Estimate and Predict¶

Strategy · All strategies
Evidence · 1 claim (1 for) · 3 studies (2 causal, 1 quant-synthesis), q3 · 1 of 3 report an effect size

Description¶

Learners make explicit, recorded predictions before a task — how well they will perform on an upcoming project, how long an assignment will take, or what score they expect — and then compare those predictions against actual outcomes. The comparison cycle, repeated across tasks, trains the accuracy of learners' judgments of their own knowledge and performance.

Design Implications¶

Prediction before learning is itself a learning event: attempting an answer or forecast, even unsuccessfully, prepares learners to encode the corrective information that follows Pretesting and unsuccessful retrieval attempts enhance subsequent learning. [+M]. The larger payoff, however, is metacognitive calibration — learners systematically underestimate task duration (the planning fallacy) and overestimate their mastery, and only feedback on prediction accuracy closes that gap. Estimates must therefore be recorded in advance and revisited with outcomes; prediction without a structured comparison produces no calibration benefit.

Context¶

Requirements¶

• Tasks with outcomes concrete enough to score against a prediction (test scores, rubric ratings, actual time-on-task)
• A lightweight recording mechanism (estimate column, prediction log, LMS quiz field) captured before the outcome is known
• A structured reflection step comparing estimate to outcome and prompting a causal explanation for the discrepancy
• Repeated cycles across multiple tasks — calibration improves with practice and feedback, not a single instance

Constraints¶

• Learners who are not motivated to reflect honestly (or who treat estimates as a formality) gain little; anonymous or low-stakes recording can reduce self-presentational bias
• Feedback on actual outcomes must be timely — delayed or absent feedback breaks the calibration loop
• Poorly defined tasks (no measurable outcome) make accuracy comparisons meaningless
• Repeated failure feedback without guidance can depress self-efficacy rather than improve calibration; pair discrepancy discussion with concrete adjustment strategies

Implementation Variability¶

• Confidence ratings: learners predict a score and rate confidence, distinguishing calibration from mere optimism
• Time estimation: predict duration for a task, log actual time; especially useful for project planning and exam pacing
• Pretest predictions: predict answers on a prequiz before instruction, leveraging the pretesting effect for content learning as well as calibration
• Peer prediction: learners predict a peer's performance, which can be more accurate than self-prediction and surfaces criteria for judging quality

Target Learners¶

• Novices, who typically show the largest miscalibration (overconfidence with low knowledge) and benefit most from explicit comparison cycles
• Students developing time-management and project-planning skills, where the planning fallacy is strongest
• Less useful for highly experienced learners who already calibrate well in the domain [~M]

Target Learning Goals¶

• Metacognitive accuracy: improving judgments of learning and self-assessment
• Self-regulated learning: planning, monitoring, and adjusting effort allocation
• Study strategy selection: learners with accurate calibration allocate study time where it is actually needed

Instructions¶

1. Before the task, have learners record a specific, quantified estimate (expected score, expected duration, expected difficulty ranking) — a brief Check-In or prediction field works well.
2. Optionally begin with a low-stakes pretest or attempt before instruction to gain the encoding benefit of prediction Pretesting and unsuccessful retrieval attempts enhance subsequent learning. [+M].
3. During the task, have learners monitor against their estimate (Assess Performance).
4. After outcomes are known, run a structured Individual Reflection: Was the estimate accurate? What caused the gap? What will change next time?
5. Repeat across tasks and show learners their own calibration trend over time so improvement is visible.

Related Strategies¶

• Activating Prior Knowledge — prediction forces retrieval attempts that surface what learners actually know before instruction
• Self-Assessment — estimation is self-assessment made falsifiable: a specific prediction that outcomes can confirm or disconfirm

Related Elements¶

• Assess Performance — supplies the outcome data that makes estimates comparable
• Check-In — a lightweight moment for recording predictions before work begins
• Articulation — learners explain the reasoning behind their estimates, exposing faulty self-models

Examples¶

• Exam prediction logs: students predict their score on each unit test and graph predicted vs. actual across the term; most see their error shrink by mid-semester.
• Time estimation in project courses: before each build sprint, teams estimate hours per task and log actuals in a tracker (e.g., a shared Trello board with estimate/actual columns), then run a retrospective on estimation error.
• Prequiz predictions: before a statistics lesson, students answer and confidence-rate items they have not yet been taught; the instructor uses the predictions to launch discussion and students use the reveal to correct misconceptions.

Key Sources¶

• Buehler, R., Griffin, D., & Ross, M. (1994). Exploring the "planning fallacy": Why people underestimate their task completion times. Journal of Personality and Social Psychology, 67(3), 366–381. doi:10.1037/0022-3514.67.3.366
• Kornell, N., Hays, M. J., & Bjork, R. A. (2009). Unsuccessful retrieval attempts enhance subsequent learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35(4), 989–998. doi:10.1037/a0015729
• Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students' learning with effective learning techniques. Psychological Science in the Public Interest, 14(1), 4–58. doi:10.1177/1529100612453266
• Nelson, T. O., & Dunlosky, J. (1991). When people's judgments of learning (JOLs) are extremely accurate at predicting subsequent recall. Psychological Science, 2(4), 267–270. doi:10.1111/j.1467-9280.1991.tb00147.x