Put simply, the activist/academics who dominate the discussion of alcohol policy in the most prominent 'public health' groups have decided that tobacco-style bans and taxes are the way to deal with alcohol-related health problems and no amount of evidence is going to deter them.
Take Peter Anderson, for instance, a temperance academic who wears many hats, including that of project leader at the EU-funded ALICE-RAP project...
The fact that Anderson et al. (2012) present tax increases and bans on alcohol advertising as two of the most effective measures does not only contradict the conclusions that Anderson and Baumberg (2006) had drawn a few years earlier: ‘Looking across Europe more thoroughly there is no apparent correlation between the revenue from alcohol-specific taxes and per capita consumption’, but also those of Babor et al. (2003), who had rated advertising bans as rather ineffective. It is interesting to observe here, regarding the relationship between alcohol taxes and per capita consumption, that real alcohol prices in Austria dropped by around 50% over the past four decades, while per capita alcohol consumption did not increase; rather, it fell by 20%.
On the selective omission of facts:
An illustration of how it is possible to present statistical findings in a way that best supports the policy conclusion desired is that of Edwards et al. (1994) using data from an article by Hurst (1973), who had reanalysed data stemming from the influential Grand Rapids Study. In a well-planned roadside study, Borkenstein, Crowther, Shumate, Ziel, and Zylman (1964) had shown that drivers with moderate alcohol levels around 0.3 g/l blood alcohol concentration (BAC) had a significantly lower risk of committing an accident than completely sober drivers, an effect referred to as the ‘Grand Rapids dip’. When Hurst reanalysed these data separately for groups with similar drinking habits, he could show that the Grand Rapids dip was an artefact due to a Simpson paradox (where heterogeneous samples are analysed, the overall effects can be the opposite of the effects in all homogenous subsamples), leading to erroneous interpretations. In effect, the risk of committing an accident rises almost linearly with the amount of alcohol consumed, which is a positive result for those who support Zero-BAC-limits for drivers. However, Hurst also found that the risk of committing an accident in a completely sober state is four-fold for abstainers compared to daily drinkers, with daily drinkers reaching the risk level of abstainers after they exceed a BAC level of 1.0 g/l, which is twice the legal limit for drivers in most European countries. This finding about the ‘dangerous abstainers’ is hard to explain and is certainly an unwanted result for alcohol control activists. Edwards et al. (1994) chose to document the pleasant effect and hide the unwelcome effect by standardising the risk at Zero BAC level to 1 for every group. They could thus demonstrate the linear increase of risk with increasing BAC-levels, but hide the differences at Zero BAC levels: a brilliant trick for an advocate, but hardly excusable for a scientist.
On the deliberately misleading cost-of-alcohol studies...
A popular argument raised to support a strict alcohol control policy is the high social cost of alcohol use for society. Anderson and Baumberg (2006) stated the loss to be 1.3% of GDP. It can easily be shown that the logic of such cost calculations is systematically flawed, since the costs are, to a large extent, not paid for by the non-involved others or from the public purse. Alcohol-induced premature death undoubtedly causes high intangible costs for the persons drinking excessively (internal costs) and their near environment, but dead persons – similarly to children not born due to contraception or young immigrants not allowed to enter the country due to strict immigration laws – neither produce nor consume any goods... These figures are a perfect advocacy tool for alcohol control activists, and they help preventionists, therapists and researchers justify funding for their activities.
On presenting correlation as causation...
An illustrative example by Anderson and Baumberg (2006, p. 147) is the way in which the relationship between heavy drinking and suicide is presented in the following terms: ‘Heavy drinking is a major risk factor for suicide and suicidal behaviour among both young people and adults’. Knowing that heavy alcohol consumption is commonly a consequence of psychiatric problems such as depression, an adequate causal interpretation would be that persons with a tendency to heavy alcohol use have an increased suicide risk, and that it is likely that drinking excessively increases the suicide risk even further. There is also little doubt that some suicides are a consequence of depression caused by drinking alcohol in individuals who otherwise would not have killed themselves, but what fraction of suicides is due to the primary disease and what fraction is due to alcohol problems cannot be answered properly if only the association is recorded. A certain percentage of depressed persons who eventually commit suicide and use alcohol to self-medicate their depression may even have lived longer since this form of self-medication made them postpone their suicide. It cannot be denied that excessive alcohol may cause depression and suicide, but to interpret and present associations in the usual way, without considering widely known facts and cultural contextual factors and without presenting different explanations is misleading.
Finally, some wise words about the bluff of 'evidence-based policy' which really just means 'evidence that the policy we propose will do the thing we want it to'...
Examples from key publications supporting evidence-based alcohol policy show that much of it should more realistically be labelled ‘policy-based evidence’... By taking a stand against certain value-driven positions which unjustly claim a degree of proof that is not factually justified, the present article argues for greater modesty and not at all for moving back to well before the age of enlightenment, when relying uncritically on authority, experience and intuition was the name of the game.
To demand that political decision-makers should implement evidence-based policies is particularly misleading, since it suggests that policy (what should be done) can be derived from empirical research (what is), which is a popular misconception, labelled a ‘naturalistic fallacy’ by Moore (1960). All empirical claims raised in discussions about policy should correspond as closely as possible to reality, but practical decisions are always rooted in ethical–political value decisions related to the image of humanity, society and the world (Uhl, 2007). This aspect is camouflaged by the term ‘evidence-based policy’. The question of how paternalistic an administration should be, or how far the views of the majority of the population in a democratic society should prevail (that is how far it is legitimate to coerce individuals in a society to behave in a healthy way) is an ethical–political decision to be decided very carefully and is not simply factual–scientific.
Well said, and that's without even mentioning the cesspool of minimum pricing 'research'.
No comments:
Post a Comment