Nearly half of students with a longstanding physical or mental health condition say they perform worse in traditional exam halls than in online or remote assessments, according to new UK research — raising fresh concerns about whether current assessment models are fit for an increasingly diverse student population.
Exclusive data from The Invigilator, a global assessment and exam integrity platform supporting almost one million students worldwide, suggests that in-person exams are acting as a structural barrier for disabled and neurodivergent learners, rather than a neutral measure of academic ability.
The findings come as demand for special educational support reaches record levels across the UK, from schools through to higher education.
Exam environments, not ability, driving poorer outcomes
The research shows that 46% of students with a longstanding condition say they perform worse in traditional exam halls compared with online or remote settings. This compares with 32% of students without a condition, indicating that the physical and sensory demands of exam environments may be disproportionately affecting disabled learners.
More than a third (38%) of students with a longstanding condition also said they would have performed better at school if they had been allowed to study at their own pace — pointing to the impact of rigid assessment formats on students managing fatigue, anxiety, chronic pain or neurodivergence.
Practical barriers also remain significant. One in three students (33%) reported that the cost or distance of travelling to education has limited their ability to participate fully.
Record SEND demand highlights system strain
The findings sit against a backdrop of rising demand for additional support across the education system. Department for Education figures show that the number of pupils with an Education, Health and Care Plan (EHCP) reached 483,000 in 2025, an 11% year-on-year increase.
In higher education, disabled students now make up around one fifth of the student population in England, yet continue to report a weaker overall experience than their non-disabled peers — including lower satisfaction and greater barriers to assessment and progression.
AI emerging as an accessibility tool, not a shortcut
The research also challenges assumptions about the use of artificial intelligence in education. Students with a longstanding condition were found to be more than twice as likely to use AI tools to support accessibility needs — such as transcription, dyslexia support or structured explanations — compared with students without a condition (28% versus 11%).
Importantly, these students were less likely to say they relied on AI to the extent that they would struggle without it, suggesting targeted, assistive use rather than dependency.
Nearly four in five students with a longstanding condition (78%) said their institution allows or encourages responsible AI use during assessments, compared with just 35% of students without a condition.
Assessment design under growing scrutiny
The findings arrive as AI-supported learning tools move further into the mainstream. Ministers have previously indicated that AI tutoring could be introduced across UK schools by the end of 2027, with estimates suggesting up to 450,000 disadvantaged pupils could benefit.
As this shift accelerates, the focus for institutions is increasingly on assessment design — how to balance flexibility and inclusion with academic rigour and integrity.
Nic Riemer, CEO of The Invigilator, said the data shows the exam environment itself is becoming a hidden barrier:
“When nearly half of students with a longstanding condition perform worse in exam halls, it raises serious questions about whether traditional formats are measuring knowledge — or endurance. Assessment design is now a core inclusion issue, not a technical one.”
Rethinking exams without lowering standards
The Invigilator’s digital assessment model is designed to support remote, hybrid and venue-based exams, including in low-bandwidth or disrupted environments. The platform uses identity verification, audit trails and exam controls to maintain standards, while allowing students to access assistive technologies where needed.
For policymakers, educators and employers alike, the findings add to growing evidence that inclusion cannot be bolted on at the margins of assessment systems.
As disabled and neurodivergent students make up a larger share of the education pipeline — and future workforce — the question is no longer whether traditional exams disadvantage some learners, but how quickly institutions are willing to redesign assessment to reflect that reality.