Predict Observe Explain¶

Strategy · All strategies
Evidence · 5 claims (5 for) · 17 studies (9 quant-synthesis, 5 causal, 2 review, 1 associational), q2–q4 · 9 of 17 report an effect size

Description¶

Predict Observe Explain (POE) is a three-phase instructional strategy, originally developed by White and Gunstone (1992), in which learners first commit to a written prediction about the outcome of a demonstration or event, then observe the actual outcome, and finally explain what happened — especially any mismatch between their prediction and the observation. The commitment to a prediction before observation is what distinguishes POE from ordinary demonstration: it surfaces prior conceptions and creates a reason to resolve discrepancies.

Design Implications¶

POE works because committing to a prediction activates prior knowledge and exposes it to testing, and because observed discrepancies between prediction and outcome create the cognitive conflict that motivates conceptual change Cognitive disequilibrium motivates conceptual change. [+M]. The strategy also leverages the testing effect: making a prediction is a form of retrieval and generation that strengthens subsequent learning even when the prediction is wrong Retrieval practice improves retention more than restudying. [+S]. The explanation phase is where most learning accrues; without it, learners may dismiss or rationalize discrepant observations Self-explanation prompts improve learning. [+S].

Context¶

Requirements¶

• A demonstration, simulation, or event with a genuinely uncertain or counterintuitive outcome — if the outcome is obvious, the prediction phase is busywork
• A mechanism for every learner to commit privately to a prediction (written, clicker, or digital poll) before observation, so public consensus does not suppress misconceptions
• Time and structure for the explanation phase, ideally with peer discussion before instructor resolution (Class Discussion)
• Instructor knowledge of common misconceptions in the topic, so the chosen event targets them
• An event or demonstration with a genuinely uncertain or counterintuitive outcome — if the outcome is obvious, prediction adds nothing
• A mechanism for every learner to commit to a prediction (written, clicker, or app-based) before observation, so commitment is real rather than performative
• Time and structure for the explanation phase, ideally with peer discussion before instructor consolidation
• Instructor knowledge of common misconceptions in the topic, to design events that target them

Constraints¶

• Ineffective when learners lack the prior knowledge to generate a meaningful prediction — they guess randomly and learn little from the discrepancy [~M]
• Discrepant observations alone do not change conceptions; learners frequently reinterpret the observation to fit their existing belief unless the explanation phase is carefully facilitated [-M]
• If the instructor reveals the answer immediately after observation, the explain phase collapses into confirmation and the conceptual-change benefit is lost [-M]
• Poorly chosen events (ambiguous outcomes, noisy data) generate confusion rather than productive disequilibrium [-W]
• Ineffective when learners can predict correctly from surface cues without engaging the underlying concept — the event must discriminate between conceptions
• Learners with fragile prior knowledge may guess randomly, producing no productive conflict to resolve [~M]
• If the observation is ambiguous or the explanation phase is skipped, learners may rationalize the discrepancy away or leave more confused than before [-M]
• Poorly designed events can entrench misconceptions if the "surprise" is attributed to experimental error rather than conceptual error [~M]

Implementation Variability¶

• POE with peer discussion: after private prediction, learners discuss predictions in pairs before observation (a Peer Instruction-style variant)
• Simulation-based POE: virtual labs and simulations (e.g., PhET) allow repeated observation and manipulation after the initial prediction
• POE as assessment: White and Gunstone designed POE partly as a probing tool — teachers use written predictions and explanations as formative evidence of students' conceptions (Assessment)
• Delayed observation: in lecture settings, the observation can be deferred to a later session to add a spacing benefit
• POE with simulation: virtual labs such as PhET allow repeated observation and manipulation after the initial prediction
• FADE variant: prediction only, with explanation folded into follow-up problem solving, for time-constrained settings
• Written POE: full individual written sequence used as a formative assessment artifact

Target Learners¶

• Learners who hold intuitive but incorrect conceptions of a phenomenon — the strategy is explicitly designed to surface and confront these [~M]
• Intermediate learners with enough background to make a reasoned (not random) prediction; complete novices benefit less [~M]
• Less effective for learners with strong correct prior knowledge, for whom prediction adds little beyond routine application
• Learners holding intuitive but incorrect conceptions (misconceptions) that conflict with the scientific account [+M]
• Intermediate learners with enough prior knowledge to generate a meaningful prediction; complete novices benefit less because they cannot commit to a reasoned guess [~M]
• Less effective for advanced learners whose predictions are already accurate — the conflict mechanism has nothing to act on

