Drinking and off licences

From the BBC:

Alcohol abuse linked to number of licensed premises

Scottish neighbourhoods with the most licensed premises have the highest rates of alcohol-related illness and deaths, according to a new study.

Death rates are more than double in areas with the most bars and off-licences, according to Alcohol Focus Scotland.

Sigh. We've been over this before. It is sometimes said that people shouldn't be allowed to become politicians until they've spent a few years in business. It is tempting to say the same of people in 'public health'.

Look, areas of high demand will have more supply. That is as true of alcohol as it is of any other product. If supply was sufficient to create demand then alcohol retailers would open more and more off licences everywhere until they were evenly distributed across the country on a per capita basis. They don't do that because they follow demand instead. Off licences are more densely concentrated in certain areas for the same reason that bookmakers, coffee shops and Polish food shops are more concentrated in certain areas, ie. because that's where the customers are.

Just as the alcohol retail industry cannot expect to increase aggregate demand by opening a new off licence, temperance activists should not expect to reduce aggregate demand by closing them down. Alas they do expect that because they're simple folk who believe, like every temperance crusader before them, that demand for alcohol is somehow created by the sneaky alcohol industry. Hence their quixotic battle against availability and advertising that flies in the face of everything we know about the workings of markets.

Today's study is a very similar to a study produced by Alcohol Concern three years ago that was widely, and rightly, derided. I wrote at the time...

This is such a blatant conflation of correlation and causation that even Ben Goldacre—who never criticises 'public health' bad science and sometimes defends it—emerged to poke fun at it.

A red-faced Don Shenker knew exactly what he was talking about and replied...

To which Goldacre rightly responded...

As Straight Statistics points out, Alcohol Concern quite explicitly did claim causality:

Under the heading Methodological Qualifications, the new report states: “This study did not set out to establish cause and effect.” Yet the previous page asserts that nearly 10 per cent of all alcohol specific hospital admissions in England, excluding London, are directly attributable to off-licence density, “meaning availability rather than any other external factor is the cause of one in ten of such harms”.

On the last page of the new study, there is a brief acknowledgement that "we cannot conclude that the relationship is causal". Such a tokenistic note of caution was not reflected in the lead author's comments to the press, which strongly implied causation and emphasised the policy implications that the study was obviously designed to push:

"The strong relationship we found between alcohol outlets and related health outcomes leads us to suggest that reducing outlet numbers, particularly in the highest availability neighbourhoods, could have health benefits for the Scottish population."

Meanwhile, the head of Alcohol Focus Scotland (a state-funded sock puppet pressure group) told the BBC...

"If we want fewer people to end up in hospital or lose their lives because of alcohol, then we have to be concerned about the high number of alcohol outlets in our neighbourhoods.

"Licensing boards have a key role to play in regulating the overall number of licensed premises and their decisions should be informed by studies such as this."

It would be unwise to base any decision on this study. At best, it appears to state the bleeding obvious, but it is actually worse than that. It doesn't even make sense on its own terms. The obvious way to conduct a study on outlet density is to look at how many people there are for every outlet. This study doesn't do that. Instead it looks at the number of outlets in an 800 metre radius with no regard for population size.

Using this methodology, densely populated cities like Glasgow and Edinburgh are shown to have denser concentrations of off licences. This is pretty much inevitable. It would not be surprising if there were more outlets in areas where there are a disproportionately large number of (heavy) drinkers, but this study doesn't even show that. It just shows that there are more outlets in areas where there are more people.

But if you look at it on the basis of people per off licence, the picture is very different. Glasgow has an off licence for every 1,158 people and Edinburgh has an off licence for every 1,059 people. In Orkney and the Shetlands, however, the figures are 559 and 533 respectively (I have derived these figures from the data in the study—the authors do not cite them directly).

It may well be the case that large, industrial cities have higher levels of alcohol-related disease than small, rural communities, but there are plenty of factors you could attribute that to before you resort to blaming large populations for needing a large number of shops.