This Bristol UCU paper/blog post provides 11 principles to guide academic workload modelling across the University of Bristol.
With very few exceptions, academic Schools now use an explicit, formal workload model. There are commonalities across these, but also differences. It is rare that there is one model in a Faculty. Yet important workforce planning occurs at the Faculty level, without any real sense of how workload varies between Schools. At university level, decisions are similarly made about the allocation of posts without a real sense of comparative workload between Faculties and Schools.
While Bristol UCU welcomes increased attention to Staff Student Ratios (SSRs) in recent years, these do not provide an adequate proxy, given that different subjects have differing requirements. It is hard to understand how the Establishment Review Group can proceed, other than by induction from past patterns, given the lack of robust data on workload. Attention to Russell Group median SSRs is positive in that it has brought resource to hard-pressed departments, but better modelling of workload remains essential.
From a university perspective, closing the gaps in workload model provision, and establishing common principles for workload modelling can inform good workforce planning. From a Bristol UCU perspective, a common approach is equally required. All Pathway 1 staff need time to produce research; all pathway 3 staff need time to do scholarship. Staff are assessed against a common set of university criteria for promotion and progression. Natural justice and the ambitions of the university’s Vision and Strategy likewise point to common standards (eg 40:40:20) that should apply across all six Faculties. Common principles in workload modelling will enable the university to identify areas where workloads preclude staff spending appropriate amounts of time on research and/or scholarship. If done realistically, it will also enable an informed critique of existing structures and processes based on a serious reckoning of the time spent on various managerial and administrative tasks.
These points are not new. What has hampered discussion in the past, however, is framing the issue in terms of a ‘single workload model’. This terminology is unhelpfully ambiguous, and immediately creates understandable concerns about the implementation of a ‘one size fits all’ solution. We need to distinguish between a common platform, common principles, and a common model. If common model refers to a single set of items with precisely the same weighting applied across all subjects, there are good reasons to reject this. There are real differences between PhD supervision in the Arts and PhD supervision within a research group in the hard sciences, not least in terms of the relationship between the PhD supervisor’s own research and that of the PhD student. This is why workload models in the Arts tend to weigh PhD supervising more heavily than those in the Sciences. A common platform, such as the software Simitive provide, allows considerable flexibility in modelling. Currently, our luxurious variety of workload models largely sit on a common platform, namely Microsoft Excel. A common platform has no necessary implications for how modelling works.
Objections to a common model do not, however, hold for common principles of modelling. There are, for instance, very good reasons to use hours (or hours translated into credits as a means of turning 4 digit into 3 digit numbers) rather than to use a % model, as the latter says nothing about the actual or relative volume of work. This paper sets out a set of common principles that should inform workload modelling across the University of Bristol. The paper is informed by hundreds of conversations with academics across the University about workload and workload modelling, by a review of (anonymised) data from several of the workload models currently in use, and by UCU’s national work on best practice in workload modelling.
1. Workload models should measure time.
Time is what a workload model measures, not money. The weighting attached to tasks should be solely derived from the time taken to accomplish them. A good model does not ‘incentivise’ behaviours by weighing some tasks more heavily than others, regardless of the relative time required to complete them, on the grounds that some tasks are more profitable than others. We should not, for instance, weigh research less heavily within a model as it is less lucrative than teaching overseas students. This principle needs to be consistently upheld in modelling.
2. The currency of the model should be hours not percentages.
As noted above, models based on percentages simply fail to deliver key requirements of a workload model.
3. The hours assigned to tasks should be realistic.
Some workload models currently in use at Bristol do not adequately reflect the realities of work. This is evident, for example, in the time allocated for marking, which often does not align with pressures to provide better feedback for students. It is, of course, the case that individual academics even in the same subject area will vary in the time needed to perform certain tasks. Hours allocated should reflect the time needed for a competent member of staff do the task properly. This is best determined through discussion with staff. This approach is both rooted in the reality of how long work takes while also providing a useful yardstick to staff: if the marking is taking much less time than the model suggests (not in practice a common experience!) you are probably not doing it properly; if it is taking far longer, you may be providing more feedback than is actually useful to a student.
