Risk, Education and Politicians

As usual I’m a trifle behindhand in my reading, so only now am I catching up with the Darwin Lectures  on Risk, a series of lectures given in 2010 and now available as a book or on your Kindle). This is a fascinating compendium of chapters looking at risk from many different angles, ranging from my Cambridge colleague Mary Beard on the ancients’ view of the dangers of life, to a much more modern take, as manifested by Oxford’s Lucia Zedner discussing Terrorism and Counterterrorism.  The introduction says that during the actual lectures there was even a short film about ‘the risk to one’s personal life of being too passionate and engaged in one’s work’, a risk to which I suspect I am definitely prone.

David Speigelhalter, another Cambridge colleague and whose formal job title is the Winton Professor of the Public Understanding of Risk, contributed another chapter, presented in his usual clear-headed style. He discusses that useful measure of risk, the micromort, a concept introduced by Ronald Howard, a professor at Stanford, several decades ago.  A micromort represents a one-in-a-million chance of death, and provides a useful way to compare different dangerous activities. It is, as Spiegelhalter describes it,  ‘the average ration of lethal risk that people spend each day, and which we do not unduly worry about’. Anything from giving birth to riding a bicycle for a mile can be quantified, at least approximately, in terms of this unit. (In case you’re wondering the answers are 0.05 micromorts for the latter; for the former it’s 80 micromorts in the UK, but more than twice that in the USA, something I think that says more about their healthcare provision for the less well-off than the dangers in a middle class hospital.)

Elsewhere Spiegelhalter has recently taken to pieces the statement that there is ‘a 20% increased risk of developing pancreatic cancer per 50 g of processed meat eaten per day’ as being really not a very helpful way of presenting the data. By looking at the  risks of eating a breakfast including bacon as an example compared with a healthier alternative, he shows this statement corresponds to an increase in deaths from pancreatic cancer of only from 5 to 6 for every group of 400 people eating one of the two alternative breakfasts all their lives. He has not, to my knowledge, evaluated the risk of being sent to jail if you are an Italian geophysicist who is deemed to have given expert advice to the Government, but he does have quite a lot to say about the horrible absurdities of the situation for the L’Aquila scientists. In terms of taking pains to make sense of risk for those less expert than him (which is just about everyone), he cannot be faulted; he is always worth reading.

The trouble is, most of us – scientists included – have a hard time of working out just what a particular figure really means for us as an individual. You are told you have a 60% chance of success in clearing up a medical problem if a certain procedure is carried out but also a 2% chance of death, what do you do? A doctor told me some figures along these lines a few years back, although I’m glad to say further investigations demonstrated that this was not at all the right procedure for my condition. I felt the figures were not at all helpful for me to work out what to do, although the risk of death felt uncomfortably high, and the success rate miserably low. Data for the population is hard to relate to oneself, where you may not be ‘typical’, just as the figure I quoted above for giving birth in ‘a’ US hospital, will mask huge variations depending on which hospital.

If you are not a scientist or mathematician, reasonably confident with probabilities, statistics and the rest, you will struggle even more. The British Academy has just released a quite extensive position statement regarding Quantitative Skills in the Social Sciences and Humanities. The BA is worried, not only because they see the lack of such skills threatening the UK’s status as a world leader in research and higher education in these fields, but because ‘quantitative methods underpin…effective evidence-based policy’ -  and we don’t see enough of that do we!  They are not talking about the majority of the UK population in this statement, but about a fairly privileged sub-set who still struggle to handle and interpret the data appropriately. I have discussed some of these issues previously, following a meeting at the British Academy, where I was trying to set the scene about the limitations of the UK’s maths education: as I pointed out then and as this recent statement reiterates, our post-16 maths education is unimpressive. Out of a survey of 24 countries carried out by the Nuffield Foundation, we (strictly speaking England) are the only one  with a participation rate of less than 20% in maths for the 16-18 age-group, and many students taking social science subjects at A level will barely be tested in quantitative methods during their assessments, although there is a significant variation across awarding organisations.

When it comes to what the Statement calls ‘skills for a modern participating citizenry’, they quote from the Royal Statistical Society’s ‘getstats’:

At every turn we make decisions to which risk, probability and sampling are relevant. Choosing what to buy or planning a journey, deciding on medical treatment and listening to the advice of the doctor; calculating earnings and benefits and how to save and invest: these are statistically rich decisions. We need numerical data to understand the performance of schools and hospitals, but also to understand why league tables can be distorting and unfair.

