Calibration, Feedback and Repair
What this concept is
Calibration compares how well you expect to perform with how well you actually perform. Confidence is a prediction, not a personality trait or a score to maximize. An error log is useful when it changes the next action, not when it becomes a catalogue of self-criticism.
Why it matters here
Without a clear criterion you may grade effort, neat notes or familiarity instead of the skill. With a criterion, a failed attempt can identify the next small repair.
Concrete example
Before four short prompts, predict that three will be correct. Score each against a rule written beforehand: the answer must state the result and explain the relevant step. Only two pass. Record “predicted 3/4, observed 2/4,” then inspect which two failed. This tiny sample does not establish a stable overconfidence trait.
If you remembered the loop rule but used an exclusive comparison, repair boundary interpretation. If you could not begin because the variable names were unfamiliar, repair reading the data model. Rewatching the whole course addresses neither diagnosis precisely.
Common confusion / misconception
Do not change the scoring rule after seeing the result or treat a copied correction as an independent success. One good day does not prove a study technique caused improvement: topic difficulty, prior knowledge and rest may differ.
How to use it
Use this cycle: specify the outcome; predict; attempt; compare; classify the gap; choose a repair; retry a fresh variant. Gap categories can be prerequisite, concept, procedure, question interpretation or execution slip. They are working hypotheses, not diagnoses. Test one by changing the input or explaining a step.
If two attempts make no progress, reduce the task or seek a targeted hint. The number two is a practical starting rule, not a scientific threshold. Revise it for your needs. Keep independent and assisted results separate.
Check yourself
A hint helps you solve a task. Can you now mark it “independent”? No. Record assisted success and arrange a later attempt on a fresh variant without the hint.
Mini drill or application
Audit one recent exercise: write the pass rule, confidence prediction, actual result, gap hypothesis and a repair that addresses that gap. Explain what observation would make you reject your hypothesis.
Read this only if stuck
Deslauriers and colleagues (2019) found that perceived learning and measured learning diverged in their introductory physics setting. That supports checking performance rather than relying on feeling alone; it does not establish that every difficult activity is effective.