4. The aim is to capture the full workload.
A model that undercounts workload is not a good workload model. As well as realistically modelling time required to perform a given duty, the model should seek to capture the full range of duties, including research and scholarship. This does not mean that a model should seek or claim to be exhaustive: some important aspects of academic life, such as a student coming to see a member of staff outside consultation hours, cannot be predicted in advance, and the costs of monitoring this activity would be both prohibitive and undesirable. The workload model should include an allowance for this unscheduled activity of at least 160 hours per annum.
5. The workload model should be developed to the highest standards of EDI.
Staff often note the tendency of WLMs to undercount the time involved in certain activities: teaching; teaching management; personal tutoring. By contrast some work is rarely undercounted – consultation with staff suggests research management roles are usually appropriately weighted. There is here a gendered pattern: roles that are under-counted are those disproportionately undertaken by women. In building a WLM, robust scrutiny of both the categories adopted and the tariffs included from an EDI perspective is essential. This is itself an argument in favour of more comprehensive approach rather than heavily trading off comprehensiveness in favour of simplicity: adopting the latter strategy is more likely to undercount women’s work, and hence to perpetuate inequality and injustice.
6. The workload model should be transparent and shared amongst those whose workload it captures.
There is already some good practice at the University of workload data being shared at School level and made available to all those whose workload is included in the model. Where this has happened, the experience has been positive: the capacity to see the workloads of others in the School has driven up the quality of information in the model, made for greater equity in workloads, and reduced (at least somewhat!) ill-informed comment upon the workloads of others.
7. The details of costings (eg how many hours does it take to supervise a PhD student in Chemistry?) should be built from ‘the bottom up’ through discussion amongst staff in the relevant unit.
8. There should be a single model at an appropriate level of unit, which will be at least that of the School, but would better be that of the Faculty.
There should be a single model in a School, rather than multiple models. It is, though, both possible and preferable to have a single model across a Faculty. This does not mean, for instance, that there is no scope for acknowledging particular unit types that require additional teaching; this already happens where there is a Faculty-wide model (eg Arts). Given the role that Faculties play in determining staffing levels, a single model per Faculty would be the ideal. The fact that at least one Faculty (Arts), and a very diverse Faculty at that (consider the differences between Spanish, Philosophy, and Film) has implemented a single model gives the lie to claims that this is impossible. It might be that the model, even within a single School, weighs the same task differently: it might, for instance, be the case that within the School of Geographical Sciences some PhD supervision conforms to an ‘Arts’ mode, some a ‘Science’ mode; the model should recognise that difference. Any such Faculty-wide model would be derived through principle 7.
9. Models should explicitly include time for research (P1 and P2) and for scholarship/pedagogy (P3) in accordance with contractual expectations.
All P1 staff regardless of Faculty are expected to conduct research. Workload models across all Faculties and Schools for P1 staff should include hours for research that reflect these contractual demands. This is likely to be of the order of 40% (ie c. 600 hours), given the ambitions of the University’s strategy for research, and should be similar regardless of School/Faculty. All P3 staff are expected to pursue scholarship/pedagogy, and hours in WLMs should reflect this. This is likely to be in the order of 30% (ie c. 450 hour), given the contractual demands on such staff, and the ambitions of the University’s strategy for educational innovation.
10. Core research time should be treated as a single block of time.
Some existing models (eg Arts) largely treat core research time for staff as a single block. Others (eg SPS) build up core research time from specific tasks (eg writing a grant proposal; writing a paper for a journal). While the latter approach is appropriate to understanding the time involved in teaching and management work, the former better reflects the autonomy, and flexibility, inherent in academic research. It also reduces the set up and maintenance costs of running a WLM.
11. Buy out for research should not be secured by reducing research time for others.
Many of the most difficult conversations around workload modelling concern buy out, usually for research, of staff time, and how this should be accounted for in the model. External grants are only one source of money for research and are not a meaningful proxy for the quality or quantity of research undertaken by either an individual or a group. There is wide variance in the significance of research grants as a funding stream across different disciplines within the university. Core research time should not be restricted to those with active grants. It may, however, be appropriate to award grant holders additional research time beyond core time, as long as this does not drive up the workload of others to levels where ‘real’ research time is reduced below the level set out by the ‘core’ research tariff.