Predict–Observe–Explain with Simulations

Predict observe explain (POE) is one of the simplest ways to make students think before they look. Students commit to a prediction, watch what really happens, and then explain the difference. Interactive simulations suit POE very well, because you can run the "experiment" in seconds, repeat it on demand, and stop students from peeking at the answer before they commit. This guide shows how to run POE with a simulation from start to finish. It includes a fully worked example on free fall with the exact questions and answers, three shorter examples from chemistry, biology and math, a bank of prediction prompts you can copy, and the moves that make the discussion work.

What is Predict–Observe–Explain?

POE is a three-step routine from science education research that surfaces what students really believe.

1. Predict. Students state what they think will happen and why. The "why" matters as much as the prediction. It shows the mental model behind the answer.
2. Observe. Students run or watch the experiment and record what actually happens, without explaining yet.
3. Explain. Students reconcile the prediction with the observation. If they match, they check that their reason holds. If they don't, they work out which part of their thinking was wrong.

The power is in step 1. A student who says "the heavy ball lands first, because heavier things are pulled harder" has put a specific idea on the table. When the balls land together, that idea has to be fixed, not just set aside.

Why POE works: misconceptions don't move on their own

Many science misconceptions are stubborn because they come from everyday experience, and everyday experience mostly supports them. A rock does fall faster than a leaf. Heavier things do feel harder to push. Telling students the correct rule rarely displaces these ideas. They learn the rule for the test and keep the intuition for everything else.

POE works for three reasons:

• Commitment. Once students have written a prediction, they care about the result. A prediction is a small bet.
• Conflict. A clear mismatch between prediction and observation creates the moment of "wait, that's not right". That moment is when students are most willing to change their minds.
• Repair. The Explain step makes students build a better model in their own words, with the evidence in front of them.

Simulations add control: you can remove air resistance, show the graph as it builds and run the same experiment ten times. A clear observation produces a cleaner conflict.

How to set up POE with a simulation

Use this checklist when you plan any POE activity.

1. Pick a discrepant event. Choose a result that a large share of your class will predict wrongly. If everyone predicts correctly, you have a confirmation exercise, not POE.
2. Fix the starting values. On the simulation's Share page, create a link for the class and pin the parameter values for the scenario. Every student then sees the same experiment.
3. Write the predictions as questions. On the Questions tab, build a question set and attach it to the link. Up to three questions can be prediction questions. These are asked before the simulation starts, and the simulation stays locked until the student commits.
4. Ask the predictions again afterward. Prediction questions can be asked again after the experiment, so students can compare their before and after answers.
5. Add explain questions. Use the remaining slots (a set holds up to five questions) for explanation and measurement: a short text answer for "why", a number question checked with a tolerance for a value students read off the simulation.
6. Plan the discussion. Decide which wrong prediction you will put on the board first. The results table shows you which one is most common.

Students only need a nickname to answer. Ask for first names or nicknames, never full names or emails.

Worked example: does the heavier ball land first?

This is a classic POE for ages 11–16. It takes about 25 minutes. It uses the free fall and air resistance simulation, where two spheres of the same material but different mass are dropped together from the same height.

Setup

Create a link named, for example, "Free fall POE – 9B" and pin these starting values:

Parameter	Value
Drop height	45 m
Acceleration due to gravity	9.8 m/s²
Mass of object 1	5 kg
Mass of object 2	0.2 kg
Enable air resistance	off

Instructions for students (paste into the question set): "Answer the predictions first. Then drop the two balls with air resistance off, then turn air resistance on and drop them again. Watch the landing times and the v–t graph."

Prediction questions (locked before play)

P1 (multiple choice). Two balls of the same material are dropped from 45 m with no air resistance. Object 1 has a mass of 5 kg and object 2 has a mass of 0.2 kg. Which lands first?

