Self-Paced Retention Study Protocol

Status: draft for human review (P5-01, #530). Supersedes the facilitated runbook in pilot-runbook-20-40.md (#20). Recruiting, consent, item validation and interpretation are [HUMAN-REQUIRED].

1. Question

After completing one QuickStart on their own, how much do learners still remember at least 7 days after their last module? This is the headline outcome of FR-19. It is a delayed test, not an in-session score, because performance during study is a poor guide to lasting learning (Soderstrom & Bjork, 2015).

Secondary questions, reported separately:

Did learners choose the right starting QuickStart (absorbs prod-04)?
How useful and comfortable did learners find the material? (Self-report; never the headline, since it tracks satisfaction more than learning: Sitzmann et al., 2010.)

2. Design

Self-paced. No facilitator, no live sessions. Learners work in their own browser on their own schedule.
One QuickStart per learner. Which one is the learner's own choice from the hub (FR-03 stays descriptive, never placement). Report results per QuickStart and per stage site.
Comparison. Phase 1 is single-arm: scores are described with an effect size and interval against the pre-test of the same learner. A between-site comparison (P9-06) is a later, separate analysis.
Sample. Target 20–40 adults. Too small for efficacy claims, so results are labelled preliminary (see §8).

3. Timeline

When	Step
Day 0	Consent, short baseline quiz (pre-test) on the chosen QuickStart's content, QuickStart choice recorded
Days 0–N	Learner completes the QuickStart at their own pace; last-module date is recorded by the learner
Last module + ≥7 days	Delayed retention test, same construct, different items (see §5)
Same day as delayed test	Self-report survey (separate form) and the "right QuickStart?" question

A reminder is sent by the researcher on day 7 after the learner reports finishing. Learners who test earlier than 7 days are excluded from the headline analysis and kept in a sensitivity analysis.

4. Participants and Safety

Adults 18+, recruited by a human. No clinical claims; this is not therapy.
Exclude the shadow-work QuickStart from this study (self-paced shadow work needs the stop rules and screening in the safety skill; run it only after safety-review signs off).
Participation is voluntary and withdrawable at any time, with data deleted on request.

5. Measures

Headline: delayed retention score. Percentage correct on the delayed test. Items are retrieval questions (free recall and short application), drafted with AI and validated by a human (P5-02, [HUMAN-REQUIRED]). Two parallel item sets (A and B) are used so the pre-test and delayed test differ; set assignment is randomised per learner.

Process (from the learner's browser, shared only by export, P10-03): reviews completed within 2 days of due, calibration gap, pathway completion, return within 7 and 30 days.

Self-report (separate table, never headline): perceived usefulness, comfort with the stage language, and "Did you pick the right starting QuickStart?" (yes / no / unsure, plus a free-text reason).

6. Data Handling

No identifiers in the data file; learners get a random participant code.
Process metrics leave the browser only through the learner-controlled export, after they see its contents (NFR-05).
No reflections or card answers are collected. Test responses are entered into the test form only.
Store data outside the repository. Publish only aggregates.

7. Analysis

Primary: mean delayed score with a 95% confidence interval; standardized effect size (Cohen's d, paired, with interval) of delayed vs. pre-test.
Report per QuickStart and per stage site, with n shown. Suppress cells with n < 5.
Self-report reported in its own table with a clear label.
Mis-placement: share answering "no" to the right-QuickStart question.
Dropout reported at every step; analyse completers and intention-to-treat.

8. Reporting Rules

Label the report preliminary. One study of this size supports no efficacy claim; state limits (self-selection, no control, small n).
Never present self-report as evidence of learning.
Record deviations from this protocol in the report (P5-03, absorbs #21).

9. Review Checklist

Protocol reviewed by a human [HUMAN-REQUIRED]
Consent wording approved
Sample size and exclusion rules accepted
Hand off to P5-02 (item drafting and validation)

References

Soderstrom, N. C., & Bjork, R. A. (2015). Learning versus performance: An integrative review. Perspectives on Psychological Science, 10(2), 176–199.
Sitzmann, T., Ely, K., Brown, K. G., & Bauer, K. N. (2010). Self-assessment of knowledge: A cognitive learning or affective measure? Academy of Management Learning & Education, 9(2), 169–191.