Assessment Policy

Every attempt in Competency Labs belongs to exactly one of two tracks. The track is stored on the attempt record and cannot be changed after submission.

• AI Practice — self-service simulations inside your Development Profile. Graded by an AI rubric grader. Produces a Practice Performance Score and formative feedback. Never eligible for a Competency Passport. Attempts are enforced practice-only at the database level, and retakes always create a new attempt without overwriting prior feedback.
• Verified assessment — invited attempts administered under identity-verified conditions. AI produces a first-pass analysis; a human reviewer makes the final scoring decision. A passing verified attempt with an approved integrity disposition is eligible for a Competency Passport.

Both tracks use the same two-layer rubric so that practice feedback and verified assessment are directly comparable:

1. Competency layer — an ordered scale for each competency in the framework being assessed (e.g. Source Documentation, ICH-GCP fluency, Monitoring Visits).
2. Entrustability layer — a decision about the level of supervision a candidate would require to perform the competency in a live setting, following the entrustable-professional-activity model used in clinical education.

For verified attempts, the reviewer records both an initial machine score and a reviewer-adjusted score, so candidates and employers can see where the human changed the AI's read.

Verified attempts are administered under the following conditions:

• The candidate must complete identity verification before the attempt starts.
• The attempt is timed and time-spent is persisted with the record.
• The AI grader emits integrity signals (unusually generic phrasing, apparent LLM-authored answers, paste-pattern anomalies); a human reviewer adjudicates each signal.
• Reviewers record an integrity disposition (clear / concerns / disqualified) alongside the score.
• Multiple simultaneous attempts by the same candidate on the same casebook are not permitted.

Practice attempts are not held to these conditions and cannot be converted into verified evidence, no matter the score achieved.

Practice retakes are configurable per casebook (typical limit: 3 attempts). Every retake creates a new attempt record with an incremented attempt_number; prior feedback is preserved and visible in the attempt history. Retakes never overwrite earlier results.

Verified retakes are handled differently. A candidate may re-take a verified assessment only after an appeal has been considered or after a defined cool-down period, and only when framework or evidence gaps warrant it.

If you believe a verified score, integrity disposition, or credential decision is materially wrong, email appeals@competencylabs.com within 30 days of the decision. A reviewer who was not involved in the original decision will review the attempt and respond within 15 working days. Appeal outcomes are recorded on the credential audit trail.

If you have a disability or health condition that affects your ability to complete a timed verified assessment, email support@competencylabs.com before the attempt starts. We will discuss adjustments that do not compromise the integrity of the assessment, such as extra time, screen-reader-friendly formats, or rest breaks.

We will update this policy when we change rubric scales, integrity rules, or retake limits. Existing credentials are always evaluated against the policy in force at the time the attempt was scored — see the Credential Verification Policy for versioning details.