The eminent statistician Professor Sir David Spiegelhalter OBE FRS was on Radio 4’s Broadcasting House on Sunday morning, helping to review the papers. Prof Spiegelhalter is Winton Professor of the Public Understanding of Risk in the Statistical Laboratory of the University of Cambridge. He is concerned with the use and misuse of statistical data in public discourse. He looked at several reported scientific surveys (“a crop of rubbish”) with headline-grabbing results and reiterated the question, “Who commissioned the survey?” He was making a point about bias. One of the other panel members cut in with a throwaway comment to the effect, “Well we know that statistics are 99% made up”, to which Spiegelhalter replied, with more than a hint of sarcasm, “Well I’ve never heard that before!” Paddy O’Connell in the chair told us that the professor pulled a weary face – useful to know when the media platform is radio. It’s the old adage attributed to Mark Twain, “There are lies, damned lies, and statistics”, which is often interpreted as a description of a downward spiral of misinformation and disinformation. But there is another way of interpreting the adage. There are, on the one hand, lies and damned lies; and, on the other, there are statistics. I venture to say that this is Prof Spiegelhalter’s message. The fact is that in this age of fake news, data are under attack, not just because they may be false, but because they may be, accidentally or wilfully, misinterpreted.
It so happens that on Thursday evening I was in Glasgow University to hear the inaugural Bowman Lecture, in honour of a modest and eminent man, courtesy of the School of Mathematics and Statistics, given by the self-same Professor Spiegelhalter. This public lecture was entitled Trust in Numbers. The Prof is a skilled communicator and it was no surprise that the lecture was a sell-out. It was also noticeable that the audience members filling the Sir Charles Wilson Hall were of all ages (I wish I could provide a graph here) including a substantial number of school students. Given that statistics is sometimes considered a dry subject, this was heartening.
The professor is on a mission; and it is one of public education. I have to say that I felt right at home listening to his lecture, not just because it was full of medicine (breast cancer, prostate cancer, processed meats and bowel cancer, safe alcohol consumption limits), but precisely because he was enthusiastically conveying a message, in much the way that I vividly recall the Australasian College for Emergency Medicine championing the Emergency Medicine cause during the 1980s and 1990s. It is always a pleasure to hear somebody who feels passionately about some cause, particularly if they possess the great qualities of rhetoric – coherence, fluidity, responsiveness, and humour.
The difference in statistics between relative risk and absolute risk was colourfully depicted. Some of the lurid Red Top banner headlines misconstruing then sensationalising a piece of medical research caused great hilarity within the hall. I was interested in the professor’s view on the latest medical advice about alcohol consumption. Over the years the “safe limit” has steadily declined. 28 units a week, then 21, then 14, for both male and female. And then, zero. No consumption is safe. Actually, as that graph extrapolates to zero, the evidence becomes distinctly hazy. Prof Spiegelhalter announced his intention to have a glass of wine after the lecture. He said, “I’ve created a new statistical unit – ‘NND’. Number needed to drink.” He added, “The medics like that one.” That is because we are familiar with “Number needed to treat”, that is, the idea of the number of patients needed to treat with Therapy A before a positive gain in a single individual is achieved. (To this, parenthetically, might be added the much less researched notion of ‘number needed to harm’ – the number of patients needed to undergo Therapy A before someone experiences a significant unwanted effect.
I was very intrigued by his concept of “inoculation”. This was a kind of pre-emptive strike. Some piece of misinformation is doing the rounds and you disable it by saying, “Isn’t it awful that everybody thinks…” and you reproduce the false information, as if you are supplying the recipient with a little piece of toxic material in order to protect them from the full effect.
After the talk, there were questions from the floor. One of them was whimsical (or maybe not): “Would the public understanding of statistical data be enhanced if Eton were abolished?”
Mischievously, the Prof asked, “Is this being recorded?” And then, tactfully, “I’m uncertain.” This chimed, because he had spoken about uncertainty, and the importance of owning up to uncertainty. Here, I was reminded of the antepenultimate episode of a BBC series nearly half a century old, The Ascent of Man, presented by another great scientific communicator, Jacob Bronowski. To me, Episode 11, Knowledge or Certainty is a little masterpiece. It starts with an exploration of the portrait of Stephan Borgrajewicz by Feliks Topolski – “Lines, possibly, of agony” – and ends at the crematorium at Auschwitz. On the way, Bronowski discusses Heisenberg’s Uncertainty Principle and the idea that our failure to pin down, say, the exact position and the exact velocity of an electron is not due to the crudity of our instruments, but integral to the world of quantum mechanics. Bronowski preferred to call this idea the Principle of Tolerance. Of course, at the time these ideas were coming to fruition, the 1930s, tolerance was vanishing from Europe.
(Parenthetically, if you get a chance to see the Bronowski, you will notice the very limited use of “background” music. I can recall the sparing use of a solo clarinet – is it from Olivier Messiaen’s Quartet for the End of Time? – composed, and first performed, in a concentration camp. These days, televised scientific documentaries are rendered unwatchable, certainly unlistenable, by their incessant background of musical drivel.)
George Santayana told us to study the past in order to avoid repeating it. But history never really repeats itself. So while the 2020s will not be like the 1930s, they might be even worse, if we lose the urge to separate truth from falsehood, and we become ever less tolerant. Yet Professor Spiegelhalter is a beacon of light.
So is Glasgow University. How wonderful to be back in the hallowed cloisters of Gilmorehill. I was back in these regions again on Saturday. A medical student sought some career advice. Why a millennial should choose somebody with my ramshackle CV to impart “wisdom” I can scarcely imagine. I wouldn’t dream of telling anybody what to do. Risky business! But I hope I was a reasonably effective sounding board. And I am quite sure I benefited more from the encounter than she did.
jamescalumcampbell 