Alcohol advertising, again

The usual lot want to ban alcohol advertising in sport, to protect kids.

I'll excerpt from my submission to the Ministerial Forum on Alcohol Advertising and Sponsorship of a couple of years ago; the whole thing is here. They were focused on evidence since 2010 because they took the Law Commission's prior report as starting point, so that explains the 'since 2010' bits quoted below.

Nelson (2010) examines whether alcohol advertising bans affected alcohol consumption in a panel of 17 OECD countries over the years 1975-2000. His modelling is careful: he begins by controlling for the underlying factors giving rise to country-level restrictions on alcohol advertising before estimating the effects of those restrictions controlling for the underlying factors that cause advertising restrictions and controlling for the overall stringency of alcohol regulation. He finds that bans on alcohol advertising have no effect on total alcohol consumption. It is rather important to correct, as he does, for the underlying factors that predict countries’ adoption of advertising restrictions: Gallet and Andres (2011) demonstrate that countries with a greater proportion of youths, with greater life expectancy, with higher income, and with greater Muslim populations are more likely to adopt advertising restrictions. If countries that are generally healthier, as demonstrated by life expectancy, are more likely to adopt restrictions on alcohol advertising, correlations between health outcomes and advertising bans could easily be spurious. Nelson also notes that while, from the 1980s onward, most countries liberalised their restrictions on alcohol advertising the period since the 1980s has also seen reasonable declines in total alcohol consumption. Similarly, the period of liberalisation in New Zealand, from 1989 onwards, also corresponds with a period of substantial decline in per capita consumption.

Some evidence since 2010 has suggested that increased exposure to alcohol advertising is associated with increased risk of youth drinking.

Morgenstern et al (2011) finds that German students who were better able to recall the names of alcohol brands from popular advertisements had a higher risk of drinking and of binge drinking. However, they are unable to demonstrate a temporal effect: that is to say, they cannot demonstrate whether advertising recognition preceded drinking, or came subsequent to drinking. Consequently, they cannot tell us whether students most able to recall alcohol brands are 2.3 times as likely to engage in binge drinking as those least able to recall alcohol brands, or whether students who binge drink are 2.3 times as likely to remember alcohol brands. As they also measure exposure to non-alcohol advertising, it would have been interesting to see if, for example, students who take on more duties at home were better able to recall the detergent brand, whether students with mobile phones were better able to remember the T-Mobile brand, and whether students who go hiking were more able to recall the trekking-clothing ads.

Jones et al (2011) argue that youths who indicated having seen alcohol advertisements were more likely to initiate drinking and to have consumed alcohol in the past four weeks. However, their adjusted odds ratios frequently fail to achieve statistical significance. One significant effect was that having seen TV advertisements for alcohol halved the risk of having consumed alcohol in the past 12 months. But none of the odds ratios there reported should be taken particularly seriously: the substantial reduction in odds ratios after correcting for a very small number of covariates, coupled with the substantial decline in statistical significance after such correction, suggests that uncontrolled confounding could easily explain the remaining variation. Table 3 presents correlations between advertising and alcohol initiation across 32 different advertising / gender / age cohorts. They find three associations are significant at the 5% level: barely more than we would expect by sheer chance with so many separate regressions. Results in Tables 4 and 5 are rather similar: where adjusted odds ratios are significant, three show that alcohol advertising reduces the likelihood of regular or recent alcohol consumption, seven show an increased risk, and 42 show no significant relationship whatsoever. If anything, we should take this study as providing reasonably strong evidence of the absence of a relationship between having seen alcohol advertising and the initiation of alcohol consumption. Further, while some of the aggregates presented in Table 2 provide statistically significant associations between some alcohol advertising exposure and drinking, they are unable to rule out reverse causality where those who are more interested in alcohol to begin with are more likely to pay attention to and to remember alcohol advertisements. Look past their discussion of their results to what they’ve actually shown: there is scant evidence on which to hang their conclusions.

