Initial question 1: How to share subjects, sub-disciplines and fields?

Archived Page.

This page related to stage 1 of the project, which concluded in September 2014.

Whether to share subjects is one of the fundamental decisions in the design of the new subject coding scheme. Read through the issue and the solutions, and let us know what you think the solution should be, and why.

The issue

However one divides up disciplines, subjects and fields, some of the lower level entries are not unique to one higher level one. Some subjects and fields simply occur in more than one discipline or even group of disciplines.

Part of the difficulty of dealing with such shared subjects is whether software as well as humans can handle the structure, and whether the same subject studied or researched in a different discipline can really be said to be synonymous. For example ‘statistics for archaeologists’ may have very different techniques and emphasis from ‘statistics for economists’.

Solution 1a. unique subjects

In 1a, no subjects or other terms are shared. One term has only one broader term to which it is related, and any overlaps are left as they are.


isolated hierarchies


  • simple
  • familiar to many users


  • misses a lot of generalisations and other information at lower levels
  • makes a for a voluminous, redundant coding system
  • potentially difficult classification decisions in multi-disciplinary areas

Solution 1b. Shared subjects

In 1b, subjects are shared equally between broader terms, and therefore occur in more than one hierarchy. For statistical purposes, a shared subject counts equally as a narrower term for both broader terms.

Unlike 1a, the term has to have exactly the same definition in both hierarchies, and remain so.

Double hierarchy

Double hierarchy


  • small number of codes
  • preserves links across disciplines


  • requires tight coordination across disciplines
  • requires sophisticated user interface design to avoid confusion and inconsistent use
  • does not deal with change over time well- either in the nature of disciplines, or policy decisions about them
  • may not be supported well by all systems

Solution 1c. Synonyms

1c is a subtle variation of 1b. Instead of being placed equally under two broader terms, a narrower term has only one ‘real’ broader term. In 1c, ‘Social policy and administration’ under ‘Social Sciences’ is a synonym for the real term which is placed under ‘Health and social care’. It can be used under ‘Social sciences’ for convenience, but its actual position is under ‘Health and social care’ and will only be counted there. Such reuse can be indicated in a note.



  • small number of codes
  • preserves links across disciplines


  • requires tight coordination across disciplines
  • may cause ownership disputes
  • requires sophisticated user interface design to avoid confusion and inconsistent use
  • does not deal with change over time well- either in the nature of disciplines, or policy decisions about them

Solution 1d. One flat list, many hierarchies

In 1d, the coding scheme is a finite, flat list of subjects. Many hierarchies can be composed on top of the flat list; one for each specific purpose or sector body. Examples include the JACS3 hierarchy, disciplines structures for marketing, subject clusters for funding or links that mark double honours degree programmes. By the same token, stakeholders can add or link up more fine grained subtopic or field vocabularies ‘underneath’ the scheme.



  • doesn’t require much coordination between different disciplines
  • simplifies classifying
  • offers significant flexibility in post coding analysis or policy making
  • de-couples the coding structure from policy decisions; what ‘counts’ as a hierarchy becomes a decision on rules, not a matter for re-coding.
  • appears to follow the realities of current coding practice


  • could be seen as a radical departure
  • less scope for fine-grained common coding


View more initial design questions on the Design page

We would like to thank those people who commented in Stage 1. We have taken your comments on-board. The project deliverable PD03, “HEDIIP NSCS Structure and Candidate Scheme” (available from the HEDIIP web site) shows how we settled on a design.


3 thoughts on “Initial question 1: How to share subjects, sub-disciplines and fields?

  1. This is a fairly fundamental issue, isn’t it? Are we classifying subjects of study, or the disciplines that study them? A given subject may be studied in many disciplines, but presumably it remains the same subject.
    Disciplinary categories are appropriate for staff (usually), but I would say are not suitable for subjects of study, especially not if the classification is to be used for research. Therefore a solution like 1d or 1c seems best. Staff can be assigned to disciplines by Cost Centre if we ever want to know that.
    However a completely flat list is going to present issues about the level it is flat at. A flat list suitable for coding modules, or research projects, would look nothing like the flat list suitable for coding courses.

    • I would like to see some empirical evidence on the level of granularity used by institutions when they code programme and modules currently. HESA may be able to help here. If, as has been suggested in other discussion groups, most HEIs do not use the richness and granularity of JACS3 when coding, then that raises a question about the usefulness of that level of detail. If, however, colleagues are applying very precise codes, especially when coding modules, then, are they doing that because they taking care to get the code exactly right or because they are using that level of granularity in their own analyses and need that fine level of detail.
      The NHS have, over a long period of time argued that the codes for nursing are not precise enough and threfore they are not fit for purpose. Do we know exactly what is wrong? Are there other areas where this can be argued?
      One of the reasons that I like the idea behind solution 1d is that you can design a coding frame that meets the majority of people’s requirements. Where there truly is a need for a more granular code, then you design a secondary structure, and in that subject area, everyone knows that they have to be more precise when applying the coding frame.

  2. I note that few respondents have felt the need to use an international classification but I suggest that the European Grid Infrastructure “Scientific Discipline Classification” ( could be used for classification of courses, skills, projects etc. Its almost identical to the OECD Frascati Fields of Science Classification (which despite its name does include Arts, Humanities and Social Science) except a) it makes 2 small changes at the 2nd level, b) adds a complete 3rd level, c) gets rid of all the ‘other’ classes in the 2nd level of Frascati. There are 6 top level classes, 38 second level and 333 at the 3rd level. Frascati FOS is a long-established and widely-used classification in Europe. In these days of shared research, teaching and publishing is it not possible to use an international discipline classification?

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s