The argument for banning smoking in public places was largely won on the basis of ‘passive smoking’. Once it had been shown that smoking was harmful to non-smokers, a convincing case could be made for a ban in the interests of protecting the innocent bystander.

Yet, the scientific evidence was far from watertight. Many studies had shown no link between exposure to second-hand smoke and ill health; indeed, more frustratingly for health campaigners, some even suggested that a non-smoker was slightly better off living with a smoker than in a smoke-free household. Although counterintuitive, this finding is not impossible – there are quite a few areas of human health where mild exposure to toxins is better than none at all.

Bluntly, the smoking ban was justified on the back of some rather dodgy statistical science – more ‘policy-based evidence-making’ than ‘evidence-based policy-making’. A few scientists dissented. Interviewed on Radio 4 late in life, Sir Richard Doll, who was among the first group of scientists to establish a link between smoking and lung cancer in 1950, pronounced himself completely unconcerned by the presence of smokers.

And yet, the smoking ban has proved to be a huge success.

The smoking ban is an example of a right thing done for the wrong reason. Writing in his latest book Under the Influence, Professor Robert Frank, a friend and one of my behavioural heroes, has proposed a much more robust justification for the smoking ban – and a far better explanation for why it worked so well. The act of smoking encourages other people in your peer group to smoke. Hence, driving smoking out of mainstream social life removed the copycat instinct that made starting smoking easier and quitting harder.

As Frank convincingly points out, the harm that arises when you allow people to smoke freely in public arises far more from social contagion than from chemical exposure. The visible presence of any smoker in a group of friends hugely increased the odds that others in the group would take up smoking, or that smokers would fail to quit.

Even after correcting for the natural tendency of people who share a vice to cluster together, research cited by Frank shows that, in a group of five people where one smokes, if one additional person takes up smoking in that group, it will increase the likely number of smokers (or non-quitters) in the group by two. And this ignores any effects the new smoker may have on people outside the group.

But, even though it is scientifically far more convincing, we probably would not have accepted the argument as sufficient to justify a ban. It strikes at our deep-felt belief in individual agency. The unshakeable notion that we are – and should be – the exclusive authors of our own behaviour. I’m as guilty of this delusion as anyone; I worked with Mark Earls when he was writing Herd, yet it took me some years to realise that he was fundamentally right.

Yet, most large-scale changes in attitude and behaviour have perhaps succeeded more as the product of social contagion than individual persuasion. The widespread acceptance of same-sex marriage. The decline in drink-driving. The rapid decline in overt racial prejudice. Not dropping litter. Wearing denim. All have come about more through the wider social force of imitation than individual agency.

Yet, it remains the force that dare not speak its name. Economics is based on the idea of methodological individualism. This means that, when we measure the effect of legislation or incentives, we only look for the initial effect on individual behaviour – not the magnified social effect that comes later.

This may cause us to hugely underrate the value of marketing – and to misdirect it. It seems entirely plausible to me that, when I buy an electric car in April, it will cause two other people to buy one (frankly, I’m only getting one because my brother has one). Likewise, if I were to install solar panels, my action may be three times more effective at cutting CO2 emissions than I think, because it encourages my neighbours to do the same.

Yet, in the very early stages, advertising for solar panels might appear to be very ineffective, precisely because they are rare, and seeding an unusual behaviour is hard. But that is precisely when mass advertising may work best. Later, when solar panels become more commonplace, and that sigmoid adoption curve starts bending steeply upwards, my advertising will look extraordinarily effective, when much of the heavy lifting is being done by social imitation instead.

If this is so, conventional approaches to measuring advertising may cause us to do it least where it is needed most. And vice versa.

Rory Sutherland is vice-chairman of Ogilvy UK

This article was first published in the April 2020 issue of Impact.