Smoking legislation works, claims new study

A study funded by the Department of Health shows that raising the age at which teenagers can buy cigarettes from 16 to 18 had a dramatic effect on smoking prevalence in the under-18s.
According to the study, published in Addiction by Jenny Fidler and Robert West of University College London, smoking in this group fell from 23.7 per cent pre-legislation to 16.6 per cent afterwards, a fall of 7.1 percentage points.
Excellent news if true, but comparison with ONS smoking prevalence data raises some questions. The ONS data for 2006 (pre-legislation) shows that 20 per cent of 16-19 year-olds smoking; for 2008 (post-legislation), the figure is 22 per cent.
These age categories are of course different. The new study looked at those aged 16-17, while the ONS data covers 16-19 year-olds. Could this difference obscure big changes in those aged 16 and 17? The data also comes from different sources: the Addiction data from monthly surveys carried out as part of the Smoking Toolkit Study, the ONS data from the General Lifestyle Survey (Smoking and Drinking in Adults, 2008, Table 1.1). The ONS covers Great Britain, the Addiction study England only.
However, if we combine data from the lowest two age groups in each survey, so as to measure smoking in the group aged between 16 and 24, a direct comparison between the two surveys can be made. The Addiction study suggests that pre-legislation 33.5 per cent of this age-group were smoking, while the ONS data for 2006 registers only 25.5 per cent. After the legislation, the Addiction study shows 29 per cent remained smokers, while the ONS figure for 2008 is 26 per cent.
There’s no telling which set of data is more nearly right. But I’ve a suspicion the new study may be exaggerating the effect of the legislation. For example, it shows a relative fall in smoking prevalence in 55-64 year-olds of 19 per cent (from 21.5 to 17.4 per cent) although this group was unaffected by the legislation. For comparison, the relative decline measured in 16-18 year-olds was 30 per cent.
There seems no reason why prevalence should have fallen so fast in the older age group, especially as the ONS data shows no change in the 50-59 age group and a 1 percentage point increase for the over-60s between 2006 and 2008.
The authors of the Addiction study consider but dismiss the possibility that those aged 16-18 might be reluctant to admit smoking post-legislation, once buying cigarettes had become illegal (though smoking them had not). I’m not quite so sure about this.
Future ONS data should show if there really has been a dramatic fall in smoking prevalence, as those who gave up grow older. But in general the decline in smoking is a slow process, and the changes shown in the new study appear too large to be entirely plausible.
It's taken a decade for overall smoking prevalence to have fallen by 7 percentage points. So it would indeed be remarkable if a decline of the same size had been achieved in 16-18 year-olds by a single piece of legislation over just a couple of years.