Thinking Routines / Predict, Observe, Explain
Predict, Observe, Explain
Commit to a prediction, watch what actually happens, then explain the gap. The predictions lock once the demonstration starts, so nobody can quietly improve them.
Opens on the whiteboard to model with the class · no sign-up · printing is optional
The sheet it prints — optional
What is Predict, Observe, Explain?
Predict, Observe, Explain is a three-step science routine. Children predict what will happen and why, observe what actually happens without interpreting it, then explain the difference. It was described by White and Gunstone in 1992 and is standard practice in primary science.
How do you run Predict, Observe, Explain with a class?
Take predictions in writing before anything is set up, and make each one carry a reason. Run the demonstration and take only observations. Then put the predictions back up beside what happened and ask what the gap tells them.
Why do predictions have to be locked?
Because children who see the answer will revise what they thought they believed, usually without meaning to. A locked prediction preserves the disagreement between expectation and result, and that disagreement is the only thing in the lesson worth explaining.
What makes a good demonstration for Predict, Observe, Explain?
One where the obvious answer is wrong. Two objects of different weight dropped together, ice melting in water against ice melting in oil, a balloon that will not pop over a flame. If the class predicts correctly, nothing needs explaining.
I used to take predictions verbally and it never worked, because the confident child answers first and everyone agrees with them. Written and locked, before any equipment appears, you get a genuine spread — and the moment where the class sees eleven of them predicted the same wrong thing is worth more than any explanation I could give. It also stops the quiet rewriting of history: nobody gets to say they knew all along, because their prediction is still on the board.
Which demonstrations work best?
The ones where the obvious answer turns out to be wrong.
- Forces — two objects of different weight dropped together.
- Materials — ice melting in water against ice melting in oil.
- Heat — a balloon over a flame that will not pop, because of the water in it.
- Light and sound — anything involving a mirror, a lens or a vacuum.
What usually goes wrong with Predict, Observe, Explain?
Taking predictions out loud. The first confident answer becomes everybody's answer, and the spread you needed disappears. Take them in writing, before any equipment is visible, and do not let anyone speak until they are all down.
What do you do with it afterwards?
Keep the predictions. A class that can see eleven people predicted the same wrong thing has something worth explaining, and revisiting that page when the topic ends is the clearest evidence of learning you will get. It also makes the next round of predictions more careful.
How often should you use Predict, Observe, Explain?
Whenever you have a demonstration worth watching, which in primary science is most weeks. It only works when the result can surprise them, so choose the demonstration first and check that the obvious prediction is wrong before planning the lesson around it.
Which year groups is Predict, Observe, Explain for?
Predict, Observe, Explain works from Year 2 upwards, grade 1 in the US. Younger children can predict and observe but find explaining the gap hard, so accept 'I thought it would sink but it floated' as a complete answer. From Year 5 insist the explanation uses the science they have been taught.
| Year group | US grade | What to change | How long |
|---|---|---|---|
| Years 2–3 | Grades 1–2 | One simple demonstration. Predictions as one sentence. | 15 min |
| Years 4–5 | Grades 3–4 | Insist every prediction carries a reason. | 20 min |
| Year 6 and up | Grade 5+ | Ask them to explain the class's wrong predictions, not just their own. | 25–30 min |
Predict, Observe, Explain compared with Think, Puzzle, Explore
Predict, Observe, Explain needs something to happen in the room. Think, Puzzle, Explore does not, so it suits topics you cannot demonstrate. Both surface what a class thinks it knows; only one of them proves it wrong in front of everybody.
Free · no sign-up to run it · printing is optional
This one is ours — Predict, Observe, Explain was described by White and Gunstone in 1992 and has been standard practice in science teaching ever since. It is not a Project Zero routine.