A prediction is an expected result worked out from evidence. A claim becomes testable when it names something that can be measured. Small errors in starting measurements limit how far ahead weather can be forecast.
A law or a model states how quantities are related. If you know some of the quantities now, you can calculate others before they are measured. That calculated result is a prediction.
From the laws of motion, you can calculate where a kicked football will land from its launch speed and angle. From a balanced chemical equation, you can calculate how many grams of carbon dioxide 10 g of calcium carbonate gives when heated. In both cases the result is known before the event.
A guess has no stated reason. A prediction comes from a stated law or model and from measured starting values. If someone asks why you expect a result, a prediction has an answer; a guess does not.
A prediction is also specific. It says what will be measured and roughly what value is expected. That is what makes it possible to check.
Varsha says: “It will rain this afternoon because the clouds look dark.” What would you measure to turn this into a claim that can be tested?
Measure quantities that have been recorded before past rain: the relative humidity, the air pressure and whether it is falling, the wind speed and direction, and the temperature trend.
Each one gives a number that two people would read the same way. Each can be compared with the values recorded before earlier rain. “The clouds look dark” is a description, and two people could disagree about it.
From a claim to a test. A useful question names something that can be measured. Two people measuring it should get the same number, and the number can be compared with past records.
If the measured result matches the prediction, the law or model has passed one more test. Confidence in it increases.
If the result does not match, something in the chain is wrong. It could be a measurement, an assumption, or the model itself. Scientists check each of these in turn. A failed prediction shows where to look, which is why it often teaches more than a successful one.
A forecast starts from measurements of temperature, pressure, humidity and wind at many places. A computer model then calculates how the atmosphere changes, step by step.
Every measurement has a small error. In the atmosphere, a small difference in the starting values grows larger at each step of the calculation. A system with this property is called chaotic. After about one to two weeks, the growth makes two forecasts that started almost identically completely different.
The widget is a schematic, not real weather data. It shows the pattern: the gap between the two lines grows faster with each day. This is why a forecast for the next few hours is usually accurate, while a forecast for a particular day next month is not possible. The model is not the problem. The limit comes from the growth of small measurement errors.
A message spreads online: “Food becomes harmful if eaten during an eclipse.” How would you check this claim?
Start with what an eclipse is. During a solar eclipse, the Moon passes between the Sun and the Earth and blocks some sunlight. Nothing else reaches the food.
There is no change in the food that could make it harmful, and no measurement showing one. The method was to ask what mechanism the claim needs, then check whether that mechanism exists.
Ask what the mechanism is. A solar eclipse is the Moon’s shadow falling on part of the Earth. Food in an ordinary shadow does not spoil, so an eclipse gives it no new way to become harmful.
A law or model has been tested only over a certain range of conditions. Outside that range, it may fail. More precise measurements can also reveal small errors that were hidden before.
When that happens, an idea is changed or replaced because of the new evidence, not because of anyone’s opinion. No scientific idea is beyond testing.
1In science, a prediction is…
2Which question is the most useful for testing “it will rain today”?
3A careful measurement does not match a prediction. What is the scientific response?
4Which of these make “it will rain this afternoon” testable?
More than one answer is correct.
5Why is a weather forecast for the next few hours more reliable than one for next month?
6To check the claim that food becomes harmful during an eclipse, the best first question is…
7A well-tested law may still fail when…
8Scientists reject an idea not because of opinion, but because of ____.
Next: how to get a useful answer when exact numbers are not available.
Noescio