Scientists often claim discoveries are based on results that are said to be ‘statistically significant’. This is widely taken – even by scientists – to imply that the effect they’ve found is real; for example, that the cut in death rate produced by a new drug isn’t just a fluke. In fact, it merely means that the chance of getting so big an effect by fluke is less than 1 in 20.
Statisticians have repeatedly warned that this is not the same as the chances that the effect really was a fluke. Working that out requires an assessment of the plausibility of the result, which is rarely ever done. As such, despite its widespread use, statistical significance is an unreliable measure of whether an effect is genuine.
Read more:
- What is maths?
- What’s the simplest unsolved maths problem?
- Why can’t we predict prime numbers?
- How do we know that the value of Pi goes on forever?
Authors
Robert is a science writer and visiting professor of science at Aston University.
