# Licet learning validation protocol

Status: proposed study, not completed research. The learning hub has not demonstrated measured learning gains. Last updated 11 October 2026.

## Learning objectives

Learners should trace attributed institutional permission, detect exact-action and version mismatch, locate enforceable control seams and bypasses, distinguish permission from execution and assessed effectiveness, handle known revocation and freshness limits, and assign a bounded next decision with evidence and ownership.

## Formative instrument in the site

- Initial diagnostic: three unhinted decisions on permission, execution and change. The first committed response is retained.
- Continuous Meridian case: five chapters, four professional workspaces, visible consequences and revisable decisions. Practice feedback is instructional, not an independent outcome measure.
- Unfamiliar transfer: five structured criteria on a purchasing, infrastructure or data-operations case. No hints before submission; variants rotate across retries. The first variant is assigned locally at case creation. Rationale and confidence are recorded before feedback.
- Rationale review: four criteria scored by the learner or a facilitator, each 0 absent, 1 partial, 2 supported. Reviewer provenance is explicit. The site does not automatically understand or grade prose.
- Delayed recall: available 24 hours after the latest transfer attempt, using a different variant. This product interval is not a validated optimum; a formal study should use prespecified longer follow-up such as 7 days and record elapsed time.

Items are not psychometrically validated or established as difficulty-equivalent. Do not interpret a raw pre/post score difference as a causal learning gain. Completion, coached retries, first-attempt performance, and independent human review must be reported separately.

## Proposed comparative study

1. Conduct initial usability sessions with participants across counsel, federal authorizing officials, cybersecurity and development. Recruit novice and experienced participants; record prior familiarity with Licet and authorization concepts.
2. Freeze the curriculum and item versions. Prespecify primary outcomes, scoring, exclusion/missing-data handling, analysis and an appropriate sample size based on a power calculation. A small convenience pilot is not sufficient for an efficacy claim.
3. Assign participants within role and experience strata to the case-based experience or the same concepts in a static presentation. Keep content coverage and allowed learning time comparable; prevent study-arm crossover before assessment.
4. Counterbalance parallel transfer variants between arms and visits. The site's local rotation is for practice; formal counterbalancing and participant assignment require facilitator control outside the site. Avoid reusing an already exposed variant as an unseen outcome measure.
5. Give baseline and immediate unseen-case assessment without coaching. Capture decision accuracy, unsupported claims, rationale quality, completion time and confidence. Blind rationale reviewers to study arm where practical.
6. Collect delayed unseen-case performance at the prespecified interval. Record attrition, exposure to further practice, and actual elapsed time.
7. Use two reviewers for a defined subset, report agreement and resolve disagreements with a documented process. Report outcomes by role and prior experience, with uncertainty and limits on generalization.

## Rationale rubric

| Criterion | 0: absent | 1: partial | 2: supported |
| --- | --- | --- | --- |
| Artifact-specific rationale | No decisive artifact facts | Mentions a mismatch vaguely | Identifies actual proposal, earlier approval and governing-version mismatch |
| Revocation reasoning | Ignores known revocation | Notices outage/freshness but misses its priority | Explains known revocation removes permission despite cache age |
| Permission versus execution | Equates ALLOW with success or non-execution | Notes uncertainty without specifying missing observation | Clearly limits the claim and requests independently sourced executor outcome |
| Bounded next step and owner | Blanket approval or no next step | Partial containment or owner/evidence request | Stops the changed side effect, identifies review of exact authority, accountable ownership and complete reconciliation |

A structured 5/5 and rationale 8/8 indicate supported performance on these specific teaching criteria, under the stated review conditions. They are not certification, general professional competence, a compliance finding, or agency authorization.

## Facilitator prompts

During usability practice, ask participants to think aloud: Which fact changed your decision? What does this artifact establish? What remains unknown? Who owns the next judgment? Do not ask leading questions during the unhinted outcome assessment.

## Privacy and records

The public site sends no learning answers, notes or analytics to a server. Case state and results remain in browser storage. The learner can download a JSON record containing choices, scores, confidence and rationale. Review for personal or confidential information before sharing. Formal research recruitment, consent, collection, retention and data governance must be established separately; the website does not implement a participant-study backend.