• 1. The 5 kg ball, clearly first
• 2. The 5 kg ball, just slightly first
• 3. Both at the same time ✔
• 4. The 0.2 kg ball

P2 (multiple choice). Now air resistance is switched on. Which lands first?

• 1. The 5 kg ball ✔
• 2. Both at the same time
• 3. The 0.2 kg ball

P3 (short text). Explain your answer to P1 in one or two sentences.

Set P1 and P2 to be asked again after the experiment.

Observe

Students drop the balls with air resistance off. Both land together, and the info line shows the fall time t = √(2h/g) = 3.03 s for both. On the v–t graph the two lines lie exactly on top of each other: a straight line with slope 9.8 m/s².

Then they tick air resistance and press "Drop again". Now the 5 kg ball lands at about 3.09 s and the 0.2 kg ball at about 3.20 s. The light ball's line curves away from the straight line, and the info line shows the terminal velocities: about 62 m/s for the heavy ball and 36 m/s for the light one.

The gap at 45 m is small, which is useful: ask students to make it bigger. With the drop height at 120 m, the heavy ball lands at about 5.2 s and the light one at about 5.7 s. Lower object 2 to 0.05 kg and it takes about 6.2 s.

Explain questions

Q4 (short text). Why do both balls land at the same time when there is no air? Model answer: the heavier ball is pulled harder, but it also needs more force to accelerate. The two effects cancel, so every object has the same acceleration g.

Q5 (number). With air resistance on and the starting values, what terminal velocity does the simulation show for object 2? Answer: 36.4 m/s, tolerance ± 0.5 m/s, unit m/s.

Add an explanation shown after answering Q5: "At terminal velocity the drag force equals the weight. The lighter ball is smaller and has more surface area for its mass, so drag balances its weight at a lower speed."

What usually happens

In many classes, a large group picks A or B for P1. After the experiment, many of them switch to C, but their written P3 reasoning often still says "heavier is pulled more". That is your discussion target. Students have accepted the observation but not yet the explanation. Spend the Explain step there.

Three more POE examples across subjects

Chemistry: what happens to pH when you dilute an acid?

Use the dilution and pH simulation. Pin the solute to HCl, the initial concentration to 0.1 M and each step to 10-fold.

• Predict: "0.1 M HCl has a pH of 1. What will the pH be after one 10-fold dilution? After three?" Many students say 10, or "it halves".
• Observe: each 10-fold dilution raises the pH by exactly 1 unit: 1, then 2, 3, 4.
• Second predict: "Will 0.1 M CH₃COOH change by the same amount?" It starts at about pH 2.88 and rises only to about 3.38 after one 10-fold dilution, about half a unit.
• Explain: strong acids are fully dissociated. A weak acid ionizes more as it is diluted, which partly makes up for the dilution.
• Bonus prediction: "If you keep diluting HCl, can the pH go above 7?" It approaches 7 but never crosses it, because water itself ionizes.

Biology: does a cooked enzyme recover?

Use the enzyme activity simulation with salivary amylase at 37 °C and pH 7.

• Predict: "We heat the enzyme to 60 °C, then cool it back to 37 °C. What happens to the rate of product formation?" Most students predict that it returns to normal.
• Observe: above 55 °C the enzyme denatures, and the status shows DENATURED. Cooling back to 37 °C does not bring the activity back. Students must press Reset.
• Explain: heat changes the shape of the active site. The protein doesn't refold when cooled, so the substrate no longer fits.
• Extension prediction: "Pepsin works in the stomach. What will happen to its activity at pH 8?" It almost disappears.

Math: which total is most likely with two dice?

Use the two dice simulation.

• Predict: "You roll two dice and add them. Are all totals from 2 to 12 equally likely? If not, which is most likely?" Many students say all totals are equally likely, or that 6 and 7 are "about the same".
• Observe: choose the event "Sum = k" with k = 7. Six of the 36 cells light up, so P = 6/36 ≈ 0.167. With k = 2, only one cell does (≈ 0.028). Then set N to 2000 and press "Roll N times". The experimental probability for 7 settles close to 0.167.
• Explain: (1; 2) and (2; 1) are different outcomes, so sums near the middle can be made in more ways.

