You can't kill a bad stat

The New Zealand Drug Foundation should know better. And yet, here we are.

The NZDF put up a Bingo card for those who support tougher regulations around alcohol with the catch-phrases they expect to show up in the debates around the Alcohol Reform Bill. The last page of it has a few problems. Alcohol Reform Bill Bingo  Among the facts, as NZDF sees things:

It's a small number who are ruining it
Since when is 785,000 a small number? That's how many "uninhibited binge drinkers" there are in New Zealand. A further 622,000 people are "constrained binge drinkers". Together that's 1.3 million people, about a third of all drinkers in New Zealand who are not drinking responsibly.

nzdrug.org/smallnumber

Ok. First off, there are 4.4 million people in New Zealand. They're claiming 1.3 million is a third of all drinkers. So they're claiming that we have 3.9 million drinkers in the country. At the 2001 census, there were about 848,000 people aged under 15. So the New Zealand Drug Foundation is claiming that every single person aged over 15, along with 41% of all of those aged 0-14, are drinkers. That seems off.

So I hit their provided link. It's a redirect that goes to an article by Sally Casswell. That article provides results from a survey where they asked a representative sample of people a bunch of questions, including whether they know anybody who they consider to be a heavy drinker. 29% of the sample reported knowing at least one heavy drinker. Now it would be a really really big mistake to extrapolate from the number of heavy drinkers known by people in the sample to a population estimate of the number of heavy drinkers because respondents could easily be referring separately to the same heavy drinker. And, Casswell doesn't do that. In fact, there's absolutely nothing at the link to back up NZDF's claim here. I think this is likely a redirect error: the study backs up the number in the next factoid, where they say "about a third of New Zealanders have at least one heavy drinker in their lives" (note that 29% is almost equidistant from a quarter and a third for rounding purposes. Since a third is 33.333% and a quarter is 25%, a third is just a pinch worse for rounding purposes than a quarter, if you want to be accurate and you like fractions. But you always round up if you want the scarier fraction).

But, because I work in this literature, I know where their number comes from. It's from an old ALAC report, cited here. What counts as an "uninhibited binge drinker"? It's awfully hard to tell from the report. They define it as:

"adults, 18+, who are less concerned with the effects of their drinking and less inhibited than Constrained Binge Drinkers. They drink regularly (often every day) and binge, mainly to unwind, and for the "buzz" and enjoyment."

By my best guess, adults who drink regularly, who sometimes binge, count as "uninhibited" if they don't seem sufficiently contrite about it in the survey questions. At page 25, they say that the main difference between constrained and unconstrained bingers are "demographic and attitudinal characteristics", with the unconstrained exhibiting a more reckless disregard for downside costs of heavy drinking. What counts as binge drinking?

"where an adult reports they have consumed the equivalent of seven (7) or more glasses of alcohol during a single drinking session."

I'll assume that they're there meaning standard drinks. Seven standard drinks, 70 grams of alcohol, is the equivalent of most of a bottle of standard strength wine (a bottle usually has 8-9 standard drinks). A half-litre of Emerson's JP is 3.27 standard drinks. So less than two pints of the JP makes you a binge drinker. Since I have about 0.6-1.3 standard drinks per typical day and perhaps once every month or two have something more like a couple pints of JP (over 3+ hours), and because I really can't see any negative consequences from my drinking pattern, I might count as an uninhibited binge drinker - it depends how often you have to binge to fall into the category. Figure 4 of that report shows that 56% of "uninhibited binge drinkers" had 3 or fewer drinks on their last drinking occasion; 25% of that category consumed 7 or more drinks on their last drinking occasion. So it's not implausible that a lot of people fall into the category. But does the category really tell us a ton about anything useful?

At page 75, the ALAC report compares uninhibited binge drinkers who drink every two to three days and who drank 7+ drinks on the last occasion with those who don't. The heavier drinking members of the uninhibited drinking cohort are more likely to be full time salary or wage earners, have higher income, and are more likely to drink with friends than to drink alone. Are any of those bad things?

But, the worst stat in NZ Drug's "Facts"? This one.

The status quo is fine.
The status quo means 1000 deaths due to alcohol every year. It means 785,000 binge drinkers. It means $72,000,000 in costs for Police, Corrections and health expenditure. It means $4.8 billion in costs to the taxpayer. The status quo means more harm.

nzdrug.org/alcoholreform

I don't know about the 1000 deaths; I hope that it's net of the lives saved by moderate drinking, but I doubt it. The 785,000 binge drinking figure seems a bit off, as noted above. The $72m figure is included in the $4.8 billion figure. And, the $4.8 billion is not a cost to the taxpayer. Again, here's BERL's cost tally:

• $1.52 billion in intangible costs of loss of life (the vast majority of which is reduced life expectancy among very heavy drinkers, with no accounting for increased life expectancy for moderate drinkers);
• $1.48 billion in labour costs of lost productivity (the vast majority of which are the forgone production of the prematurely deceased, and consequently is double-counted with the $1.52 billion above since the Value of a Statistical Life is inclusive of lost production);
• $699 million in "drug production costs", including every cent of excise paid to the government by heavy drinkers;
• $562 million in crime costs, with a ridiculously low threshold for determining what is an "alcohol-caused crime";
• $290 million in health care costs, under the assumption that there are no health benefits from moderate drinking and by zeroing-out the effects of drinking for those disorders where even fairly heavy alcohol consumption reduces the costs of that disorder;
• $200 million in road crash costs, consisting both of those very real costs that drink drivers impose on others and the costs that drink drivers impose on themselves by wrecking their cars;
• $42 million in lost quality of life, again mostly falling on heavy drinkers themselves.

When I'd gone through the figures it looked like maybe $967 million of BERL's $4.8 billion could count as an external cost by more normal method.

It's hard to kill a bad stat. Darned things keep popping up whenever somebody finds them to be convenient.