Invisible in the data: how London universities are failing disabled students through ambition without accountability 

Author:
Ella Malik and Dr Helen Coulshed
Published:

This blog was kindly authored by Ella Malik, Research Assistant, and Dr Helen Coulshed, Associate Vice Dean (Assessment and Quality Assurance) in the Faculty of Natural, Mathematical and Engineering Sciences and Senior Lecturer in Chemistry Education, both at King’s College London. 

In 2025, the National Students Survey (NSS) recorded lower positivity scores amongst disabled students across every domain. That same year, an Office for Students (OfS) workshop found that only 15.2% of providers systematically evaluate the effectiveness of reasonable adjustments, while 60% rely on informal or ad hoc processes. Without systematic evaluation, institutions cannot determine whether their support is effective, and outcome gaps are likely to persist undetected.   

To understand how universities are responding, we audited Access and Participation Plans (APPs) from 16 London-based higher education institutions, spanning pre-92 and post-92 providers, specialist institutions, and research-intensive universities. 

Why Disabled Students Are Hard to See in the Data 

In 2023, the OfS reformed the APP framework around a risk-based model. Under this model, institutions are required to identify specific outcome gaps, provide evidence of their causes and set proportionate targets in response – replacing broad aspirational commitments with measurable ones. For disabled students, however, this model has structural limitations. 

Unlike other underrepresented groups, disabled students often require a diagnosis to be recognised in institutional data and access support. Diagnosis rates vary significantly by gender, income, and ethnicity, and students who are awaiting assessment or face financial or cultural barriers to obtaining one may not appear in the figures that drive institution planning. 

Aggregated reporting compounds this. When all disability types are grouped into a single category, differences between subgroups, such as between students on the autism spectrum and students with ADHD, can be obscured. An institution may show no measurable gap at the aggregate level while significant disparities exist within it. In this way, the absence of a recorded gap does not confirm the absence of disadvantage. 

Patterns Across the Sample 

Most institutions in our sample acknowledge disability as a concern: 13 of 16 identified it as a risk factor, and 15 of 16 included at least one linked intervention strategy. Two institutions did not designate disability as a prioritised risk, but both maintained disability-linked provision. This is consistent with the OfS whole-provider approach, which recognises that small cohort sizes can make disadvantage statistically invisible and encourages institutions to embed inclusive provision into standard practice without requiring a formal risk designation at every lifecycle stage. 

However, this acknowledgement did not consistently translate into commitment. Only 8 of 16 institutions set numerical targets for their disability related intervention strategies. The remaining eight classified disability as an ‘area for monitoring’ or set no targets. Five institutions had already disaggregated disability types, identified a specific condition with a significant gap at a lifecycle stage, and still set no target to reduce it.  

To analyse variation across the sample, we developed a Granularity Index with three tiers. Tier 1 institutions set intersectional targets combining disability type with other characteristics such as age or ethnicity, Tier 2 institutions set targets for a single specific condition, and Tier 3 institutions used aggregated disability data. Tier 1 and 2 institutions linked their disaggregated data to condition-specific support. One Tier 1 institution, for example, set intersectional targets for minoritised ethnic students with mental health conditions, and developed specific interventions including a psychoeducation programme and anti-stigma workshop aimed at increasing help-seeking behaviour. Tier 3 institutions focus instead on systemic change through broad, aggregate targets. 

The differences between Tier 1 and 2 universities and Tier 3 institutions reflect a strategic choice between condition-specific support and systemic inclusive-by-design frameworks. But this raises a practical question: inclusive-by-design approaches benefit students without a formal diagnosis who would otherwise be invisible in the data, but applying them without disaggregation can mask the very disparities they are meant to address. How can institutions reconcile these two facts? 

What Effective Practice Looks Like 

The variation across the sample points to two recurring limitations: students without a formal diagnosis are often excluded from support and absent from outcome data, and aggregated disability reporting conceals differences between subgroups. The strongest APPs addressed both directly. 

One institution removed the requirement for medical or diagnostic evidence to register for bespoke disability support, reducing the barrier for students awaiting assessment or unable to obtain a diagnosis. Critically, this did not come at the cost of analytical precision: UAL continues to collect condition-specific outcome data, meaning broader access to support does not prevent the institution from identifying where targeted intervention is still needed. 

Disaggregation by disability type alone, however, does not account for disadvantage that compounds across multiple characteristics. Another institution addressed this by applying logistic regression at each lifecycle stage to identify characteristics independently associated with risk and isolating the specific combinations that predict disadvantage. Mandating this kind of intersectional analysis as a standard component of APP development would shift the burden of proof: rather than waiting for compounded disadvantage to appear in headline figures, institutions would be required to actively test for it.  

Together, these examples illustrate that improving outcomes for disabled students is not solely a question of what support is offered, but of how precisely institutions can identify who needs it and why. That precision depends on the quality of the analytical commitments institutions make, and this is where the current framework has a measurable gap. 

Classifying disability as an ‘area for monitoring’ is not inherently problematic, but it requires specification. Institutions should be able to state what data is being collected, what threshold would trigger a target commitment, and what evidence would indicate that no further action is required. Without this, monitoring records awareness of a potential problem rather than a structured plan to address it. For students whose disadvantage is already difficult to see in the data, the difference between monitoring and acting is not an administrative one. 

Get our updates via email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Comments

Add comment

Your comment may be revised by the site if needed.