The Norm Chronicles (7 page)

You know what’s coming next. A cough in everyday life is not what you might call a statistically significant event. But you’re watching
Casualty
. And so you know that this will be a triple heart bypass.

And you are right. Likewise, a pan of boiling water and a toddler can mean only one thing in fiction, and it must involve screaming. In stories, do fires ever just go out?

Risk is much the same. It is almost always framed by the thought of events, not by non-events, framed by things that happen, not things that don’t. As soon as we talk about the risk of heart disease, we are
thinking of those who will die of heart disease, not those on the other side of the odds who will be fine. The whole discussion of risk is primed by the thought of the bad stuff that happens. As Chekhov should have said, everything you hear about risk begins with a revolver on the wall.

All this may be true, you say, but how else can we talk about risk? What’s the alternative? The alternative is to approach the whole subject with our focus on what doesn’t happen, the non-events, the coughs that are just coughs and the fires that go out. Then risk would no longer be about the deadly, unhappy endings, it would also be about your chance of being fine if the risk didn’t, in the event, burn down the house.

This is an unorthodox take on the subject, admittedly. You’ll look in vain for the headline ‘No children killed on the way to school today’. Non-events are by definition systematically neglected in news coverage. Imagine the newsroom conversation that begins: ‘Anything interesting not happened today?’

But is the concept so daft? As we said earlier, a probability like 30 per cent implies a non-probability of 70 per cent. There are really two numbers, not one: two sides to the odds. This reminds us that the way risk is usually framed places one number in front of the other, with no equivalent word for thinking of them the other way round. What is the opposite of risk’s emphasis on events? Not safety, exactly. That doesn’t quite describe the nothingness we’re after. The dictionary and thesaurus are not much use, offering a meagre list of synonyms for non-event such as ‘damp squib’, which hints at how alien the concept is. We scarcely have words for it. But that’s because it is unfamiliar, not because it’s ridiculous. How would a newspaper describe the thing that didn’t happen, if it ever did? Would it see the point?

Take a real example. ‘A daily fry-up boosts your cancer risk by 20 per cent’, said the
Daily Express
in early 2012,
¹
about the effect on pancreatic cancer of processed meat. Not only does the
Daily Express
put the cancer revolver on the wall; when it also says that the risk is ‘up’, it caresses the trigger.

This ‘20 per cent up’ is a calculation based only on the things that happen. It begins with people who have or will have cancer and ignores all those who don’t or won’t. Five people in every 400 typically develop
this (very aggressive) cancer during their lifetime. If we take the figures at face value, the 20 per cent increase in risk if everyone –
every one of those 400 people
– eats an extra fry-up every day – and good luck to all who try – takes that toll from five cases to six. That is how we calculate a rise of 20 per cent. Twenty per cent of the original five is one additional case. We have gone from a ‘relative risk’, the 20 per cent, to a change in ‘absolute risk’ from 5 in 400 to 6 in 400, or 0.25 per cent. Relative risks make things look much more important.

Note that we have ignored – except by implication – what happens to the several hundred people in our sample who were unaffected before and are unaffected now, as did the
Daily Express
, as does just about every news report of cancer risk.

Let’s change this. Let’s begin instead by focusing on those who do not or will not have this cancer, the people for whom the revolver hangs idly on the wall and is never fired, those for whom the fire is never lit. Now we notice the 395 out of 400 who don’t ordinarily get pancreatic cancer. If all 400 eat the extra fry-up every day for the rest of their lives, one more will have pancreatic cancer, but 394 will still be fine.

Now it looks less risky, don’t you think? But we can make the numbers more striking by drawing attention to the fact that 399 out of 400 are unaffected by an almighty pig-out change in diet. There is, remember, only one extra case of this cancer, and the 399 were either going to get it anyway, or they were not and will be fine. Either way, 399 out of 400 are non-events as far as the extra fry-up goes. Yet this is exactly the same risk as the 20 per cent increase in risk if we concentrate only on the events and ignore the non-events.

The point is to ask if the way risk is usually talked about leaves the cup of danger pessimistically half-emptied by death, not half-full of life or survival. Except that in this case the cup is 399/400ths full of life, or at worst 394/400ths – but the newspapers look only at how much it empties.

Perhaps that’s what people want when they talk about danger. We want to talk about the 1, not the 399. But this is a choice, even if we are unaware that we have one.

We could, as we say, talk about survival instead. And by concentrating
on those who are unaffected, the danger in this case diminishes. It isn’t even our focus any more – living is – and the excellent prospect of survival is not much changed. We could also describe the pancreatic-cancer effect of an extra fry-up every day for life as roughly a 99.75 per cent chance of still being fine.

Figure 5:
If 400 people all have a fry-up every day, with an associated 20 per cent increased risk of pancreatic cancer, then one extra person (0.25 per cent) will get this cancer during their lifetime

In
Figure 5
above, is the risk of pancreatic cancer best measured by the people in grey, who get it, or by the people in white, who are fine? The extra chance of pancreatic cancer if all 400 eat a fry-up every day is
shown by the person crossed out: is that best expressed as an increased risk of 20 per cent for the greys or a 0.25 per cent decrease in the chance of being fine for the whites, or as a 99.75 per cent chance of still being fine for the whites? It’s all in the framing.

Although, following the work of psychologist Gerd Gigerenzer, there is a way to avoid percentages altogether, and talk about the pure number of people affected. Then, whether you want to talk about the increased risk, or the change in the chance of being fine, the measure is the same: one person in 400.

