Thursday, 8 March 2012


Remember that survey of a baker's dozen young smokers that Otago healthists used as basis for recommending banning smoking outside of bars? The Otago Daily Times editorial page wonders, if costs to the health system are the reason for increasingly onerous restrictions on smokers, if there's any logical basis for stopping at smoking.
What social freedoms can be safe in the face of such overweening zeal?
They have a point, particularly as to risk and its relative social costs.
If, for example, costs of medical care are to be pre-eminent criteria in instituting prohibition, should society really continue to allow enthusiasts to ride motorcycles - or to indulge in any number of adventure sports, as a result of which death can be immediate rather than drawn out.
Such logic can be taken to absurd ends: given the high rates of melanoma in this country, and the concomitant social and medical costs, should not sunbathing be prohibited?
Or equally, to combat high levels of diabetes-inducing obesity, should pies and chips be banned?
The serious point is that there is a fine line between prohibitions that seek to minimise harm and social and economic cost, and those which curtail the individual freedom of choice which is a hallmark of developed, democratic societies.
I'd quibble here: any sane analysis finds that smokers pay roughly three times as much in excise as they cost the New Zealand health system, even leaving aside savings to the superannuation system.

The four authors of the paper surveying thirteen people (I wonder who got stuck surveying 4 instead of 3 respondents) respond in a letter to the ODT 6 March that proves rather too much. [Print edition; haven't seen an online version, alas.]
THE claim (ODT editorial, 28.2.12) that smokers make "conscious knowing choices" suggests they make a fully informed decision to become addicted to a behaviour that will kill half of them prematurely. Aside from lacking face validity, this assertion has no empirical basis. 
While most smokers agree that smoking is harmful, few understand all the specific harms it causes. Very few know the level of risk that they will suffer those harms or their prognosis if they do.
It's not necessary to know all of the specific harms to get to the right answer on whether smoking is bad for your health. I don't know exactly how drinking arsenic would kill me, but I know to stay away from it. If smokers systematically underestimate the risks involved in smoking, there can be a case for policy intervention. But what does the data say? Here's Kip Viscusi writing in 1990:
Both smokers and nonsmokers greatly overestimate the lung cancer risk of cigarette smoking, and the extent of the overestimation is much greater than the extent of underestimation. These risk perceptions in turn significantly reduce the probability of smoking, as suggested by an economic model of risky consumption decisions.
There's some evidence that smokers view themselves as being less likely to see the downside outcome themselves, but the same overconfidence bias tends to apply to all such risk elicitation surveys; there's consequently little basis for paying particular attention to smoking as compared to other risky behaviours. If rugby players accurately assess the overall risks involved in the sport but think they're less likely than others to have bad outcomes, does that give cause for government intervention? Back to Hoek, Edwards et al:
Using slippery slope logic to scaremonger avoids evidence-based decision making, something we'd suggest is fundamental to a "developed, democratic society". Adventure sports don't kill around 5000 New Zealanders every year and spurious attempts to argue by analogy aren't a substitute for research evidence.
Except that slippery slope logic is evidence-based. Anti-smoking policy shifted from protecting non-smokers to eradicating tobacco, with each step along the way bringing denunciations of the "slippery slope" people warning of the next step likely to follow. Anti-alcohol policy is starting to be seen as an extension of anti-tobacco policy. Anti-alcohol folks are learning strategy from the anti-tabbac; fat and sugar are next in line. Rizzo and Whitman also show how paternalist policy induces these slippery slopes. As for adventure sports, we have had calls in New Zealand for mandatory ski helmets; Nova Scotia's headed in that direction, with explicit citation of the public health costs. Scaremongering? Hardly.

Hoek, Edwards et al:
Our paper, to which the editorial referred, explored social smoking among a diverse group of young adults; all but one supported making outdoor areas of bars smokefree because this would reduce the pressures they felt to smoke.
It's worth remembering that that one person was 7.7% of the selected sample of 13. But it's also plausible that you could get majorities in sounder surveys agreeing that bans are good ideas. A better idea is letting those folks avoid bars with outdoor smoking areas.


  1. I'd quibble about one point: I'd say there is reduced freedom of choice in case of addiction. In the case of, say, drinking beer or avoiding arsenic you and I make an informed decision that balances our perception of risks and benefits. The incidence of alcohol, arsenic or skiing addiction seems to be much smaller than for nicotine and I'd suggest freedom of choice is larger.

    1. I know I've seen evidence on teenagers overestimating the probability of becoming alcoholics conditional on drinking, but I can't recall seeing anything on survey estimates of kids on likelihood of becoming addicted conditional on consuming cigarettes. If the risk assessment is wrong, first cut policy is information provision on actual risk, right?

  2. Very hard reading for us Eric to find out that a survey of 13 is meaningless ;
    and that Statistics are political,;
    and we can't eat chocolate unless we ask permission.