Number question for this one: "What is the theoretical probability that the sum is 7? Give a decimal." Answer 0.167, tolerance ± 0.005.

Ready-to-copy prediction prompts

Use these stems with any simulation. Each one pushes for a reason, not just a guess.

• "What will happen to ___ when we increase ___? Explain why."
• "Rank A, B and C from fastest to slowest. What did you base your ranking on?"
• "Sketch the graph of ___ against ___ that you expect to see."
• "Will ___ go up, go down or stay the same? How sure are you (1–5)?"
• "If we double ___, what will happen to ___: double, halve, stay the same, or something else?"
• "Which of these two results do you expect, and what would it mean if you were wrong?"

Subject-specific prompts that reliably produce wrong predictions:

• Physics: "Does a pendulum with a heavier bob swing faster?" (simple pendulum). "Does a block with more mass always sink?" (density – sink or float).
• Chemistry: "Does crushing a solid into powder make the reaction faster, slower or no different?" (reaction rate – collision model).
• Biology: "A potato strip sits in pure water for an hour. Will it gain mass, lose mass or stay the same?" (diffusion and osmosis).
• Math: "If you change only c in y = x² − 4x + 3, does the vertex move left, right, up or down?" (graphing quadratics).

Use multiple choice for predictions where you can, with the common wrong answers as options, so the results table shows which idea students hold. Keep one short-text question for the reasoning.

How to run the Explain discussion

The Explain step is where learning happens, and the one teachers most often rush. Plan at least 10 minutes for it.

1. Show the before and after. Open the results table for the link (or across all classes that use the same question set). Put the prediction counts on the board without names: "14 of you said the heavy ball lands first. After the experiment, 3 still do."
2. Start with the most common wrong answer. Ask: "Why might a sensible person predict that?" It makes being wrong safe and surfaces the intuition to repair.
3. Think, pair, share. Students write one sentence alone ("I now think … because …"), compare with a partner, then share.
4. Go back to the simulation. When an explanation is vague, test it: "If drag causes the gap, what should happen when both balls have the same mass?" Set object 2 to 5 kg and drop them again.
5. Name the model. State the scientific idea in student-friendly words, next to the original wrong prediction.
6. Check transfer. Next day, give a new case: "A bowling ball and a tennis ball are dropped from a window. What happens?"

Sentence starters help students who are new to explaining:

• "I predicted ___ because I thought ___."
• "What actually happened was ___."
• "The difference shows that ___."
• "This would change if ___."

Never grade the prediction. Grade the explanation, if anything. Students punished for wrong predictions learn to predict cautiously, and POE stops working.

Where POE fits in a lesson

POE works as a lesson opener, as the Explore phase of a 5E lesson, or as an end-of-topic check. For a full lesson structure, see the 5E lesson plan with interactive simulations. For using the prediction results to adjust your teaching across a unit, see using simulations for formative assessment. For other ways to bring simulations into lessons, start with how to use interactive simulations in the classroom.

FAQ

How many prediction questions can I add?

Up to three per question set, inside a set of up to five questions. Prediction questions are asked before the simulation starts and can be asked again after the experiment.

Can students see the simulation before they predict?

No. When a question set has prediction questions, the simulation stays locked until the student commits a prediction. This keeps students from testing first and "predicting" afterward.

Does POE work with a projector instead of devices?

Yes. Collect predictions on mini whiteboards or by a show of hands, then run the simulation in present mode. You lose the automatic before-and-after record, but the routine works the same way.

What if most students predict correctly?

Then the event wasn't discrepant enough for that class. Push further with a follow-up prediction, such as "What if we raise the drop height to 120 m?" or "What if the acid is weak?", or move on and keep the activity for a younger group.