‘Facts are stubborn things, but statistics are pliable’ (Mark Twain) by Janette Rutterford

How important is it for those in government to be numerically and mathematically literate?  The answer is obvious. Unfortunately, the British educational system does not prioritise the acquisition of these skills. Fifteen year olds in the UK are ranked 17^th in a table of 78 countries with respect to their mathematical understanding[1]. Worryingly, less than half of a sample of adults in the UK could correctly answer four simple questions to do with adding the costs of a few purchases; calculating a cost per litre; estimating a percentage discount; and interpreting a graph[2].  This may reflect the fact that only a tiny minority of British students study maths after the age of 16.

The Prime Minister, Boris Johnson, and Michael Gove, Minister for the Cabinet Office in charge of the practicalities of Brexit, were the two key figures in the pro Brexit campaign.  Johnson studied Classics and Gove studied English, both at Oxford University. They both became journalists. This career trajectory may not have helped their ability to critique models put forward by scientists, to understand how algorithms work, and to foresee issues likely to arise when dealing with large amounts of data. Instead, the government has shown contempt for numbers and their importance.

Johnson and Gove spearheaded the Brexit campaign and endorsed the claim painted on the campaign bus that £350m a day could be used to fund the NHS after Brexit instead of being paid to the EU, a figure based on the UK’s gross contribution instead of the much smaller net contribution which was the relevant figure.

With the pandemic, this cavalier attitude to numbers has only increased.  Numbers are being used as propaganda rather than as facts, echoing the use of statistics by Germany during WW1 to encourage take up of their national War Loan to fund the war.  The Germans believed that facts comparing Germany with its enemies would engender confidence – and hence increase subscriptions for War Loan. A 1915 poster entitled ‘We Barbarians’ depicts a variety of statistics on education and welfare in Germany, France and England. It shows Germany with the smallest proportion of illiterates, the biggest expenditure on education and pensions and the highest number of Nobel Prize winners.[3]

The British preferred  a more emotive approach to persuade investors to buy British War Loan, with an emphasis on helping those on the battle front. But they understood the power of simple, large, eye-catching numbers. A propaganda pamphlet entitled, modestly, The World’s Largest Loan, claimed that there had been 5 million subscribers and £1,000 million new money raised for the third British War Loan issue.  The pamphlet failed to mention that over two million of those subscribers had bought War Savings Certificates, not War Loan, and that only £850m was new money.  This ‘economy with the truth’ was even more stark for the first British War Loan in 1914. The then Prime Minister, Lloyd George, claimed the loan was fully subscribed when, in reality, it failed to reach its £350 million target by a whopping £113 million[4].

As with War Loan, over-optimistic forecasts have been central to the management of the pandemic. Targets for the number of tests per day have gone from 100,000 at the end of April, to 200,000 by end May, and the current target is 500,000 by end October.  Government claims that the end-April target was met, with 122,347 tests on 31^st March, were shot down in flames.  Of that total, nearly 30,000 were in fact uncompleted test kits sent to people’s homes and a further 12872 were tests simply being moved within the laboratory network.

Indeed, the government was publicly reprimanded by the UK Statistics Authority, for not defining clearly enough in advance what a ‘test’ having been done meant in practice.  It asked: did the target number reflect capacity to test, actual tests administered , test results received or the number of people tested?  This same independent Authority had already written to Johnson to criticise the £350 million Brexit bus claim – clearly to no effect.

The two most recent fiascos involve algorithms and Excel spreadsheets.  As the government cancelled end of school exams (‘A Levels’) in June 2020, it replaced them with an algorithm which used the teacher’s grade estimate and the past performance of the student, of cohorts of students and of the school.  Government insisted it did not want grade inflation from the previous year, so teachers’ usually optimistic grade forecasts were downgraded in 40% of cases, hitting underprivileged students the most.  After a public outcry,  teachers’ estimates were used in lieu of the algorithm.

The Excel problem arose from the fact that-  as any researcher knows – Excel spreadsheets are limited in size, with a maximum number of rows and columns. As the number of test results input into an Excel spreadsheet to be used in the Track and Trace system increased, they reached the limit, with the excess results dropping out of the system. Thousands of test results were not entered each day for several days. In total, 16,000 who had tested positive were not told, nor their contacts traced.

Understanding maths and statistics is vital for any government in order to hold modellers, developers, and specialist advisers to account.  Numbers and models are opaque and need challenging. We need a government that aims to give the facts without twisting the numbers and one that is prepared to challenge the numbers generated by others.

[1] https://data.oecd.org/pisa/mathematics-performance-pisa.htm

[2] ‘The financial skills of adults across the world.  New estimates from PIAAC’, A. Bhutoria , J. Jerrim, and A. Vignoles, March 2018. PIAAC Working paper.

[3] http://www.iwm.org.uk/collections/item/object/20677.  Poster by Louis Oppenheimer.

[4] R. Roberts, Saving the City, Oxford University Press, 2013, pp. 188-9.