heriot-Watt, Bath Spa and Sussex have each fallen by more than five points over the same period.

A falling ratio is good news for students in the arithmetic sense – there are fewer of them per member of academic staff. But the data can’t tell us why, and there are three candidate explanations that look very different from a student’s point of view.

The first is that the university has hired. The second is that its students have gone – and given what has happened to recruitment, and how many providers have run redundancy programmes, it would be brave to assume hiring without checking. Where a ratio has fallen sharply while a redundancy programme was running, the student numbers must have fallen faster than the staff numbers did.

The third is franchising. Franchised students have never counted in the Guardian’s staff:student ratio, on the assumption that the staff teaching them aren’t reported by the registering university either. But this year’s methodology says the Guardian has found “secondary effects from hosting franchised provision that lead to distorted and inconsistent results”, and has responded by detecting any association between a franchising department and its cost centre statistics and applying “a tighter credibility threshold” to every data point in it. In plain English, some of the numbers reported by the sector’s big franchisors have been cleaned differently this year from last.

The Guardian doesn’t say which providers or departments that touched, or in which direction, so I can’t tell you how much of the movement at London Met, De Montfort, Bedfordshire or Wrexham is students, staff or spreadsheet. What I can say is that the like-for-like comparison I promised above is a little less like-for-like at exactly the providers where the ratio was highest to begin with, and that if a university with a large franchise operation tells us its staffing has improved in this year’s tables, the first question is whether the Guardian’s cleaning did the improving.

Same fee, different subject

The more important story sits below the institutional averages.

In 2024/25 the subjects with the most students per member of staff are marketing and public relations at 21.1:1, hospitality, event management and tourism at 20.9:1, business and management at 20.6:1, animation and game design at 20.5:1, accounting and finance at 20.4:1, law at 20.3:1 and computer science at 20.2:1. At the other end, veterinary science, medicine and dentistry sit at about 9:1, languages at 9.6:1 and physics at 10.6:1. Nursing sits in the middle at about 17.5:1.

Nobody thinks a business degree should be resourced like a dentistry degree. But a home undergraduate pays the same fee for both, and the gap between them is now wider than ever at two to one.

It gets even bigger inside institutions. Across the 122 providers with eight or more subjects in the tables, the typical gap between the best and worst resourced subject on the same campus is 15.4 points – the median provider’s worst subject has 2.7 times as many students per member of staff as its best. At 109 of them the ratio between top and bottom is at least two to one, and at 48 it is at least three to one.

At London Met, accounting and finance is at 38:1 while sociology and social policy is at 4.8:1. At Trinity Saint David business and management is at 34.1:1 and music at 5.1:1. At Chester accounting is at 33.3:1 and languages at 6:1. And it’s the same subjects at the top nearly everywhere. At 83 of the 122 providers, the subject with the highest ratio on campus is one of law, computer science, economics, accounting, business, marketing or criminology.

A subject running at three times the campus average is a subject where personal tutoring, dissertation supervision, responding to feedback and marking turnaround are the first things to stop happening.

The tail

Last year I said that some of the ratios were so high they were hard to believe. This year there are even more toast-droppers.

Across the 3,572 subject-and-provider combinations in the 2027 guide, 184 have a ratio of 25:1 or worse, 53 are at 30:1 or worse, 27 are at 35:1 or worse and 11 are at 40:1 or worse. In 2018/19 the equivalent counts were 121, 27, seven and three. In 2021/22, on the old method, they were 227, 55, 18 and 10.

So the broad middle of the extreme range – the 25 to 30 bracket – has shrunk since 2021/22, which is at least partly the method change doing what it does. But the very top has kept growing through a change in method that should have pulled it down. Twenty-seven combinations at 35:1 or worse, against 18 three years ago and seven six years ago.

Computer science at East London is at 50.6:1, and so is animation and game design there because the two share a cost centre. Business and management at the London School of Management Education is at 48.4:1. Computer science at West London is at 46.1:1. East London’s accounting, business and marketing cluster is at 40.9:1. Computer science at Glasgow Caledonian is at 38.6:1. A couple of the others at the top of the list look like artefacts of how HESA cost centres map onto Guardian subjects rather than real classrooms – BIMM’s music, drama and film all report the same 47.9:1, and politics at Ulster at 47.5:1 is out of line with everything else there – which is why this data is a set of clues rather than a verdict.

Computer science

Last year I listed business, marketing, computer science, economics, law and psychology as the subjects where the ratio had risen dramatically at plenty of providers. With another year of data and a proper look at the method change, I need to correct that.

Measured against 2021/22, most of those subjects now show a lower average ratio across the providers that appear in both years – business by 1.9 points, economics by 2.0, accounting by 2.1, marketing by 1.2, law by 1.0. But every one of those comparisons crosses the change in method, which pushes the numbers down, so I can’t tell you whether the subjects have actually improved or whether the calculation has flattered them. What I can say is that they remain the most thinly staffed subjects in the sector on the new numbers, and that the biggest single-year rises this year are at providers where they dominate.

Two subjects have gone up even through a method change that should have pulled them down. Animation and game design is up 1.9 points on average since 2021/22, with 68 per cent of providers rising. And computer science is up 0.8 on average, with roughly half of providers rising and 30 of them rising by three points or more. Computer science is now the subject with the highest ratio on campus at 21 providers, and computer science at East London, West London and Glasgow Caledonian are three of the twelve highest ratios in the entire table.

If you wanted a single subject to worry about on the basis of this data, it’s that one – one that skills strategies say they want more of.

Two sectors

There is one more story in the numbers.

In 2018/19 the average overall ratio at Russell Group providers was 13.3:1, and at everyone else it was 16.4:1. In 2024/25 it is 12.3:1 and 16.2:1. The gap has widened from about three points to about four, and it has widened because the Russell Group has got better rather than because the rest has got worse.

That would be one thing if it were about subject mix – research-intensive universities teach more medicine and fewer marketing degrees. But it holds within subjects too. In computer science the Russell Group average is 15.4:1 and everyone else’s is 21.5:1. In business and management it’s 16.9:1 against 21.5:1. In law it’s 17.2:1 against 21.1:1, in marketing 17.0:1 against 21.8:1, in accounting 17.4:1 against 21.2:1, in criminology 15.2:1 against 18.9:1. The four lowest overall ratios in the country are Oxford at 9:1, Edinburgh at 9.9:1, and Cambridge and UCL at 10.5:1.

The student doing a law degree at a post-92 pays the same regulated fee as the student doing one at a Russell Group university, and gets somewhere between four and six more classmates per member of staff for it, depending on the subject. That is not a new observation, but it’s a bigger gap than it was, and the direction of travel is the wrong one.

Faulty signals, still

Everything above is about 2024/25. It has been published in September 2026, for students applying to start in autumn 2027, who will graduate in 2030. When a provider’s ratio can move by six points in a single year, a three-year lag is not a rounding error – it’s the difference between the course that was advertised and the course that gets delivered.

And still nothing in the regulatory architecture, in England or elsewhere in the UK, treats a rapid deterioration in the resource behind a course as something a student is entitled to know about, let alone something a provider has to do anything about. The Office for Students’ approach to quality leans on outcomes measured donkey’s years later. The National Student Survey asks final year students about an experience they chose three years earlier. Neither would have caught the student officers I started with, and neither will catch the next one.

What the data can do is give staff, students, students’ unions, governors and journalists a place to start asking. Look up your own provider. Look up the subject with the highest ratio on campus and ask what the personal tutoring arrangements actually are there. Look up how it has moved since 2021/22 and ask whether anyone told the students paying for it all.