Statistics is often seen as the poor relation of mathematics; the subject that people study when they are not interested in or cannot handle "real" maths.
But a new analysis of the performance of students in the highly regarded Pisa (Programme for International Student Assessment) international tests suggests that it is time to turn that assumption on its head: being good at statistics is one of the most accurate predictors of how well students go on to perform in education.
The analysis - undertaken by Maciej Jakubowski of the University of Warsaw in Poland, and published by the Organisation for Economic Cooperation and Development this month - looked at how 4,900 Australian students performed in Pisa in 2003 and whether they went on to obtain university degrees, in any subject, by 2010.
Those who had grasped statistics at age 15 were "significantly" more likely to have higher level qualifications, according to the study. It suggests that the other key thing to master at a young age is how to work with ratios and percentages.
"These items frequently require students to apply common mathematical concepts to solve multi-step, non-routine problems, think flexibly, and understand and interpret information presented in an unfamiliar format or context," according to the analysis.
Dr Jakubowski made his discovery by adjusting the way that the questions in the Pisa tests were classified. He found that being good at certain aspects of mathematics and mathematical thinking was linked with gaining higher qualifications.
The UK Royal Statistical Society welcomed the findings but said it had "major concerns" about the teaching of statistics. Too often, schools focused on instructing students in a set of statistical techniques, rather than problem-solving, a spokesman said.
"Statistics, properly understood, is fundamentally about information in context and applying quantitative skills to real problems that relate to them," he added. "It is not about a collection of techniques in isolation but about creatively applying those techniques in the context of problem-solving."
An example of how students' ability in statistics is tested in Pisa, highlighted by Dr Jakubowski, is taking election polls and asking students to use statistical reasoning to decide which is most likely to be accurate (see panel, right).
The report comes as the national curriculum in England is being redrafted for implementation from September next year.
Roger Porkess, author of a report published last year on the future of statistics in schools, said that it was not a case of dropping algebra or other disciplines but of using the time available to teach statistics more effectively.
Students needed to practise problem-solving by deciding on what data they needed and then working with it to find answers, instead of just doing simple calculations, he explained.
"It's easy to give people questions where they have to work out the mean or median, but it doesn't go anywhere," he said. "The whole point of being able to work out the mean or median is to provide you with information you need to make a decision.
"We've moved into a situation now where every form of employment is data-rich. Those people who have a good understanding of data are at a huge advantage."
The Advisory Committee on Mathematics Education, an independent UK organisation, reported in 2011 that the demands of almost all university courses had increased in terms of students being able to work with statistics. It also said that there had been a similar shift in the workforce towards jobs that required people to handle data.
Keep up to date with the latest developments in mathematics with the monthly TES podcast: www.tesconnect.commathspodcast
An example of how students' ability in statistics is tested in the Programme for International Student Assessment:
Opinion polls were conducted by four newspapers to find out the level of support for the president in a forthcoming election. Which is likely to be the best at predicting the level of support?
Newspaper 1: 36.5 per cent (poll conducted on 6 January, with a sample of 500 randomly selected citizens with voting rights).
Newspaper 2: 41 per cent (poll conducted on 20 January, with a sample of 500 randomly selected citizens with voting rights).
Newspaper 3: 39 per cent (poll conducted on 20 January, with a sample of 1,000 randomly selected citizens with voting rights).
Newspaper 4: 44.5 per cent (poll conducted on 20 January, with 1,000 readers phoning in to vote).
Answer: Newspaper 3's poll because it is most recent, has a larger sample size, is randomly selected and asks voters.