Mine’s a double!

by Chris Bertram on September 2, 2005

Via “Radley Balko”:http://www.theagitator.com/, a “rather nice BBC article”:http://news.bbc.co.uk/1/hi/magazine/4200056.stm about the crappy statistics behind the claim that the British are engaging in more binge drinking.



Charlie B. 09.02.05 at 10:18 am

I’m sorry to say it seems a pretty feeble effort to me. The suggestion that increases in over-the- counter and over-the-bar sales of alcohol are replacing home-brew is ridiculous, and no data are provided. The remainder of the article is similarly lacking a single statistical argument. I am deeply suspicious of official statistics about everything, and profoundly opposed to the way every kind of social interference is based on and driven by statistical claims – especially “comparisons” with other countries. No area of life is free of bureaucratic meddling linked to statistical measures. Unfortunately, this article contributes nothing to the necessary debunking.


Jack 09.02.05 at 12:06 pm

There is a big statistical argument there. My favourites are:

Use of gross figures not per adult capita for overall growth of consumption.

Poor definition of binge.

Other unquantified effects such as non-inclusion of homebrew, period of measure, are not quantified in the article but also not in the report. This is a serious failing in the report. Readers of such reports should have higher expectations.


Mike Otsuka 09.02.05 at 12:27 pm

Wasn’t the alcohol to see in the millenium purchased in 1999, thereby undermining the author’s case against choosing that year as the baseline?


Matt Daws 09.02.05 at 1:27 pm

Mike and charlie,

I think you miss the point. The author isn’t trying to argue that the report (I think it’s this report with a summary from Mintel) is wrong. She is simply pointing out that it uses statistics very badly.

Were the report well-written, it would be possible to check and (presumably) refute all of her suggestions. On the 1999 base-line issue, the report *should* have given alcohol sales for every year since, say, 1995, so we could tell what the trend is. That said, I agree that her specific point doesn’t actually work: if there was a peak in 1999 because of the millenium, and then there was still an increase of 5% to 2004, then that means we’d expect an even bigger increase between 1998 and 2004, say.

Note that the report fails to even *mentions* the figure for 1999, instead giving the “5%” claim and the figure of “over 8 billion litres” of “alcohol” in 2004 (so quite how they compare wine with beer with spirits is unclear).

Similarly, ignoring per capita effects is clearly dodgy. Her last point about the silly definition of “binge” is also fair.

So I don’t think she was trying to refute the report. She was simply pointing out that the report was pretty useless! One certainly gets the impression that the “5%” figure is vaguely massaged, and that a more careful analysis would probably get a figue closer to 1-2%, that is, little to worry about.

To make a more serious point, this does show the lack of any decent stats in news reporting. Mintel issues a press release which lacks any important information on methodology, and lacks any supporting data. News organisations then pick up the press-release, vaguely re-word it, and publish it. If we’re lucky, a broadsheet newspaper will actually buy the report and print some of the data, but that’s only if we’re lucky. It’s interesting to look at the contents summary of the full Mintel report which does, for example, carry “per capita” data. I would bet that the full report would deal with pretty much all of Valerie Pegg’s issues. The problem is that the media report doesn’t.


Charlie B. 09.02.05 at 5:23 pm

Thanks Matt. I agree. The same lousy report about the sales increase report is repeated in one news source after another. I didn’t find it too hard to get some of the missing data myself (ONS mainly) and indeed, it does not make for a neat summary – not least because the statistics vary very considerably between regions (if the reality does I can’t of course say…). The problematical definition of “binge drinking” is remarked in all the serious literature – it is basically a less emotionally-laden phrase for “get drunk” – as opposed to drink without getting drunk. What they mean when they say “binge drinking” is getting worse is that “drunkenness” is increasing – which it probably is.


Matt Daws 09.03.05 at 4:15 am


What they mean when they say “binge drinking” is getting worse is that “drunkenness” is increasing – which it probably is.

Yeah, I agree I think. The problem comes when the media or government or moralisers in general complain about the rise in “binge drinking”. I then look at the figures, work out that my half bottle of wine over dinner counted as a “binge drinking session” and then generally ignore the message. I’m fairly sure a lot of people do this as well: it’s the same old story of kids being told “all drugs are evil” and then them working out that actually not all drugs are intrinsically bad, and some, used in moderation, are quite fun. If the definition of a binge drinking session was more realistic, then I do think people might listen to the message more. Drunkenness is getting worse I guess, but it’s not people drinking a few pints!


William 09.03.05 at 7:21 am

I consider the entire concept of “binge drinking” as its defined to be hopelessly useless to anyone other than college deans who want to justify clampdowns on alcohol. Much of the work done to operationalize the current term “binge drinking” is done by Henry Wechsler of the Harvard School of Public Health College Alcohol Study. A link to their published articles follows:


Weschler defines binge drinking as consuming 4 drinks for women and 5 drinks for men in one episode of drinking. AFAIK he does not take into account the duration of drinking, or estimate BAC. A man who downs 5 shots of vodka back to back is treated the same as a woman who drinks 4 beers at an all-day Superbowl party. It may be that as a long-term health issue there isn’t much of a difference . . . but Weschler isn’t really dealing with long-term effects of alcohol on the human body.

Instead, he’s concerned about the behavior of binge drinkers and the impact binge drinkers have on other folks. It might be *possible* that simply drinking 4 drinks in 24 hours will permanantly rearrange the personality of women, but its more likely that any differences in behavior that are related to use of alcohol have to do with, you know, folks being drunk. So why use the 4 and 5 drink cut-offs without reference to BAC or time spent drinking?

They use this standard because there are correlations between someone “binge drinking” over the past 2 weeks and that person causing social harm or being dependent on alcohol. The research, as far as I know, doesn’t actually say that drinking 4 or 5 drinks is harmful. Just that folks who do will be more likely to (at some point, perhaps on *another* occasion when they drink 7 or 8 drinks) do something obnoxious. But the bad behavior isn’t correlated to any specific incidence of “binge drinking”.

Some of Weschler’s stuff is really funny . . . in one survey he includes “unplanned sexual activity” as a alcohol-related problem, along with having a hangover or missing a class, or requiring medical treatment. In another he finds that colleges that have no stores selling alcohol within a mile of campus have lower rates of drinking. And what makes it great is the way his research combines widely different things, like unplanned sex and getting your stomach pumped, or bars offering $US10 all-you-can-drink specials with bars offering US$2 pints.

Don’t laugh too much, though. Weschler thinks the only way to prevent this terrible scourge of “binge drinking” is for colleges, communities, and states to create more restrictive laws and regulations against drinking. So all the hijinks with correlations and laundry lists of possible negative outcomes and the overly-broad definition of “binge drinking” that doesn’t measure actual drunkeness server a darker, puritanical purpose.

At the end of the day, however, a 180 pound adult male who drinks 5 beers in 2 hours will have a BAC% of 0.06. Not really safe to get behind a wheel of a car, but also not intoxicated enough to be a big threat to society.

Comments on this entry are closed.