I try not to get to irritated by poor standards in stats in journalism as much as I used to. Random Daily Mail reports based on flaky sources I can just about get past without more than a few seconds of seeing red (mostly by not looking, I admit). But, sometimes you hear something from more credible sources that feels surprising enough to want to check whether it really is true.
Over Christmas, I heard reports that the UK was the second most obese country in Europe, behind Turkey. Whilst I suspect obesity is an issue, and UK average diet is hardly good, I was surprised to hear it was relatively that high. It also felt fairly hard to measure on a like for like basis, raising the risks of dodgy stats
And, it wasn’t the Daily Mail; I saw it on the BBC ‘what the papers say’ (see here), with reference to the Telegraph (see here). Though the Daily Mail did cover it too, here. And the Times here, and the Guardian here. They all point to an OECD report here, which I would expect to be a pretty solid source. But, reading the source info showed a bit more light.
Page 59 is where the relevant data is graphed (See here). And, it does clearly show the UK as second highest. But, it also flags the UK, along with Hungary, Ireland, Luxembourg, Czech republic and Slovak republic as being measured rather than self reported. 5 of these are in the top 6 countries. And, there is a callout on the previous page noting that “Estimates from health examinations are generally higher and more reliable than from health interviews”.
What it doesn’t do, which feels pretty naughty, is to give any indication on what correction one should allow for, given that these are effectively two different statistics. However, there was a review done in Canada in 2008 (see here) that did do both measures, which found that errors in self reporting moved the percentage of those classed as obese from 15% to 23%. If one had the base information you could probably approximately correct for this in the 6 that were measured, but by good fortune for this purpose, the UK is listed 24.7%, which is pretty close to 23%, so at a first order, one could assume that the UK was roughly 16.5%. The average will also be overstated by the contribution of these different metrics, but as it stands it is 16.7%.
So, from fag packet maths, I would judge the UK is pretty much average on obesity. Not that good for an avoidable issue that causes harm, but nothing notable. Which feels more like what I expected.