Many of the assessment design conversations that I am involved in at the moment are framed as optimisation problems. How do we improve security without harming experience? And how do we increase accessibility without compromising validity?
These are valid questions, but potentially a little misleading.
In reality, these are not optimisation problems, they are trade-offs, and not necessarily ones that are neatly resolved.
Every assessment is ultimately a series of decisions about what we choose to prioritise, and what consequences we are willing to accept as a result. The reality is that most of these trade-offs are already being made. The question is whether we are making them consciously.
Assessment systems today are expected to do more than ever. They must be secure enough to withstand scrutiny, accessible enough to be inclusive, intuitive enough to be fair, and robust enough to measure something meaningful.
Individually, each expectation is entirely reasonable. Taken together, however, they can pull assessment design in different directions.
And if we do not address these tensions, they don’t simply disappear, they reveal themselves in the candidate’s experience, in appeals or contested results, and they can lead to a wider mistrust of our assessment processes.
Whose definition of “normal”are we designing for?
Many of us have had to develop high security environments to detect suspicious behaviour. But the definition of “suspicious” can be narrower than we think.
For some candidates, particularly those with disabilities or accessibility challenges, behaviour that falls outside the norm is not an anomaly, it is a necessity. Assistive technologies, involuntary movements, or the need to step away from the screen can all trigger flags in tightly controlled environments.
At that point, the system isn’t just protecting integrity, it is making a judgement about what “normal” looks like. And when viewed from that perspective, it does pose the question: Whose definition of “normal” has been embedded into the assessment process?
The cumulative weight of assessment security
Security controls are rarely experienced individually. Candidates experience them cumulatively.
On its own, browser lockdown may be fine. Identity verification alone may be fine. Restricted navigation alone may be fine. But together they create a different assessment experience.
Each intervention is defensible. But combined, they can fundamentally change the assessment experience.
Under pressure, every bit of friction matters.It increases cognitive load, heightens anxiety, and can shift focus away from the task itself.
So, as part of our design process, we need to ask ourselves, at what point does securing the assessment begin to interfere with the very performance we are trying to measure?
When accessibility risks changing the construct itself
Accessibility is often framed as an unambiguous good.
That is true.
But in practice, it can also force assessors to make difficult design choices.
Adapting content for accessibility may involve simplifying or removing elements that are not easily translated across formats. In some cases, those elements are peripheral, but in others, they can be central to the skill being assessed.
If, for example, a task is designed to measure the interpretation of complex visual information, what happens when that visual complexity is removed?
We may improve access, but are we also changing what is being measured? And if we are, can we still claim that the assessment is measuring the same thing?
Authentic assessments create their own barriers
I have, for many years, been a passionate advocate for more “authentic” assessments, assessments which more accurately mirror the real world.
Whilst this is the way forward for many organisations as we seek to adapt to the digital environment that we now operate in, such assessments can also risk introducing their own barriers.
It is critical that we design these assessments carefully to ensure that we are not asking candidates to navigate unfamiliar interfaces, manage multiple inputs, and interpret system behaviour alongside the task itself.
While we have a responsibility to prepare candidates for the environments in which they will ultimately work, we also have a responsibility to ensure that we are measuring capability rather than familiarity with a particular interface, level of digital confidence, or tolerance for friction.
Standardisation was never the whole story
For decades, standardisation has been the foundation of fair assessment: the same test, under the same conditions, for everyone.
But accessibility challenges that model. Meaningful inclusion often requires flexibility: adjustments to timing, format or interaction. And the more flexibility we introduce, the harder it becomes to demonstrate that outcomes remain comparable.
It is here that we can often put strain on our processes and on our teams that support them. Not because flexibility is wrong, but because many of our traditional assumptions about fairness were built around uniformity rather than equity.
As if these tensions were not already complex enough, AI is adding another layer to the discussion. Measures designed to prevent inappropriate AI use may increase surveillance, restrict candidate behaviour or alter assessment design. At the same time, failing to address AI may undermine confidence in assessment outcomes.
Again, there is no perfect solution, only a series of trade-offs that need to be made consciously and transparently
Where do we go from here?
Nothing here is new.
And none of it means we should abandon security, accessibility, usability, or validity. Quite the opposite. The challenge is recognising that these objectives do not always align perfectly.
There will be times when strengthening security introduces friction. Times when improving accessibility raises questions about comparability. Times when increasing authenticity creates new barriers for some candidates.
The challenge is not to eliminate these tensions, but to acknowledge them, examine them, and make conscious decisions about how they are managed.
Because every assessment reflects a set of priorities.
Trade-offs are inevitable. We simply have to ask ourselves whether we have been deliberate enough in making them.
Hear more at the International e-Assessment Conference 2026
These questions are worth exploring in the open, and that is exactly what the Accessibility Special Interest Group panel is designed to do.
Robbie will be joining fellow assessment professionals on the Main Stage on 9 June for The Future of Assessments: Why Integrity will be Pivotal to Fairness and Inclusion, a discussion on how integrity can support, rather than hinder, inclusive assessment design.