Target Learning Goals¶

• Conceptual change: replacing intuitive misconceptions with scientific conceptions
• Causal reasoning: linking observations to underlying mechanisms
• Metacognition: monitoring one's own understanding by comparing expectation against evidence
• Scientific epistemic practice: experiencing hypothesis-testing as a way of knowing
• Conceptual change: replacing intuitive models with scientific ones
• Epistemic practice: treating evidence as arbiter between competing claims
• Formative assessment: the prediction and explanation phases surface thinking the instructor can respond to Assessment for learning improves achievement by surfacing and responding to learner thinking. [+S]

Instructions¶

1. Predict: Present the setup of a demonstration or event (without the outcome). Each learner privately commits to a prediction and a reason, in writing or via poll. Do not collect correctness at this stage.
2. Observe: Run the demonstration, simulation, or data reveal. Keep it short and unambiguous.
3. Explain: Ask learners to write whether their prediction matched and why. Use Peer Instruction-style pair discussion, then whole-class Class Discussion, before the instructor resolves the science.
4. Consolidate: Name the target conception explicitly, connect it to the discrepant event, and follow with application (Application of Knowledge) or a second POE on a related phenomenon.

Related Strategies¶

• Peer Instruction — shares the commit-then-discuss-then-resolve structure, applied to conceptual questions rather than physical events
• Case-Based Learning — similarly anchors learning in a concrete, discrepant scenario
• Think-Pair-Share — the discussion structure most often layered onto the predict and explain phases
• Comparing Cases — alternative route to discriminating conceptions through structured contrast Comparing and contrasting cases improves learning. [+S]
• Concept Probing — diagnostic questioning without the observation component
• Use PDEODE tasks (Predict-Discuss-Explain-Observe-Discuss-Explain) to facilitate conceptual change in chemistry topics

Examples¶

• Physics lecture (Mazur's Peer Instruction lineage): Students predict whether a heavy and a light ball dropped together land simultaneously, vote with clickers, discuss, then observe the drop — the same predict-commit-discuss cycle Crouch and Mazur documented at Harvard.
• PhET Interactive Simulations (University of Colorado Boulder) — teachers run POE cycles around simulations such as circuit construction: predict bulb brightness, then test in the sim.
• Chemistry misconceptions: The classic "mass of dissolved salt" POE — students predict whether dissolving salt increases the mass of water, observe a balance reading, and confront the conservation-of-mass misconception.
• Physics: forces on a coin on a rotating turntable — learners predict the coin's path when released, observe it, and confront the centrifugal-force misconception.
• Chemistry: mass change in a closed vs. open system during a reaction — predictions typically split; the observation forces engagement with conservation of mass.
• PhET Interactive Simulations — simulations such as Circuit Construction Kit support POE sequences: predict bulb brightness, run the circuit, explain the result.
• CLUE (Chemistry, Life, the Universe and Everything) — a redesigned general chemistry curriculum built around predict-observe-explain cycles.

Key Sources¶

• White, R., & Gunstone, R. (1992). Probing understanding. Falmer Press.
• Chi, M. T. H., Bassok, M., Lewis, M. W., Reimann, P., & Glaser, R. (1989). Self-explanations: How students study and use examples in learning to solve problems. Cognitive Science, 13(2), 145–182. doi:10.1207/s15516709cog1302_1
• Crouch, C. H., & Mazur, E. (2001). Peer instruction: Ten years of experience and results. American Journal of Physics, 69(9), 970–977. doi:10.1119/1.1374249
• Gunstone, R. F. (1994). The importance of specific science content in the enhancement of metacognition. In P. J. Fensham, R. F. Gunstone, & R. T. White (Eds.), The content of science: A constructivist approach to its teaching and learning (pp. 131–146). Falmer Press.
• Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. doi:10.1111/j.1467-9280.2006.01693.x
• White, R., & Gunstone, R. (1992). Probing understanding. London: Falmer Press.
• Gunstone, R. F., & White, R. T. (1981). Understanding of gravity. Science Education, 65(3), 291–299. doi:10.1002/sce.3730650308
• Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–112. doi:10.3102/003465430298487
• Freeman, S., et al. (2014). Active learning increases student performance in science, engineering, and mathematics. PNAS, 111(23), 8410–8415. doi:10.1073/pnas.1319030111
• Vosniadou, S. (2013). Conceptual change in learning and instruction. In International Handbook of Research on Conceptual Change. Routledge. doi:10.4324/9780203154472.ch1