Facts and attitudes

This morning I was prodded by the scourge of epidemiocracy, Chris Snowden, to read this piece by Theodore Dalrymple. What most struck me was not the main argument (I find predictable agreement almost as wearing as disagreement) but this piece of supplementary information:

A higher proportion of the Dutch population smokes than average for a developed country (27 percent), and fewer Dutch people are aware of secondhand, or second-lung, smoke — that breathed in from other people's tobacco — than any other comparable country.

Why should that be? I think it demands an explanation. Certainly the Dutch population cannot easily be classed as ill-educated or poorly-informed. (I have been sworn at by a drunk tramp on an Amsterdam tram who switched instantly to English invective when he realised that it was going to be more effective in my case.) My mind leapfrogged towards ideas about the Dutch liberal tradition. They choose not to know, because they do not like to hassle people about their private behaviour, perhaps...

Unfortunately there are no sources quoted. When I looked for stats and background info, I found something even odder. That remarkable factoid contains no truth.

The OECD statistical library agrees with that 27% average - if it is actually daily smoking for males 15 and over. But it places The Netherlands fractionally below average, equal with Germany and slightly above Belgium for the proportion of males who smoke (26%), with slightly more women than either (20%).

How about "awareness of second-hand smoke". The points in the article about "relatively high" Dutch smokishness appear in less critical articles such as this one in Salon. (Which itself hints that it relies as a source on one Lies Van Gennip, director of the national tobacco control center.) Here we have a hint of the source for the "awareness" figure.

In a global survey on smokers' awareness, only 61 percent of Dutch smokers agreed second-hand smoke was dangerous to non-smokers — much lower than smokers elsewhere, including Mauritius, China, Brazil and Mexico.

"Dutch smokers are among the least informed about the harms of smoking and second-hand smoke," said Geoff Fong, at the University of Waterloo in Canada, who heads a program that monitors smoking policies worldwide.

Note the built-in interpretation: failure to agree counts as being ill-informed. I googled down the global survey mentioned. It appears in the BMJ for 4 April 2011 under the headline "Dutch smokers are "alarmingly" ignorant of harms of passive smoking, study finds." The original findings do indeed appear under the aegis of the University of Waterloo here (pdf) But are published on behalf of 'The International Tobacco Control Policy Evaluation Project: ITC Netherlands Survey' — the presentation of which suggests it is intended to drive Dutch policy, and the naming of which suggests we should worry about that 'global survey' point. Some (more) cherry-picking, perhaps?

Inspecting a bit further suggests there is reason to worry. See here. The ITCPEP doesn't do a global survey. It surveys different countries at different times using different methods. The most recent French survey (2009) was a telephone poll with respondents reimbursed; the most recent Dutch one (2011) was an online survey. The critical common question is not "Do you think your smoking harms others?" but "How often, in the last month have you thought about the harm your smoking might be doing to other people?" In both surveys, the critical question is preceded by questions about respondents' attempts to give up and their degree of addiction, but in the Dutch survey that is the beginning. In the French survey there is much prequalification apparatus including emphasis of the importance of the survey itself, and information sought about the individual respondent's household. Longitudinal comparisons on a single country might make sense if individual surveys are consistent; but comparing attitudes in two countries on this basis does not.

We know nothing from the ITCPEP about the beliefs of either Dutch or French smokers concerning the dangers of environmental tobacco smoke. They were not asked. But we cannot even compare their relative preoccupation with whether they may be harming others—what they were asked—because they were asked at different times, in different ways, in different contexts.

The only reason for making the comparison in the first place was to condemn Dutch views as "ignorance", but even the variance in views is a pseudo-statistical phantom, if you can be bothered to look into the detail.

I am interested in variation in public attitudes and in political culture and their relationship to policy. But it is devilish hard to find out about those relationships when even critical discussion, such as Dr Dalrymple's, is predicated on 'facts' whose selection and interpretation is determined by the attitudes of interested researchers. Even specialist commentators are seldom suspicious enough to do more than re-word the press release and cast it in the light of their own attitudes.