Data dredging (or torturing the statistics until they confess)

I mentioned last month that I feature in the March/April issue of Tobacco Asia.

Based on a series of Q&As, Simon Clark - The Passive Smoking Myth is now available online.

Talking of which, I was reminded last week of something the late Lord Harris once wrote.

Chairman of Forest for 20 years until his death in 2006, Ralph was highly sceptical of the claim that passive smoking is a threat to non-smokers.

He first addressed the issue in a book, Murder A Cigarette, that he co-wrote with Forest researcher Judith Hatton in 1998.

Six years later, as part of our campaign against the proposed public smoking ban, he wrote an essay - Smoking Out The Truth - that directly challenged the chief medical officer on the subject.

Written in Ralph’s characteristically flowery style, it nevertheless made a strong case for the argument that environmental tobacco smoke (aka passive smoking or secondhand smoke) was not the deadly killer it was alleged to be by anti-smoking campaigners and the CMO.

In also highlighted how difficult it was - and still is - for anyone in the medical or scientific community to question the passive smoking narrative:

Privately I have encountered Very Important Persons in the medical world who, in response to my earnest enquiry about ‘passive smoking’, have dropped their voices and looked around furtively before assuring me there was ‘nothing in it’, except for a possibly adverse effect on serious asthmatics. Since my VIPs would prefer not to be quoted I conclude with two prominent authorities who did venture to put their doubts on the record – and brought retribution upon their heads.

Take the plight of a venerable scientist commanding universal respect as the first to establish a link between cigarette smoking and cancer, namely Sir Richard Doll. In 2001, in the relaxed atmosphere of the BBC Radio 4 programme Desert Island Discs, this 90- year-old doyen of epidemiologists confided to presenter Sue Lawley: ‘The effect of other people smoking in my presence is so small it doesn’t worry me.’ When the gist was publicly quoted there was an immediate outcry from the anti-smoking propagandists at ASH, in response to which Sir Richard felt obliged to state he was only ‘speaking personally’. But of course, how else would an honest man – even a conscientious anti-smoker – be expected to speak?

Since then I was surprised to hear that Sir Richard seems to have overcome his doubts. Following a 2004 newspaper report in which he appeared to have retreated from that view, I wrote to him seeking clarification. He replied courteously as ever that he had not abandoned his belief that relative risks between 1.0 and 2.0 are ‘difficult to establish’ but he now added that they were not ‘impossible to establish’. He reaffirmed his long-standing view that ETS ‘should be accepted as a cause of lung cancer, albeit one which produced only a small risk’. It suggested less than whole-hearted support for the alarmists who use statistics to justify a ban on smoking in all public places.

A more overt example of intimidation followed the publication by the prestigious British Medical Journal in 2003 of a study by two American medical scientists, Enstrom and Kabat, who drew on data from the American Cancer Society which tracked 118,000 Californians over 40 years (1959-1998) and concluded: ‘The results do not support a causal relationship between ETS and tobacco-related mortality ...’

Publication provoked a fierce barrage of 140 letters, many of which treated the editor to lectures on ‘the evils of tobacco’ and accused him of being ‘naive’, ‘stupid’, ‘mad’, ‘irresponsible’ or having ‘darker motives’. Every known lobby joined forces in an attempt to bully the editor to withdraw.

Instead of retraction or resignation, which the lobbyists impudently demanded, the then editor, Dr Richard Smith, was stung into a robust rejoinder. He explained that, although ‘passionately anti-tobacco’, the BMJ was not ‘anti-science’ and ‘the question [of whether passive smoking kills] has not been definitively answered’. (Little over a year later, at a fringe meeting organised by the Institute for Public Policy Research during the 2004 Labour party conference in Brighton, he appeared to have retreated from that view, although without revealing his reason.)

But the passage I was reminded of last week concerned Ralph’s response to the claim that hundreds if not thousands of non-smokers were dying each year from secondhand smoke.

According to the noble lord:

If laymen dare to question any of these guesstimates and projections, the sophisticated statisticians take refuge behind their computers which have been heavily programmed to incorporate a variety of elaborate assumptions and statistical techniques. And since researchers have discovered that the bigger the reported risk the better the chance of attracting funding and getting their results published (known in the trade as ‘publication bias’), they have exerted much ingenuity in what is known as ‘data dredging’ – that is, torturing the statistics until they confess!

‘Torturing the statistics until they confess’. What a lovely line and how true of so much of what passes for ‘research’ these days.

But back to Murder A Cigarette. Published in 1998 it was promoted as follows:

The aim of this book is not to encourage others to smoke, but to sort out fantasy from fact in the debate on the freedom of adult individuals to smoke - always with courtesy and considerations for others.

The authors acknowledge smoking as a risk factor in cancer; but as only among several leading predispositions, including heredity, diet, personality, lifestyle, social class and location.

They aim to expose the exaggeration and deception of the "medical mafia" and the "cancer establishment" and show "passive smoking" to be a bogus invention designed to stigmatise smokers.

How we could do with an updated edition that covers the last 20 years. Given the Government’s commitment to a ‘Smoke Free England’ it would be a welcome voice of common sense based on facts not propaganda (aka estimates and calculations).

Sadly Ralph and Judith died within two years of one another, Ralph in 2006, Judith in 2008. They were both in their eighties (Judith possibly older, I didn’t like to ask!) despite a lifetime of smoking.

Rightly or wrongly they went to their graves believing the propaganda about smoking and passive smoking was greatly exaggerated.

Were they wrong?