Skip to main content

Assessment and Calibration

Fresh scenario​

A learner rates confidence at 90% after reading a worked example. The next day they can recite its conclusion but cannot solve a changed case. They propose reading the same page five more times and mark the topic complete. Write a better learning check, a possible explanation for the failure and a targeted repair. Then defend your own lab record.

Three required dimensions​

Score each 0 (missing/incorrect), 1 (partly usable or assisted without clear labels), or 2 (complete and justified). All three must reach 2; use a fresh variant after repair.

  • L1 — Evidence design: prompts distinguish recall, explanation and changed-case use, with criteria set before scoring and a preserved independent attempt.
  • L2 — Calibration and repair: prediction and result are compared; a gap hypothesis leads to a specific repair and a check that could challenge it.
  • L3 — Delayed observation: a real later attempt, its interval and assistance status are recorded, with a defensible next action. Pending is not complete. Process quality may pass despite a failed answer; technical mastery is scored in the original module.

Calibration — read after attempting​

Confidence alone is insufficient. Ask for an explanation and a fresh variant without the source, then compare with a predeclared rule. Possible causes include shallow recognition, a missing prerequisite or misunderstanding the new condition. Choose an experiment before assuming which applies. A targeted worked step followed by a new independent attempt can be appropriate; five rereads do not themselves demonstrate transfer. One failure does not diagnose a fixed ability or prove the method ineffective for everyone.

For your record, a repaired answer without the original attempt earns at most 1 on L2. A planned check with no observation cannot pass L3. Reusing the programming retention record is expected; point to the relevant entries.