Guidance on the use of scaling was developed by a working group of the Board of Undergraduate Studies and approved by the Senate at its meeting on 5 July 2017, for implementation in 2017/18. The Guidance on Scaling is intended to complement the University's Guidance on Moderation and applies at both undergraduate and postgraduate levels of study.
For the purpose of this guidance, scaling is defined as:
A process by which a set of marks is raised or lowered (normally to an agreed mean), in order to properly calibrate the performance of the cohort in terms of the achievement of learning outcomes and grade descriptors.
Scaling may be contrasted with normalisation which is defined as follows:
The variation of a set of marks in order to fulfil a pre-defined expectation, without consideration of the performance of the cohort.
In some contexts, the terms scaling and normalisation may be used interchangeably and the two functions may use the same mathematical methods. The key difference between the two functions is that for normalisation, marks are changed without consideration of the performance of the cohort. There is no justification for varying marks simply to meet a pre-defined expectation, without consideration of the qualitative performance of the cohort. Accordingly, normalisation, is not permissible in any circumstances.
The proper objective of scaling is to achieve a set of marks for a cohort of students which reflect the extent to which each student has achieved the relevant learning outcomes and has met the criteria set out in the grade descriptors.
Scaling may be used:
- For elements of assessment which are marked according to objective criteria, and which do not allow for the exercise of discretion in marking.
- To achieve consistency between the grades awarded in different modules.
- To avoid students being either advantaged or disadvantaged by choice of modules.
- Where an academic judgement is made that an assessment was either too difficult or too easy.
- Where scaling is necessary to achieve an appropriate calibration of the grade boundaries.
- Where an examination has denied candidates a fair and realistic opportunity to fully demonstrate their ability, for instance where the examiner has misjudged the work that can be completed in the prescribed time.
Scaling should not generally be used for subjectively marked assessments where the judgements intrinsic in the marking process allow for proper calibration. Scaling should also not generally be used where students taking a particular module or assessment have been prejudiced by a problem over which they had no control, such as the illness of a key teacher. Normally problems such as this should be remedied by other means, such as exceptional resits, or by exam boards discounting the compromised assessment when calculating final grades. If it is proposed to use scaling as a remedy in such circumstances, this must be authorised by the examination board, after consideration of a fully reasoned academic justification.
Examiners may be prompted to consider scaling by a number of factors, including a module cohort having an unusually high or low mean mark, or the marks being inconsistent with marks achieved by the same students in other modules, or inconsistent with the marks achieved in previous years, or by other anomalous mark distributions. However, such factors should not automatically lead to scaling in the absence of an academic judgment that the raw marks do not appropriately reflect the standards achieved by the cohort.
Scaling should not be the norm for any module. Examiners should endeavour to avoid the need for scaling by setting exams in which the achievement of a boundary mark reflects the level or knowledge and skill appropriate to the relevant grade.
Scaling is a stage in the marking process, and is carried out under the authority and supervision of the examination board. In advance of the exam board, authority for scaling should be given by the chair of the board, or the head of department or another appropriate responsible office holder within the department. Authorisation may be given on a continuing basis where appropriate.
In departments in which scaling may be used, this should be explained in course handbooks, the departmental website, or other publications for students. The information provided to students should explain, in broad terms, what scaling is, why it may take place, and the procedural safeguards which apply. The information should also indicate those modules in which scaling may be used. Some departments currently publish the algorithms used for scaling. This is recognised as good practice but is not a requirement. Information on scaling must be accessible to ensure that students understand the methods used.
Where scaling has been authorised in advance of an examination board, the marks which result from the scaling process are provisional until confirmed by the external examiner and the board. The external examiner and the exam board must be provided with full information, including the reason for scaling, the method of scaling employed and the raw marks as well as the adjusted marks.
Where a module involves more than one element of assessment, scaling should be applied, if appropriate, to each element of assessment, rather than to the aggregate mark.
Departments which anticipate using scaling, should establish appropriate processes which may take into consideration, as a factor to guide academic judgment, expected mark distributions (based on historical precedent and evidence relating to the abilities of the cohort).
Departments should retain a full record of the scaling process including the raw marks as well as the final marks.
Marks published to students should be final marks, after scaling or other moderation processes. Generally, departments should avoid releasing raw marks to students if these are to be scaled, to avoid confusion. If, in any circumstances, raw marks are released to students where there is a possibility that these may be subject to scaling, this should be clearly explained at the time when the marks are released.
Academic judgment must be used to determine the appropriate points for grade boundaries. Classifications must retain their qualitative meaning and scaling must not be used to award particular marks unless the candidates in question have demonstrated the appropriate level of ability.
An exercise in scaling must preserve the ranking of individuals within the cohort. Zero must always be scaled to zero and 100 must always be scaled to 100.
Scaling should be used with restraint and only to the extent that it is justified by the evidence.
Generally, scaling should not be applied to generate fail marks for students who would have passed on the raw marks. If scaling is used in this way, there must be a very clear academic rationale.
Scaling must not be used to reduce the marks of a particularly able cohort, or to enhance the marks of a poorly performing cohort.
In applying scaling, departments may deal separately with unusual outlying marks in order to avoid distorting the process.
A variety of scaling methods may be employed provided that they comply with the general principles set out above. It is important that the method used can be implemented simply and quickly, taking into account the resources available. Acceptable methods include the following.
Monotonic, piecewise-linear transformation preserving 0 and 100 and controlling the location of important mark boundaries based upon academic judgement and input from a two way analysis.
Departments may vary the precise method used.
A combination of adding or subtracting a proportion of marks, or adding or subtracting a whole mark.
Subject to the principle that raw marks of zero and 100 must be retained.