Journalists don't understand statistics

It has scarcely been possible to open a newspaper in recent weeks without seeing a test set for 11-year-olds in 1898. It contained such fiendishly difficult questions as "multiply 642,035 by 24,506". This was adduced as evidence of "the decline in educational standards".

No journalist had given the test (which first appeared in a weekly magazine) to any of today's 11-year-olds before concluding this was "proof" of dumbing down. Nor did anybody report what proportion of Victorian children took or passed the test. I once dug out an even older test, given to all pupils in the days of "payment by results", and tried it out in a small sample of schools. I compared the average marks achieved in 1979 (as it then was) with those achieved in Victorian times. The former were better than you might have expected.

But it will not surprise TES readers that Victorian children did better still, because the test was designed for the 19th century, not the 20th.

You can give old tests to today's pupils but you cannot give modern tests to their forebears. The 1898 test included a whole section on Latin, invited children to "name the British possessions in America" and asked them to calculate the prices of horses and sheep. Its long-multiplication sums would now be performed in a few seconds by calculator, as would the task of finding the square root of 5,185,440,100. Would Victorian children have done better if they had been given calculators? We cannot know.

Nor can we know how they would have performed if they had been presented with statistics rather than arithmetic. What we do know is that most modern journalists have no grasp of statistics whatever.

Being of a kindly disposition, I will not name the upmarket newspaper (or its distinguished political commentator) which recently reported that 77 per cent of Bush voters in America thought abortion should always be illegal while 37 per cent thought it should be legal. These weren't misprints; they were the result of failures to spot tables that should be read horizontally rather than vertically. The howler must have passed through at least six people before it got into print.

Even more common in newspapers is the failure to grasp the concept of risk.

Headlines state that the chances of getting a particular disease are increased by 20 per cent if you eat, say, a pork chop. This is a meaningless statement unless you know what proportion of the population suffers the disease in the first place. If few ever get the disease, another 20 per cent will also be few. The risk, therefore, is not worth bothering about if you really enjoy a pork chop. Millions change their diets on the basis of such spurious warnings.

Millions more take out insurance without understanding that it simply involves paying somebody else to take on a risk that you cannot afford. If your Pounds 300,000 house burnt down, you would not have the money to re-build it. If your pound;600 washing machine were to go wrong, you could afford to repair it - by taking out an overdraft even if you did not have the cash. In other words, insure the house, but not the washing machine. If you are a single person, your insurance premium subsidises people who have five children (and therefore use their washing machines more frequently), as well as giving the insurance firms money for a service you do not need.

This is elementary statistics. But millions do not understand it. If they did, most insurance companies would go bust. On that score, schools are certainly failing. I would have children chant the basic principles of insurance every morning, just as they once chanted multiplication tables.

That is what newspapers should campaign for, not the restoration of long-multiplication sums from 1898.

Peter Wilby is editor of the New Statesman

Log in or register for FREE to continue reading.

It only takes a moment and you'll get access to more news, plus courses, jobs and teaching resources tailored to you