Last Updated January 2024
Department of Computing Policy on
Moderation of Module Marks (“Scaling”)
1. General Principles
1.1. Basis
The Regulations for Taught Programmes of Study 2023-24 state (12.26) “The final stage of internal
moderation is to ensure that the scale, range and standards of marking are appropriate.” and (12.28)
“Statistical tools and techniques may be used to moderate assessment practices in order to address
underlying issues. Such techniques may not be used to change the marks of individual students however
they may be used to apply consistent changes to all work submitted for the assessment task by enabling
the department to identify assessments where the marks profile is out of line with departmental norms.”
1.2. Aims
The aim of this policy is to set out how the department will use “scaling” to moderate the difficulty of
the assessment of:
1. Any module, including core modules, in order to maintain the integrity of the overall degree
programme and ensure fairness and consistency of outcomes between different cohorts.
2. Different elective modules taken by members of the same cohort, where there is potential for a
student’s module selection to affect the outcome of their degree in comparison to their peers.
2. Departmental Norms for Module Marks Profiles
The marks of a module are outside departmental norms if they provide clear evidence that the
assessments set for all candidates were either too difficult or too easy to reflect their ability accurately.
2.1. Module Average
The module average (the mean of the weighted total mark for each student) will be used to evaluate the
overall level of difficulty. A module will normally have its marks scaled if it has been taken for credit by
at least 10 candidates and:
• it is a Level 4 module, and the average is not within the range 50-80%; or
• it is a Level 5 module, and the average is not within the range 55-75%; or
• it is a Level 6 or Level 7 module, and the average is not within the range 60-70%.
2.2. Exceptions
A module will not normally be scaled, regardless of its average mark, if any of the following apply.
i. There is a significant difference in the work undertaken by each candidate, e.g. for an individual
project. Such modules should have separate moderation processes focusing on ensuring
consistent and accurate application of the marking criteria.
ii. The assessments set are substantially the same as those set in previous years, e.g. for a
laboratory module, and the average has historically been within acceptable norms.
Last Updated January 2024
iii. The module is elective and the candidates that have selected it have a significantly better or
significantly worse average performance than their overall cohort.
3. Scaling Methods
Scaling will be used conservatively to bring the module average closer to or within the given normal
range, as given under 2.1 above.
1. If the module average is below 60% then marks will be added to the total of all candidates in
increments of 0.5 until the average is at least 59.5%. Further increments to all marks may be
made in order to reduce the number of failed candidates. The module average will be increased
to a maximum of 63%. Marks will be capped at 100%.
2. If the module average is greater than T, where T is the top of the range given in 2.1 above, then
the total mark of all candidates will be multiplied by a factor such that:
Scaled Average = (Actual Average + T) / 2
3. Marks below the pass mark will not be reduced and students who have a passing unmoderated
mark will not have their mark reduced to a failing mark.
The additive scaling method (1) provides proportionally greater benefit to students at the lower end of
the mark range. The multiplicative scaling (2) has the effect of halving the difference between the
outlying average and the target upper bound (see 2.1). It also has a lesser effect on students at the
lower end of the mark range than those at the upper end.
4. To Note
The scaling methods described in Section 3 are substantially the same as those used in 2021-2022.
However, the UG Pre-Board minutes state that “No scaling was applied to the first- and second-year
exams”. (In fact, one first year module was scaled up because too few courseworks were set.)
