English writing assessment research

Beginning in 2024-25, we tightened the conditions of assessment in National 5, Higher and Advanced Higher English writing to support teachers and lecturers in authenticating learners’ work and to improve reliability. Learners are now required to complete the first draft of their portfolio during class time under supervision, enabling teachers and lecturers to review this work before learners complete the final piece under some supervision and control. The final piece is then submitted for external assessment.

We made the changes in response to feedback from English educators that it could be difficult to verify learners’ work as their own. Feedback also suggested that it was becoming more challenging to detect plagiarism, due to the rapid development and accessibility of generative artificial intelligence (AI) tools.

In response to these challenges, and within a rapidly evolving landscape, we have completed a suite of inter-related research on the assessment of writing:

Research questions

The 'Evaluation of the changes to English writing assessment' explores whether the changes to the conditions of assessment in English writing are supporting teachers and lecturers in authenticating learners’ work, how centres and learners are managing the changes in practice, and whether they are experiencing significant or unintended challenges.

'Approaches and considerations for the assessment of writing: A review of literature from 2013-2026'  examines the different models and practices that exist for the assessment of writing as a skill. It explores how these models are being adapted in response to the emergence of generative AI and assesses whether robust or transferable elements of best practice can be identified.

'International approaches to the assessment of writing in the age of AI: Considerations for Scotland’s National Qualifications' examines how the implications of AI for writing assessment are being addressed across a range of UK and international settings.

Methodology

For the evaluation, we took a mixed methods approach. Large-scale surveys of more than 1,000 learners and 400 educators gathered both quantitative and qualitative data. These were followed up by 10 interviews with educators and a focus group with learners to explore the issues in more depth.

The literature review summarises main approaches to the assessment of writing skills within the past decade, drawing on evidence from early primary through to higher education contexts across multiple subject areas. Approaches can be grouped into four main categories: criterion-based frameworks, comparative judgement, content and authorial voice, and scenario-based assessment. The paper also focuses on how the proliferation of generative AI is being considered and accounted for in the development and assessment of writing skills.

The desk-based study adopts a comparative qualitative policy analysis approach, drawing on jurisdictional policy, strategy and guidance, and awarding body documentation. Comparator jurisdictions were selected either because they are natural comparators (that is, the other UK nations) or because they combine a pro-AI policy approach with curriculum and/or assessment features relevant to Scotland’s current or reformed qualifications.

Key findings

These papers provide insight into a central challenge in contemporary assessment: balancing authentication, reliability, and validity in a rapidly evolving AI-influenced landscape.

The findings highlight that tightening assessment conditions alone does not resolve the underlying tensions in the assessment of writing: competing demands of securing authentic learner work, ensuring fairness across centres, and maintaining valid, meaningful assessment, particularly in an AI-rich environment.

The suite of papers therefore offers evidence on how policy changes are experienced in practice, while highlighting the potential strengths and limitations of different approaches. In particular, the evidence suggests that no single assessment model fully resolves the competing demands of authentication, reliability, and validity.

The findings suggest that in relatively open assessment models such as coursework and portfolios, tightening conditions can address some concerns around authentication and reliability as AI continues to develop, but it may introduce new challenges around educator workload, assessment validity, and so on.

Implications

The recent advances in AI have significantly intensified longstanding challenges in the authentication of learners’ writing. This is not just an issue for Qualifications Scotland, but a system-level challenge faced by awarding bodies, schools, colleges and higher education institutes worldwide.

In Scotland, the recent tightening of writing assessment conditions was aimed at helping teachers and lecturers to authenticate learners’ work. We acknowledge the challenges and workload implications associated with this approach, and place high value on the partnership work done by and with teachers and lecturers.

Consistent with wider research evidence, teachers and lecturers are best placed to authenticate learners’ writing as they are familiar with learners’ work and abilities.

Moreover, the present arrangements for portfolio as coursework offer significant strengths. Evidence from the marking process confirms that learners take full advantage of the wide personalisation and choice inherent in the approach, enabling genuine creativity, personal expression, and the exploration of meaningful ideas and topics.

The emphasis is firmly on writing as a process, from initial ideas to research, drafting, reflecting on feedback, and then re-drafting — which closely reflects authentic real-world writing practices and underpins the validity of the assessment.

While some educators in our evaluation suggested further tightening of conditions, including a potential move to exam-based assessment, the evidence indicates that this would likely result in the loss of key strengths. In practice, this would potentially mean reduced learner choice, greater reliance on rehearsed responses, and a narrowing of the construct being assessed, thereby compromising validity. Similarly, further prescription around conditions of assessment may risk constraining the writing process, increasing the delivery burden, and introducing additional timetabling and resource pressures without fully resolving authentication challenges.

Next steps

We recognise that the challenges presented by AI, particularly in relation to authentication, are complex and evolving. We will continue to monitor developments in this area and their implications for qualifications and assessment.

In the short term, our priority is to maintain stability for centres and learners, while continuing to build the evidence base. We do not envisage changes to the current position in the immediate future, as the balance of evidence suggests that the strengths of the current approach, and the stability it provides, outweigh the identified risks.

However, this remains a fast-changing environment. Ongoing research, stakeholder engagement, and international comparison will be critical to informing future decisions. Consideration of writing assessment will also form part of wider curriculum and assessment reform activity.

Access the full reports and accessible summaries

This summary is intended for general audiences and simplifies complex research. Please refer to the full reports for detailed methodology, complete findings, and technical discussion.