Mental Capacity Assessment
Every tester completes a guided assessment before their first session. It verifies they genuinely understand what data we collect, how it is used, their rights, the risks, and that participation is freely chosen.
How the Assessment Works
The assessment is built into the onboarding flow. Every tester must pass it before they can access any test session. It is not a formality. It is a genuine check that the person understands what they are consenting to, aligned with the UK Mental Capacity Act 2005 principles.
Educational Walkthrough
Before any questions are asked, every tester is guided through five interactive sections that explain exactly what OpenScouter does with their data. Each section includes audio narration so testers can listen rather than read, which is important for many neurodivergent users.
Interactive consent walkthrough
What we collect
Audio available- •Facial expressions via webcam (emotions, not identity)
- •Voice transcripts (what you say during tests)
- •Browser activity (clicks, scrolls, navigation)
- •Your ND profile (shared anonymously with clients)
How your data is used
Audio available- •To improve digital accessibility for ND users
- •Clients compare which website version works better
- •Never sold to advertisers or third parties
- •Never used to diagnose conditions or assess mental health
Your rights
Audio available- •Revoke consent at any time from profile settings
- •All session data deleted on revocation
- •Stop any test at any time, no penalty
- •Request a copy of your data or deletion
Risks and your wellbeing
Audio available- •Tests involve focus, webcam use, sometimes frustrating interfaces
- •Your wellbeing comes first. If uncomfortable, stop.
- •Never penalised for stopping or for how you react
- •Emotional data handled confidentially
Payment and free choice
Audio available- •Earn per completed test, paid via Stripe
- •No "bad" results. Authentic reactions are valuable.
- •Keep all money from completed tests if you leave
- •No pressure, no deadline, no penalty for declining
Comprehension Quiz
After the walkthrough, testers answer 10 multiple-choice questions. Two questions from each of the five domains, randomised order, with shuffled answer options. Correct answers are never sent to the browser. All scoring happens server-side.
Can OpenScouter use your test data to diagnose a medical condition?
OpenScouter collects interaction and emotional data to improve website accessibility. It is never used for medical diagnosis, mental health assessment, or any clinical purpose. This is stated in the consent disclosure.
Outcomes and Safeguards
The quiz is not designed to catch people out. It is designed to catch gaps in understanding. The system handles every outcome with care.
Pass (8+/10)
Tester proceeds to their dashboard and can begin accepting test sessions immediately. Their consent is recorded with a timestamp.
Fail (7-day cooldown)
Tester receives feedback on which domains they got wrong, reviews the walkthrough again, and can retake after a 7-day cooldown. Up to 3 attempts total.
3 fails (human review)
After 3 failed attempts, the case is escalated to a human capacity review. A team member schedules a call (with a supporter if the tester wishes) to walk through the consent material together.
Anti-Gaming Measures
Server-side validation
Correct answers are never sent to the browser. The quiz API returns questions without marking which option is correct. All scoring happens server-side after submission.
Randomised questions
Questions are drawn from a pool using Fisher-Yates shuffle. Option order is also randomised per attempt, so memorising positions does not help.
Cooldown periods
A 7-day cooldown between attempts prevents rapid retry-until-pass. This ensures testers actually review the material between attempts.
Full audit trail
Every attempt is recorded: questions shown, answers given, score, timestamp. Administrators can review any tester's quiz history.
Human Capacity Review
When a tester is escalated after 3 failed attempts, they are not rejected. They enter a structured review process managed through our admin dashboard.
Review process
Case created automatically
The system logs the reason (3 failed attempts), quiz history, and creates a pending review.
Team member schedules a call
The tester is contacted to arrange a video or phone call. They can bring a supporter, family member, or advocate.
Guided walkthrough together
The team member walks through the consent material with the tester, answering questions and checking understanding.
Outcome recorded
If the reviewer is satisfied the tester understands, they mark the review as "cleared" and the tester is granted quiz pass status. If not, the tester is not onboarded.
Why this matters for clients
Defensible consent
Every tester has demonstrably understood what they consented to. This is not a checkbox. It is a scored, auditable assessment with escalation paths. If a regulator or client asks “how do you know your testers consented?”, we have the evidence.
MCA 2005 aligned
The UK Mental Capacity Act requires that capacity is assessed in relation to specific decisions (not general intelligence). Our quiz tests understanding of specific consent decisions: what data is collected, how it is used, rights, risks, and freedom from coercion.
Dignity preserved
The assessment is designed for neurodivergent users: audio narration, clear language, no time pressure, explanations after each question, and a human fallback. Nobody is excluded without a genuine attempt to support them.
Positive framing
All AI-generated reports frame findings as design problems, not user deficits. “The form creates barriers for users with dyslexia” not “dyslexic users struggled.” This is enforced in every AI system prompt.
Additional protections during sessions
- Right to pause and stop. Testers can take a break at any time. All recording pauses. They can end a session at any time and request deletion of incomplete session data.
- AI never reacts to emotional distress. The live chat AI receives an abstract tone profile (standard, supportive, moderate, restorative), not the tester's raw emotional state.
- Human review before any report. Testers confirm their own session observations before any AI report is generated. No tester is misrepresented in client deliverables.
- ND data never disclosed individually. Client reports show aggregate findings only. No individual tester's ND status, emotional state, or session data is disclosed.
- Fair compensation. Testers are paid per test via Stripe Connect, treated as domain experts providing valuable insights.
For details on data storage and protection, see our Privacy, Data Ethics, and Data Deletion pages.
Questions about participant safeguards?
Our team can walk you through every safeguard in detail and show you the quiz experience first-hand.
Talk to an Expert