Where the truth really lies with statistics
As any magician can tell you, the real trick often takes place offstage. The deck has already been stacked; the card placed into the shoe. No matter how closely you look, you’ll never spot the sleight of hand.
So it is with statistics. We often pay attention to the wrong thing, scrutinising the numbers with a forensic eye without asking about what those numbers really describe. Sometimes there is no intent to deceive; there doesn’t need to be. We deceive ourselves.
A simple example: the recent official figures showing the number of homicides recorded by police in England and Wales jumped by more than 100 deaths, or 21 per cent. What could explain this? The Labour party called it “worrying”, which it is, and promised to hire more police officers.
Crime data cognoscenti know that the police database has long been regarded as a poor measure of anything very much, because the policy on recording crimes has changed over the years.
But the true reason not to be alarmed by the rise in homicides in 2016 is that the deaths did not take place in 2016. Ninety-six people died as a result of the disaster at the Hillsborough football stadium in April 1989. After years of campaigning these deaths were recorded as “unlawful killing” 27 years later. Deaths in 1989 became homicides in 2016. Sometimes the statistics are not counting what we assume they are counting.
Unexpected definitions can affect targets as well as trends. In the UK, the most notorious target is the one that keeps being missed: a promise to keep net migration under 100,000. In 2010 the then prime minister David Cameron challenged voters to kick him out if he missed the target. He did, and in a way, so did they. Encouraged by six years of failure to hit the target as home secretary, Theresa May has, now as prime minister, renewed the promise again.
How is this to be achieved? Leaving the EU won’t do the job alone: net immigration from outside the bloc has consistently exceeded 100,000. So attention has turned to a policy that many people regard as obvious: keep low-skilled immigrants out, and prioritise the highly skilled. For example, a recent policy paper published by the lobby group “Leave Means Leave” calls for a “moratorium on unskilled visas”. The paper proposes that working visas should be issued only to those who meet certain requirements, including a job offer on a salary of at least £35,000.
But this is an interesting slippage in the use of the word “unskilled”. About three quarters of UK employees earn less than £35,000, and as Jonathan Portes of King’s College London points out, the majority of nurses, primary schoolteachers, technicians, paralegals and chemists earn less than this figure.
Proposing an end to “unskilled migration” sounds reasonable to many voters; they might find it less reasonable if they realised that some definitions of “unskilled” would exclude a teacher or an intensive care nurse.
The word “cut” can also mean something different when it emerges from the mouth — or the Twitter account — of a politician. A few days ago the Labour leader, Jeremy Corbyn, tweeted: “Labour will stop the Conservatives’ £22 billion cuts to our #NHS immediately.”
No argument about the figure of £22bn: it is based on an estimate from NHS England two years ago. The problem is the word “cut”. The budget of the NHS in England is not being cut. It is increasing by £8bn after inflation.
The difficulty is that NHS England estimated that it needed not £8bn but £30bn. One might use many words to describe the £22bn shortfall — for example, “shortfall” — but “cut” is not one of them.
Amid all these surprising, confusing or deliberately vague labels, it was refreshing to see John McDonnell, the shadow chancellor, in action early in the election campaign. He promised that Labour would ensure that “the rich pay their way more”.
Normally, such statements mean very little. George Osborne, the former chancellor, was fond of saying that “those with the broadest shoulders” should bear the greatest burden, which sounds good but could mean almost anything. One definition of “rich” would be “earning more than £35,000” — most people don’t, as we’ve seen. Another might be “assets over £2.9m”, which according to the Institute for Fiscal Studies would put a British household in the top 1 per cent. Usually when politicians say “tax the rich” they hope that voters will hear “tax someone else”.
To Mr McDonnell’s credit, he actually produced a definition of “the rich”: people making more than £70,000 to £80,000 a year. That is admirably clear. Whether or not you agree with his definition or with his policy, at least we can understand it. This is rare.
Statisticians are sometimes dismissed as bean-counters. The sneering term is misleading and unfair. Most of the concepts that matter in policy are not like beans: they are not merely difficult to count, but difficult to define. If we don’t understand the definition there is little point in looking at the numbers. We have fooled ourselves before we have begun.
Written for and first published in the Financial Times on 12th May 2017.
My new book is “Fifty Things That Made The Modern Economy” – coming soon! If you want to get ahead of the curve you can pre-order in the US (slightly different title) or in the UK or through your local bookshop.