US statistician Bradley Efron at Stanford University in California has won the 2018 International Prize in Statistics for pioneering the ‘bootstrap’ method for measuring the reliability of small data samples.

His work, which dates back to 1977, has given rise to techniques now commonly used across many scientific disciplines.

The American Statistical Association (ASA) — which administers the prize together with four other scientific societies — announced the winner on 12 November. The US$80,000 prize was first awarded in 2016 and is given out every two years; British statistician David Cox was its first winner .

Efron, who is 80, says that he was “thrilled” to receive the prize. Scientists often have to wait many years to get their “round of applause”, he says. “It turns out that’s okay — it feels great!”

Sally Morton, a statistician at Virginia Tech in Blacksburg, says that Efron is “a statistical rock star”. “He has inspired generations of statisticians and scientists,” she says.

In many branches of science, researchers often have to draw conclusions from limited data. Assessing the level of confidence that comes with such conclusions is crucial, and often difficult. In the 1970s, Efron and others realized that the increased availability of computers would make new, computationally intensive tests feasible.

The bootstrap was the first such method. It slices up a data sample in random ways and calculates whether a conclusion — such as the fact that two variables are strongly correlated — is solid. “It allows data analysts to use complicated, sometimes very complicated, methods and still be able to assess their accuracy,” Efron says.

Bootstrapping is now used in a variety of applications, such as machine-learning algorithms, says Peter Bickel, a statistician at the University of California at Berkeley. “It's also used in situations where p -values and other methods of evidence are difficult to compute,” he says, and in particular in medicine.