Friday, 16 December 2016

A world of pure imagination

From the Telegraph...

'Health by stealth' sugar tax could slash rates of childhood obesity by 10 per cent

The planned sugar tax could slash obesity among young children by 10 per cent and create “health by stealth”, experts from Oxford University say.

This is about Mike Raynor doing the Lord's work by producing yet another computer model that makes happy predictions about the sugar tax. You can read it here. As always, the words 'garbage in, garbage out' spring to mind.

If you wish to take such models seriously, that is your look out. Regardless of whether obesity rates rise, fall or stay the same, campaigners will produce another model in a few years claiming that the rate of obesity in 2018 was lower than it would have been had there not been a tax. That, too, will be treated as fact by their supplicants in the media. It is impossible to prove that nanny state policies fail. The goal posts are always moved. Scientific claims in 'public health' are unfalsifiable, which is to that they are not scientific.

But this blog post is not about the merits of predictive models. I am more interested in the media's response. The sugar tax study has been reported in virtually every mainstream media outlet in the UK with such headlines as 'Sugar levy is set to slash child obesity rates by 10%' (Daily Mail) and 'Sugar tax "will stop 10% of child obesity"' (The Times).

Evidently, a drop in childhood obesity of 10 per cent would be such big news that even a forecast suggesting it might happen makes it into every newspaper in the land. The clear implication from today's reports is that a 10 per cent decline would be a big deal and the sugar tax is therefore an important policy.

And that is rather strange, because on Tuesday the government published the Health Survey for England which contains all the official figures for smoking, drinking and obesity for 2015 (it always lags one year behind). It contains figures showing that between 2014 and 2015 there was a statistically significant decline in childhood obesity from 17.1 per cent to 14.0 per cent - a drop of 18 per cent. Among the 2 to 10 year olds whom the latest sugar tax study focuses on, the rate dropped from 16 per cent to 13 per cent - a decline of 19 per cent.

Since the sugar tax study used figures for 2014 as its baseline, the rate of childhood obesity has already fallen below what it predicted. Do you feel any better for it? Do you think your taxes are now going to be lowered as a result of obesity-related healthcare costs dropping?

You'd think there would be room in the news cycle somewhere for actual data, but I saw these figures reported nowhere. Insofar as the media reported the health survey, it was with stories about how children were less likely to smoke and drink than ever before. Obesity was not mentioned. The closest the Guardian came to mentioning the drop in childhood obesity was to say that 'child obesity remains stubbornly high', which hardly tells the whole story.

There is an element of fluctuation in these figures from year to year due to the sample size, and there is some suggestion that the 2014 figure was an overestimate (see graph below), but they are real figures, at least. They are not from a computer model.

Even if you ignore year-on-year changes and look at the longer term, it is clear that child obesity has fallen by much more than 10 per cent since its peak in 2004 and is only a couple of points above what it was 20 years ago. Why isn't this fact a cause for celebration while speculation about a lesser effect from the sugar tax is? Why do predictions about obesity get so much more coverage than actual rates of obesity?

Speaking of predictions, it is only two months since the Guardian - channelling the Obesity Health Alliance - predicted that obesity among low income boys was 'set to soar' to 43% by 2020. I have put my own money up for grabs for anyone who wants to bet that it gets anywhere close to 43%. The new health survey reports a drop from 27% to 20% in 2015 so it's looking good for me. It's only a shame nobody took me up on it.

And let's not forget the model in the Foresight report. Published in 2007, it predicted that the male obesity rate would be 36 per cent by 2015. This prediction was treated as gospel for years afterwards but this week the 2015 figure was finally published: it is 26.9 per cent. (The female figure was 26.8 per cent, but the Foresight prediction was less crazy for women; it predicted it would be 28 per cent. At the time the report was published, the rate was 24 per cent for both sexes.) Needless to say, none of the media who reported the original forecast have informed their readers that it turned out to be wildly off base.

The next milestone is 2030 when half of the adult population is supposed to be obese. As always, I will bet with anybody that it gets nowhere near this.

Time and time again in 2016 I have written about the 'public health' racket withdrawing into a world of pure imagination. The models have become more important than reality, not only for making predictions but also for remaking history. Obesity predictions get far more coverage than changes to the actual rates of obesity. Predictions about what sugar taxes might do get far more coverage than evidence about they actually do in places where they have been tried (ie. nothing). Regression models miraculously transform sharp increases in heart attack deaths into sharp declines and purport to show that policies save lives in places where lives have patently not been saved.

It is all very strange and rather discomforting. There is a narrative in place and mere facts cannot shift it. Unless you dig deep into Excel spreadsheets on government websites, you will not even find out about the most basic facts. If you rely on 'public health' campaigners and the media, there is a spiralling epidemic of childhood obesity and a sugar tax is needed to tackle it. It scarcely matters that the first of these claims is based on modelling that turned out to be wrong and the second is based on modelling that will, in all likelihood, also turn out to be wrong.   

Meanwhile, back in the real world, childhood obesity has dropped by 20 per cent in the last decade in Britain without a sugar tax but has not dropped at all in Mexico where there is a sugar tax. Never mind though, lads. Back to the models.

No comments: