Friday 22 March 2013

Tobacco control doesn't work, admits CRUK

If you want to see an example of churnalism at its most shameless, compare and contrast this press release from Cancer Research UK with this "news story" in the Daily Express. Barely a single word has been changed. It has also been reported by the BBC and the Guardian, and the gist of it is that underage smoking rates have rocketed and something must be done. Plain packaging is something, therefore it must be done.

Around 207,000 children aged 11-15 start smoking in the UK every year according to new research published today.

This means that nearly 570 children are lighting up and becoming smokers for the first time every day.

The new Cancer Research UK figures show this number has jumped by an extra 50,000 from the previous year, when 157,000 started smoking.

The youth smoking rate rose by 32 per cent in a single year?! Not only is that enormous, but it is the opposite of what should be happening. Recall that one of the anti-smoking mantras is...

A 10% Increase in Price Reduces Smoking Prevalence Among Youth by nearly 7%

There have been steep tax hikes on tobacco in recent years (shown below in £s).

Between 2008 and 2011, the average price of a pack of cigarettes rose by 22 per cent (roughly 15 per cent after taking inflation into account). Between 2010 and 2011—which is the period CRUK looked at—prices rose by 5.4 per cent. According to tobacco control dogma, this should have led to a fall in youth smoking of just under 4 per cent. Instead it rose by an incredible 32 per cent.

"Tobacco taxes are the most effective way to reduce tobacco use, especially among young people," according to the WHO. And let's not forget all those other policies—vending machine bans, graphic warnings etc.—that were implemented for the sake of the chiiiiiiiiiildren at around this time. Is CRUK's report a tacit admission that the neo-prohibitionist model is broken?

Well, up to a point, Lord Copper. As Taking Liberties points out, CRUK's figures do not actually tally with official estimates.

According to Smoking, drinking and drug use among young people in England in 2011:

... the 5% of 11-15yr olds considered regular smokers in 2011 is unchanged over 2010. The proportion of girls who are considered to be regular smokers actually fell from 6% to 5% in 2011. The proportion of 11-15yr olds who have never smoked increased from 73% to 75% (60% in 2005).

You can see the figures below for 11-15 year olds. The trend is very similar to the adult smoking rate, ie. a large decline until 2007 when a slew of anti-smoking policies began to be introduced and then almost no change. Certainly there was no rise in 2010-11.

This looks like another example of campaigners finding official estimates to be unhelpful and so concocting their own (see also the Scottish heart miracle and the Canadian MUP miracle). The trouble is that CRUK's survey seems to be much more prone to random variation than the official stats. The CRUK press release contains this telling line...

Analysis of the data showed that the 2010 figure was unusually low and this most recent figure is similar to the numbers seen in the late 2000s.

OK then. So 2010 was a blip in the data while 2011 was more typical (regression to the mean). The rate of youth smoking hasn't risen in recent years. It has stayed much the same. In other words, the tax hikes and neo-prohibitionist policies have made very little difference and CRUK's figures are dodgy.

Just another day in tobacco control then, and the media swallow it hook, line and sinker as usual.


The link from the CRUK press release to the research doesn't take you anywhere so we can't see what their figures are based on, but the following quote from their press release does not inspire confidence:

A survey among 12 year olds in 2010 found none were regular smokers, one per cent smoked occasionally and two per cent said they used to smoke.

But a year later in 2011 among the same age group of children, now aged 13, two per cent were found to smoke regularly, four per cent smoked occasionally and three per cent said they used to smoke.

Surely this wasn't the basis for their claim that youth smoking rates have risen?! Yes, 13 year olds are more likely to smoke than 12 years (and 14 year olds are more likely to smoke than 13 year olds etc.) but surely even the most incompetent/biased researcher would not use this as evidence that the number of underage smokers is rising. Or do I overestimate them? Either way, it is odd that CRUK would think that such a mundane and fatuous statistic was worth including in a press release.


I missed this explanation from CRUK. The figures are based on the official stats after a lot of jiggery-pokery. The following garbled words are the closest thing we have to a methodology...

