I was flicking through the TEF subject-level consultation document – what else would you do on a Monday morning? – when my picky classifier’s mind was caught by some jarring bits of wording. “We will use a subject classification system to define what a ‘subject’ is for the purpose of assessment and ratings… We propose a system of 35 subjects for subject-level TEF. This is based on the second level of the Common Aggregation Hierarchy (CAH2).”
Now, I know what they mean, and it’s not my intention here to criticise the overall thrust of the TEF. However, the CAH is not a subject classification system, nor does it define “subjects”. That’s the new Higher Education Classification of Subjects (HECoS). The CAH does what it says on the tin; it aggregates subjects of study, primarily the HECoS ones, but it can also be used to aggregate JACS3 stuff. So, CAH level two is a set of subject groupings, rather than subjects of study themselves.
I’ve been handling, defining and managing courses information for my entire career – and I’m nearing retirement, so that’s a fair few years, and, declaring an interest, I was part of the Cetis team that helped design HECoS and the CAH. I wrote in an earlier Wonkhe article: “Analysis of TEF or LEO is always limited by the accuracy of classification, which will be rather better using HECoS compared with JACS. Analysis also rests on the way in which subjects are grouped, clearly separated when grouping HECoS subjects into the CAH.”
Part of Cetis’s work on HECoS in the Higher Education Data & Information Improvement Programme (HEDIIP) was to address subject-based analysis in relation to adoption and governance of HECoS and the CAH. In this work we looked at what constitutes a subject grouping and the way joint, combined and interdisciplinary studies might be handled. In designing the CAH, we bore in mind that practical aggregations had to have rules that covered attributes over and above subject of study, such as subject weightings and additional attributes for things like teacher training – in this last, you have both “teaching” as a subject of study, and the subject of study content of the teaching subject.
It’s pretty easy to see that each entry in CAH2 is not a “subject”. Taking a look at the labels, we have “pharmacology, toxicology and pharmacy”, obviously an aggregation of three subjects; “subjects allied to medicine”, patently not a subject in its own right; and agglomerations such as “health and social care” and “humanities and liberal arts”. There’s also a recognition of the difficult area of balanced, combined and interdisciplinary subjects, which are to do with programme design rather than subject of study, where we have “general and others in sciences” and “combined and general studies”.
Questions to ask
Will the data identify and relate to “subjects”? One of the criteria for selecting CAH2 as the appropriate level of granularity for the subject-level TEF was that the “subjects” will “be clearly understood by students making choices about what to study”. In many cases, this will be unproblematic; for example, “nursing”, “chemistry” or “economics”. Others, particularly non-school subject groups will be more challenging. I have to look up the HECoS subjects myself for some to be clear about their content; for example, “subjects allied to medicine” compared with “health and social care”, or “engineering” compared with “technology”.
A second criterion was “group courses that are likely to be reasonably similar in teaching quality (although there will always be courses that straddle subject boundaries)”. I’m unclear that it is logical to assume that the teaching quality across different programmes in the same grouping will be “reasonably similar”. If they’re taught in the same department that might be justifiable, but across different departments, which is fairly common in the sector, not so much.
A current criticism of the KIS is that the subject groupings can aggregate relatively poor performing programmes with relatively high performing ones with the complaint that “x programme is dragging down the ratings of y programme”. CAH2 has thirty five subject groups, whereas the KIS has over a hundred, so CAH2 would increase the weight of this problem. In addition, there are some cognate subjects that are frequently delivered in very different ways in the same provider with the likelihood of different teaching quality assessments; for example, “human geography” versus “physical geography”, both in “geographical and environmental studies”.
The consultation document properly raises the thorny issue of general, combined, multi-subject, interdisciplinary provision. As my wife points out, only I would care that they’ve equated “interdisciplinary” – relating to more than one branch of knowledge, integrated programmes – to “joint and multi-subject combined courses” – programmes combining two or more subjects taught largely separately. Interdisciplinary programmes are an acknowledged problem in subject aggregation, and HESA has grappled with this within JACS for some years, with some success via weighting of subjects.
In HECoS we introduced some broad subjects, such as “natural sciences” and “humanities”, as an aid to classification. However, there is a bit of a difference between an interdisciplinary programme, potentially likely to have a singular teaching quality, and a combined subject programme drawn from two separate departments. It may be important to bear in mind these differences, though I can’t offer a simple and problem-free solution.
Policy solutions, policy problems
I strongly believe that the new HECoS classification system and the CAH make it easier to manage aspects of policy like the TEF. At least we can focus our efforts and thoughts on what the subject groupings mean and how they are to be used, and we can move away from manipulating the subject codes attached to the course in order to game the system. As a final illustration of the difficulty and because I work with the Open University on subject classification, please advise where to put the OU’s BA Open degree!
One final observation: I’ve managed to avoid mentioning any codes at all. For those interested in the codes, please take a look at #hecoscodeoftheday, lovingly nurtured by the excellent Dan Cook at HESA.