Last time I discussed ONS figures on drinking, I pointed out that in 2006 the number of units in a glass of wine was changed to reflect the growing trend to larger glasses. I had no problem with this change in calculation because it's true that on average we're drinking from larger glasses, so the definition of 'a glass' of wine needs to keep pace with this.
But it did create an apparent huge one-off jump in alcohol consumption, particularly among women. However, this was NOT an increase in drinking - it was a change in methodology. It meant that the figures in the years before the recalculation should probably have been higher, and meant that any figures coming after the change could not be compared directly to those before the recalculation to give any kind of accurate trend. At the time, the ONS said: "“It should be noted, however, that changing the way in which alcohol consumption estimates are derived [in 2006] does not in itself reflect a real change in drinking among the adult population.”
Get that? That's the ONS saying it - the people who compile the figures.
Consumption was on a downward trend before this recalculation. After the jump caused by the change in calculation, it resumed this downward trend. In other words - let me spell this out as clearly as I can, because it seems to be a difficult thing to understand - THE OFFICIAL ONS FIGURES SHOW THAT ALCOHOL CONSUMPTION/BINGE DRINKING IS IN LONG TERM DECLINE.
So why in March 2011 does the ONS then issue a press release that states: “The percentage of females consuming more than the weekly recommended units of alcohol has increased by a fifth since 1998”?
Is this true? Or did they forget the change in their own methodology that they themselves were previously so keen to point out, in order to ensure people read the figures correctly?
The answer is: yes, they forgot to take account of their own methodology change, which led to them releasing a false story about alcohol consumption to an anti-alcohol national press!
They did at least have the decency to point this out, reissuing the press release with a clarification on the front page that reads:
Corrections have been made to reported trends in alcohol consumption in this article, published on 31 March 2011. The errors are unrelated to estimates of output, inputs and productivity.
In Annex C, figure C.5 illustrated trends in alcohol consumption from 1998 to 2009, using estimates from the General Lifestyle Survey (ONS 2010), but omitted references to a change in the estimation methodology in 2006. The change means that trends over the whole period do not necessarily reflect changes in drinking habits.
Accordingly, explanatory footnotes have been added to figure C.5 and paragraphs C.2.7 to C.2.10 in Annex C. References to alcohol consumption in the main article (Table 4.2) and the News Release have also been amended.
ONS apologises for any inconvenience caused.
According to the Liberal Conspiracy blog, the ONS has apologised to the Portman Group for the error.
According to the Straight Statistics blog, which helpfully found this little clarification for us, the Portman Group wrote to the Daily Mail and pointed out this error, but the Mail has refused to print this correction to a factually inaccurate story they ran, and is no longer accepting comments on the online version.
I wonder why?
The Telegraph story is also still up online and uncorrected.
A special prize goes to anyone who can find a single UK media outlet clarifying the story with the correct data.
Thanks to Jeff Pickthall and to Dave Boyle for alerting me to this gem.