The man who ate an extra fry-up every day and didn’t get pancreatic cancer and was fine is a massively more faithful rendition of how life is. It is also ‘dull, dull, unbearably dull’. Newspapers never write it that way. People don’t tend to talk that way. But then, the whole concept of risk assumes a ‘negative frame’: risk is about the potential for bad, seldom the potential for life to go on, except by implication, an implication that must lurk in the background. Standing in the foreground is death. Is it any surprise which perspective holds our attention? So if cancer is shown hanging on the wall in Act I, then cancer frames our expectations. Thinking about what makes the risk go ‘up’ tightens the focus further. We’re mentally half-way to contracting it before we’ve even reached the last act.

Even MicroMorts do this. They too are about comparing the bad things that happen rather than the non-events. Perhaps they too need a complementary perspective, some sort of reframing. Maybe we need another new unit, an anti-MicroMort to describe the daily chance of being fine. How about a MicroNot (1MN) – a one-in-a-million chance that nothing fatal happens?

Then if we take as an illustration the average daily dose of acute fatal risk – 1 MicroMort – we can switch focus to the MicroNot to show that the average daily chance of an acute non-event – of being fine – is 999,999 MicroNots.

Then let’s say that you do something, hypothetically, to double your average daily MicroMort risk from 1 MM to 2MMs – up 100 per cent. But reframe the risk with MicroNots to talk about non-events and your MicroNot dose falls from 999,999 to 999,998, a
fall in your chance of being fine of 0.0001 per cent. Perspective, or framing, is all.

Another real news story illustrates a similar trick. Researchers found that a genetic variant – call it ‘X’ – present in 10 per cent of the population
protected
them against high blood pressure. Although published in a top scientific journal, the story received negligible press coverage until a knowing press officer rewrote the press release to say that a genetic variant – call it ‘not having X’ –had been discovered which
increased
the risk of high blood pressure in 90 per cent of people.
²
This story was widely reported, whereas talking about those to whom nothing happened wasn’t news.

Both good and bad endings are present in any risk of less than 100 per cent; otherwise there is no risk, and the future is certain one way or the other for all. But the entire framing of risk is around the bad and changes to the bad, even when there is not much bad to be had, even when the good is overwhelmingly more likely and not much dented even by extreme changes in behaviour.

Sometimes, but rarely, the bad is changed to the good. There was an advertising campaign on London Underground that proudly declared: ‘99 per cent of young Londoners do not commit serious youth crime.’ Sounds wonderful. But, taken literally, that means 1 per cent of young Londoners
do
commit serious crime, and there are around 1 million young Londoners, which implies 10,000 thugs running around. Not so wonderful. The switch between events and non-events, between fires that burn and fires that go out, can have a powerful effect on people’s decisions. DS has found that when statins are shown to reduce the risk of heart attack for over-50s by 30 per cent, people are keen on statins. But they’re not so keen when he also shows them that around 96 people in 100 will be unaffected if they take statins for ten years – they either would or would not have had a heart attack anyway – but might be vulnerable to side-effects. The risk hasn’t changed, only the way it is represented; but that changes people’s minds.

Does this make them irrational? For some, this is proof of human flakiness. Why aren’t people consistent when the risk is the same?

We think that’s unfair. A change in framing changes the context – that’s
what framing is. People unsurprisingly find it hard to weigh a risk against everything in life that’s relevant. Shown a risk in a way that emphasises danger, they react by saying it’s dangerous. Shown it in another that emphasises being fine and they say: ‘Oh, that changes things.’ Our experience is that, once they have seen it both ways, they are more consistent. Some experiments that purport to prove irrationality through a change of framing seem more like a trick than a test. Do people who change their minds when the framing is changed change them back again if the framing changes back, and continue switching every time? Of course not.

In relation to stories, think of the numbers as follows: imagine 5 bad things out of 400 possibilities. This probability consists of a numerator and a denominator: 5 is the numerator and 400 the denominator. Narratives or stories often concentrate only on numerators. We tell stories about people who do things and people to whom stuff happens – the 5 – not the people who do nothing and to whom nothing happens. This is almost the definition of what stories are. If you want to know about a story you ask: ‘What happens?’ And so we suffer from what is known as denominator neglect. We ignore the mass of people from whom our 5 happening examples are plucked.

Not all stories are so simple, it is true. Good fiction teases out and tests our expectations and sometimes frustrates them. It puts revolvers on the wall and leaves them there. It can be full of ambiguity. In 1925 Virginia Woolf distinguished her fiction from that of the past when she wrote:

If a writer were a free man and not a slave, if he could write what he chose, not what he must, if he could base his work upon his own feeling, and not upon convention, there would be no plot, no comedy, no tragedy, no love interest or catastrophe in the accepted style … life is not a series of gig lamps symmetrically arranged.
³

But this argument against orderly expectations has been used to exaggerate the modernist case. For it is also true that Hamlet’s agonies
about killing his uncle have exercised critics and audiences for 400 years with their lack of symmetrical, gig-lamp clarity, as does Hamlet’s every relationship. There are no signposting coughs in Hamlet, the gig lamps are not bright and the play is brighter for it. Good fiction has played for centuries with causality and the limits of our knowledge.

But do the stories we tell in the news media or about our own lives have the same sophistication? Life is at least as puzzling and messy as fiction. But do we, many of us, behave as if it were a bad novel, and try to impose more direction and coherence than many writers would dare? Risk is not, or should not be, a bad novel. In real life the gun on the wall usually rusts. So maybe non-events should be bigger news. The
auteur
in Mitchell and Webb’s sketch is that rare and uncelebrated thing: an artist of the denominator. Give him an Oscar.