Calibration practice

Predict your performance before each retrieval test, then compare prediction to outcome.

Why it works

Learners are systematically overconfident about what they know — the fluency of recognition inflates confidence well beyond actual recall ability. Calibration practice directly confronts this: by predicting performance and immediately checking the prediction against outcome, learners receive concrete evidence of their own misjudgment. Over repetitions, the predictions get closer to outcomes — producing accurate self-assessment, which is the foundation of effective self-directed study.

How to do it

1. Before a quiz or retrieval test, write your predicted score (0–100%).
2. Complete the test, then score it.
3. Record the gap between prediction and outcome.
4. Note specifically which items you predicted you’d get right but didn’t — those are the fluency traps.

Evidence

Overconfidence and poor calibration in self-assessment are well-established findings in educational and cognitive psychology. Calibration training — repeatedly comparing predictions to outcomes — has been shown to improve accuracy of self-assessment. (observational)

Calibration is a relative measure — improving calibration does not automatically improve performance; it improves the learner’s ability to allocate study time to what actually needs work.

Sources

• Kruger & Dunning (1999), unskilled and unaware, Journal of Personality and Social Psychology
• Dunning et al. (2003), "Why people fail to recognize their own incompetence," Current Directions in Psychological Science

Common mistake

Testing knowledge after reviewing material without first making a prediction, which allows the review to contaminate the calibration signal and leaves the learner unable to distinguish what they knew from what they just read.