Our overall weakness in maths and statistics is now being recognized in Government, and maybe there is hope on the horizon. ACME (the Advisory Council for Mathematics Education) is being asked to advise on what could and should be done for post-16 mathematics in England by the Department for Education. But it cannot be forgotten that the DfE has its own agenda, as the way the story about a project in Cambridge to create enrichment activities for student and teachers at Maths A level appears to have been ‘spun’ to the media makes only too plain. If you compare the factual story about this project on the University of Cambridge website with the way it was written up in the press (led by the Daily Telegraph, with its wholly inaccurate title ‘Cambridge University professors ‘to set maths A levels’) the discrepancy between the reality and what the Ministry appeared to want to happen is clear.

On the day this particular story broke earlier this week, with the Secretary of State, Michael Gove, visiting Cambridge to launch the initiative, I was attending a so-called Joint STEM Ministerial Meeting in London.( I refer to it as ‘so-called’ because, despite the fact that both the Universities’ Minister David Willetts, and the Schools Minister Liz Truss were intended to be present, neither of them were – albeit the former turned up later in the meeting after both KS4 and A level reform and been heatedly discussed.)  I felt obliged to point out quite forcefully to the other Education representatives in the room, that the media had got the story completely wrong, to make it clear that my University is not trying to leap into the breach of ‘setting’ A levels unilaterally.

Nevertheless, it is good news that the DfE is looking at the post-16 maths provision. We have to hope that future generations are better equipped to make sense of many aspects of their lives, risk included. That will at least represent progress.  In the Darwin Lectures, Spiegelhalter says ‘Public figures need to act with humility when communicating about [these} uncertainties’. If they started doing this, that would also represent progress.

This entry was posted in Communicating Science, Education, Uncategorized and tagged , , , , , . Bookmark the permalink.

4 Responses to Risk, Education and Politicians

  1. richard hickman says:

    Add Training to Risk, Education, Politicians.

    I just watched a recent Horizon programme where children (4/5yrs?) were offered a choice:

    1 marshmallow now, or, if they waited 10 minutes, three later.

    This demonstrates a difference between Hazard and Risk. Hazard being the loss of 2 marshmallows or not, Risk being a person’s capability to decide one way or another to the best advantage.

    But of course, do I prefer 1 marshmallow now, or three later, when I could fall foul of deceit?

    And of course, I could decide immediately after the proposition was proposed, or I could choose to take one now after 9 minutes 59 seconds of deliberation.

  2. Ralph Lucas says:

    If Universities will not take the lead in proposing the next maths curriculum, who do they suppose will do it better? Not those responsible for the current one, surely?

    Certainly not universities only, though. As extensively discussed in Ben Goldacre’s Bad Science, and as you show in this post, risk is at once a science and a humanity. The Italian geophysicists got the science right, but got the human side culpably wrong, and it is fascinating to see so many scientists, who often and rightly bemoan the media’s lack of statistical grip or honesty, fail in their turn to understand.

  3. PL Hayes says:

    When it comes to what the Statement calls ‘skills for a modern participating citizenry’, they quote from the Royal Statistical Society’s ‘getstats’:

    Where, inexplicably and inexcusably, you’ll find examples of the ill-conceived and misleading nonsense which has been damaging science and undermining (public) understanding of probability and inference for years. If you already know all about the ‘inference scandal’*; what a frequentist ‘p’ value really means and why you need to look elsewhere for a measure of evidence for a hypothesis etc. check out this page – but put your tea/coffee cup down first! If you don’t, maybe start here. The truly horrible irony in all of this is that the technically vastly superior and conceptually and mathematically well-founded ‘alternative’ system of inference (orthodox stats isn’t really a system of inference at all) is also much easier to understand.


    It is a major scandal that orthodox methods continue to be taught at all to young statisticians, economists, biologists, and medical researchers; this has done irreparable damage in these fields for decades. –E.T. Jaynes.

  4. mary beard says:

    Thanks for the name check (still not quite sure what I was doing in the volume)! But there is great stuff here and in this whole area. I always bridle at the computer programme my GP has (centrally issued I guess) which claims to show me my risk of heart attack and death within the next 10 years, or whatever… I always want to say, “.. but Spiegelhalter says..”. Annoying for the GP I guess!