Scientific uncertainty

What it is

Scientific uncertainty is an honest statement of how sure we are about something. Every scientific finding carries some uncertainty. But the amount varies hugely. Some conclusions are close to certain. Others are educated guesses.¹

"Uncertain" does not mean "unknown." And it certainly doesn't mean "anything goes." Saying "the answer is somewhere between 2 and 4" is real knowledge, even though it isn't one exact number.

Why it matters

In 1955, the physicist Richard Feynman gave a talk called "The Value of Science." In it, he described scientific knowledge as a collection of statements that are more or less certain, with none of them completely certain.² That is not a weakness. It is what makes science able to improve.

But uncertainty is easy to misuse. People sometimes say: "Scientists aren't 100% sure, so it might be false." That sounds reasonable, but it is misleading. A finding that is 99% certain is a sound basis for action, even if it isn't 100%. Doubt comes in degrees. A 95% conclusion and a 50% guess are not the same thing.

Many people also worry that admitting uncertainty makes the public trust science less. Researchers tested this in five experiments with nearly 6,000 people, including a trial on the BBC News website. Stating uncertainty, especially as a range of numbers, lowered trust only slightly.³

How scientists put uncertainty into words

Scientists have developed careful ways to say how uncertain they are. A good example comes from the IPCC, the United Nations body that assesses climate science. Its authors use everyday words with agreed, fixed meanings:⁴

They also give a separate confidence level, which describes how strong the evidence is and how much experts agree. When the evidence is overwhelming, the guidance even allows a finding to be stated simply as fact.⁴

The same guidance also warns about the opposite problem. Groups of experts can become overconfident and settle on ranges that are too narrow.⁴ Sounding confident is not the same as being right.

In everyday life

• The weather and your plans. A forecast that says rain is "likely" doesn't promise rain. But it is a good reason to take an umbrella.
• Health news. A news story says a food "may" be linked to a disease. That "may" is doing a lot of work. Ask how sure the researchers really are, and on what evidence.
• "Scientists disagree." On almost any topic, someone disagrees. The useful question is how much evidence sits on each side, not whether any doubt exists.
• Your own beliefs. You may be "sure" your train leaves at 8:05. How sure, exactly? Would you bet on it?

The common trap

The first trap is treating any uncertainty as if it means we know nothing. A range is knowledge. A finding that is very likely is still very likely, even if it isn't certain.

The second trap is treating any doubt as if both sides were equal. A 95% chance and a 50% chance are very different.

The third trap is trusting whoever sounds most certain. Overconfidence is a well-known failure, even among experts.⁴

Some ways out:

• Ask "how sure?" rather than "sure or not?"
• Look for ranges and likelihood words, and check what they mean.
• Ask what evidence would make the conclusion more or less certain.

Try it

Write down five things you believe. Mix big ones and small ones: "The Earth goes round the Sun," "Coffee is bad for you," "My train leaves at 8:05." Next to each, write how sure you are, from 50% to 99%.

For each one, write down what evidence would make you more sure, and what would make you less sure. Which ones did you feel certain about, but couldn't really back up?

Related concepts

Probabilistic reasoning · Error bars and confidence intervals · Iteration · Scientific models · Credible sources · Scenario planning · Facts vs values · Falsifiability

Sources

1. van der Bles, A. M., van der Linden, S., Freeman, A. L. J., Mitchell, J., Galvao, A. B., Zaval, L., & Spiegelhalter, D. J. (2019). Communicating uncertainty about facts, numbers and science. Royal Society Open Science, 6(5), 181870. https://doi.org/10.1098/rsos.181870

2. Feynman, R. P. (1955). The value of science. Engineering and Science, 19(3), 13–15. https://calteches.library.caltech.edu/1575/1/Science.pdf

3. van der Bles, A. M., van der Linden, S., Freeman, A. L. J., & Spiegelhalter, D. J. (2020). The effects of communicating uncertainty on public trust in facts and numbers. PNAS, 117(14), 7672–7683. https://doi.org/10.1073/pnas.1913678117

4. Mastrandrea, M. D., Field, C. B., Stocker, T. F., Edenhofer, O., Ebi, K. L., Frame, D. J., Held, H., Kriegler, E., Mach, K. J., Matschoss, P. R., Plattner, G.-K., Yohe, G. W., & Zwiers, F. W. (2010). Guidance note for lead authors of the IPCC Fifth Assessment Report on consistent treatment of uncertainties. IPCC. https://www.ipcc.ch/site/assets/uploads/2017/08/AR5_Uncertainty_Guidance_Note.pdf