Bryden et al (2012) conduct a meta-analysis on effects of alcohol advertising. They find little evidence of harmful effects of alcohol advertising. Importantly, those studies categorised as methodologically weak were more likely to find harmful effects of alcohol advertising. See discussion at 3.2.2, p. 355.

The recent literature here surveyed suggests that even the most stringent of alcohol advertising regulation, full bans, has no effect on consumption. Individual level exposure to advertising may have small effects on consumption, but those studies showing effects do not successfully disentangle brand recognition among drinkers from effects on drinking intentions among those exposed to branded advertising.

Bryden, A., B. Roberts et al. 2012. “A systematic review of the influence on alcohol use of community level availability and marketing of alcohol.” Health & Place 18: 349-57.

Gallet, C. and A. Andres. 2011. “International evidence on the determinants of alcohol advertising restrictions.” Applied Economics Letters. 18:14, 1359-1362.

Jones, S. and C. Magee. 2011. “Exposure to alcohol advertising and alcohol consumption among  Australian adolescents.” Alcohol and Alcoholism 46:5, 630-7.

Morgenstern, M. et al. 2011. “Exposure to alcohol advertising and teen drinking.” Preventative Medicine 52: 146-151. Nelson, Jon P. 2010. “Alcohol advertising bans, consumption and control policies in seventeen OECD countries, 1975-2000.” Applied Economics 42:7, 803-23.

I couldn't see much reason to expect that advertising bans in sport would do much to reduce risky drinking. I also didn't like suggestions of bans on sponsored events, teams and venues.

Bans on alcohol industry sponsorship of sporting or other events need be based on strong evidence of net harms resulting from such sponsorship. If youths or other at-risk groups were substantially more likely to engage in harmful binge drinking instances because their favourite team or concert were sponsored by particular brands, and if those harms greatly outweighed the demonstrable benefits of sponsorship to the sponsored organisations and events, then we could have a reasonable case for restrictions or bans. That evidence, however, does not exist. While there is ample evidence of alcohol sponsorship of events and sports teams, evidence of consequent harms is lacking.

Further, it is plainly evident that attendees at sponsored events often greatly benefit from that sponsorship. The Rugby Sevens are typically taken as evidence of the awful consequences of alcohol sponsorship of sporting events. And while it’s true that the Sevens are typically associated with alcohol consumption, that hardly makes the case for a ban. Survey data from the HPA (2013) demonstrates not only broad awareness of alcohol sponsorship of the Sevens, but that alcohol’s presence at the Sevens is a critical part of the fan experience. 82% of attendees surveyed at the Sevens agreed or strongly agreed that drinking alcohol made the event more entertaining; 93% agreed or strongly agreed that they attend the Sevens because of the atmosphere; 77% agreed or strongly agreed that “drinking alcohol at this event is ‘just what you do’”. Fans attending the Sevens really seem strongly to enjoy the particular atmosphere present at the Sevens. It’s also worth noting that that same survey demonstrated that alcoholic sponsor messaging was less prominent there than at other surveyed events, like the Heineken Open or the International T20, where alcohol sponsor messaging was more prominent but where the event’s culture was rather more sober.

It is particularly worrying that the Law Commission’s report called ultimately for a ban on alcohol sponsorship (19.182), but provided only one piece of evidence suggestive of potential harms from sponsorship: that survey respondents who received free or discounted alcohol as part of their team’s sponsorship arrangement felt they should drink their sponsor’s product (19.27). On the basis of that evidence, they wished to ban all alcohol industry sponsorship of events and sports teams. Presumably they found the harms self-evident.

Meanwhile, over at the Sandpit, I have a look at another paper showing some of the social benefits of social drinking. Check it out.

UPDATE: I'd forgotten to link the Cochrane Review, that said there's not enough evidence to show anything on effects of advertising bans as yet. Curiously, none of the Science Media Centre's chosen experts seemed to have heard of that one, 'cause they didn't cite it.