This new analysis is based on figures taken from the Smoking, Drinking and Drug Use Among Young People in England.

The “started smoking” figure is calculated by comparing the smoking rates at each age with the smoking rates of the same group in the year before.

So 12 year olds in 2011 are compared with 12 year olds in 2010. Both regular – one or more cigarette per week – and occasional smokers – less than one cigarette per week – are included.

There were an additional two per cent smokers in 2011 than 2010 (regular smokers from zero per cent to two per cent) but in addition one per cent of the 12 year old smokers in 2010 gave up (used to smoke up from two per cent to three per cent) so an equivalent number of smokers must have started (or else the one per cent smokers giving up and starting would cancel each other out) so there are actually three per cent new children smoking.

The three per cent is applied to the UK population to give a number of new children age 13 who start smoking in the UK. This is repeated for the other age groups and the totals added to give a figure for the number of new children.

Is anybody else reminded of this?

Basically, they're treating numbers that have been rounded up or down to the nearest percent as if they were precise figures. It is, for example, clearly implausible that there was not a single 12 year old smoker in 2010, but the figure can still be cited a 0% if the figure is 0.49% of lower. CRUK treats 0% as if it was zero people and 1% as if it was 1.0% etc. The 50,000 extra smokers they report is a rounding error that comes from inappropriately mixing lots of other rounded numbers together as if they were exact proportions of the school-age population.

It's imaginative junk science, I'll give them that, but it's junk all the same and it is totally inconsistent with the Office for National Statistics' conclusion that: "The proportion of 11-15yr olds who have never smoked increased [between 2010 and 2011] from 73% to 75%."

If this kind of thing is representative of Cancer Research's work, I wouldn't hold your breath waiting for them to find a cure for cancer.


Ian B said...

I followed the link trail from the report in the Telegraph (headlined as a massive increase in youf smoking, sigh) and the telling paragraph (to me) in the CRUK report is based on the "used to smoke" figures; basically they're using some dodgy arithmetic to boost the "new smoker" figures-

"If, from a thousand children aged 12 in 2010, 10 smoked regularly, 20 smoked occasionally and 20 used to smoke, and, from a thousand children aged 13 in 2011, 30 smoked regularly, 20 smoked occasionally and 40 used to smoke, there are clearly an additional 20 smokers in 2011 than 2010 (current smokers have increased from 30 to 50). But, in addition, 20 of the 12 year-old smokers in 2010 have given up. To take account of these children that used to smoke, an equivalent number of children must also have started smoking (or else the 20 smokers giving up and starting would cancel each other out), so there are actually 40 new children smoking."

So if (simplifying) last year 2 people smoked and 2 people used to smoke, and this year 2 people smoke and 2 used to smoke, they conjure up 2 "new smokers" since the 2 who used to smoke this year must be the 2 who smoked last year.

Combining this with an absurdly low definition of a smoker (1 per week!), anybody who occasionally used to have 1 or 2 but now doesn't at all gets added to the "new smoker" figure with which they wanted to headline in support of the packaging ban. Then the media lead with an implication (if you haven't read the report) that youth smoking has risen by this figure.


Michael J. McFadden said...

Chris, when it's tax time next year, just underpay by 50,000 pounds. If they catch you, just say, "Oh! Sorry! It was a rounding error! Perfectly acceptable you know..."

It reminds me of figures I saw on e-cigarette use a week or two ago. Evidently something like 67% of 12 year olds who had tried e-cigs were not regular smokers -- thereby proving that e-cigs were creating smokers! However, not only did the study study NOTHING about whether these "experimenters" ever went on to smoke regular cigarettes, they also overlooked a blatantly obvious fact. MOST 12 year olds are not "regular smokers" so obviously most 12 year olds who'd try a puff from an e-cig would be unlikely to be regular smokers. If they'd asked 5 year olds they probably would have found that 99% of all five year olds who'd tried a puff from an e-cig were not regular smokers either. Of course there were only three such cases in all of the UK, but it still comes out to 99% after they churn the figures!