Research on how GCSE grade boundaries in Wales are set

“In January 2024, we published a report from the Oxford University Centre for Educational Assessment (OUCEA) on how GCSE grade boundaries in Wales are set. The report described the approach to grading in Wales as attainment referencing.  We also released a statement about our approach to grading and that this approach would continue to be used for the introduction of new GCSEs in Wales, being phased in from September 2025. We have now published another research report from the OUCEA, looking at how an approach called criterion referencing could be researched for qualifications like GCSEs. Although we have no current intention for criterion referencing to be used to grade GCSEs, the approach is used for other qualifications, and the topic is likely to be of ongoing interest in discussions about grading. 

The difference between attainment referencing and criterion referencing
The use of ‘referencing’ indicates that grading decisions are made with reference to evidence and expectations.  The first report we published explains important differences between attainment and criterion referencing. In attainment referencing, experts on an awarding committee refer to both statistics and qualitative evidence (such as the exam scripts produced by learners) when setting grade boundaries. Attainment referencing is a mixed methods approach, using both quantitative and qualitative evidence.  

Criterion referencing relies on expert judgement by teachers or examiners, who reference written statements (the criteria) to decide how an assessment performance by a learner should be graded. Criterion referencing is a more qualitative approach than attainment referencing. As discussed in the first report, these differences in process link to differences in the intended meaning of the grades.  

Why we have researched how criterion referencing could be used for GCSEs in Wales
Some within education have argued for this sort of approach to be used. As the qualifications regulator for Wales, we think it is useful for us to both engage with these discussion and viewpoints, while carefully examining alternatives in an evidence-based way. 

In the case of criterion referencing and GCSEs, this is particularly important given that when GCSEs were first designed in the 1980s there was an intention to use a form of criterion referencing for grading. However, the approach failed during qualification development (see page 85 of the main report for more details).  Doing this work also means that, should there be a desire to change policy in future, we will be well prepared to understand the implications and challenges that would need to be faced.  Exploring an alternative approach also helps us think about the strengths and weaknesses of the current approach to grading GCSEs in Wales. 

If we were to use criterion referencing for qualifications like GCSEs, research would be needed to help make evidence-based decisions. The report we have published discusses many ideas for that research. The reports from the project showed us that a substantial change to grading standards – such as a switch to criterion referencing for qualifications like GCSEs - would have significant implications for the education system, including impacts on the role of teachers. 

There would also be impacts on the social functions performed using qualification grades. These social functions include evidencing the results of education, helping with school self-evaluation and managing the transition of learners across different stages of education and into the labour market. A different standards approach is likely to result in outcomes that have different properties, which in turn impacts on those functions.

Further resources from the project can be founded on OUCEA’s project page. These include videos explaining why GCSE grade boundaries change and the relationship of the attainment referenced approach to criterion referencing and norm referencing. I’d like to thank our partners in Oxford University for working with us on this project. Read the report here.