An extensive review of alcohol policy published in the Lancet concluded that: ‘Making alcohol more expensive and less available, and banning alcohol advertising, are highly cost effective strategies to reduce harm’. Unfortunately, calls to ban or restrict alcohol advertising (such as calls to increase price) have been rejected by governments in most countries. Thus, as Meier states, there is a need to provide evidence of the effects of alcohol advertising on young people in order to encourage the government to take action to reduce, or eliminate, the most harmful forms of alcohol promotion (which may, or may not, be ‘advertising’per se) .
Meier identifies a number of gaps in the current evidence base, and suggests a research agenda to address these gaps. These gaps are: evidence of the effects of alcohol advertising on specific population groups; research designs that reflect marketing complexity; and research to clarify the timing of effects. It is true that researchers consistently fail to conduct comprehensive studies that assess the complex web of alcohol marketing. A recent systematic review of longitudinal studies of alcohol advertising and youth alcohol consumption identified 13 studies , but found that the majority assessed only a narrow range of exposures and only two assessed a broad combination of above- and below-the-line marketing exposures [4,5]. However, just as we know that the most effective interventions to change individual drinking behaviour are environmental ones, I would argue that the same applies to changing researchers' behaviours. Each of the issues identified by Meier are important gaps in our current knowledge base, and each should be addressed. However, so doing requires going beyond calling on researchers to engage in this research; we also need to call on government, industry and other players to make changes that will facilitate the conduct of this research.
Governments must be prepared to fund complex, longitudinal studies that extend beyond 3–5-year projects (the standard period of funding in many countries). Although researchers can apply for subsequent grants to continue projects, this absence of secured ongoing funding seriously limits capacity to commit to the type of longitudinal studies needed to answer these important questions. In particular, studies that span 10–20 years (from drinking initiation to adulthood and the establishment of life-time drinking habits) are needed. These long-itudinal studies will enable us to move beyond demonstrating that exposure to alcohol advertising has immediate effects on consumption  and medium-term effects on adolescent alcohol initiation and consumption levels [4,5,7]. Establishing the timing of effects of removing alcohol advertising is more complex, given the ingrained cultural norms of (excessive) drinking in many countries. For example, advertising bans are likely to take generations of time to have a measurable effect at a population level; effects on some subgroups may be more rapid (e.g. as children who grow into adolescence in a world not saturated by pro-drinking messages).
Industry needs to be prepared (or governments need to be willing to require industry) to provide the level of data necessary for researchers to assess effectively the effects of the complex web of alcohol marketing. Even for the most straightforward of studies into, for example, broadcast television advertising, researchers need to commit substantial funds to the purchase of data from commercial agencies  and to the purchase of advertisements for follow-up testing . The industry has such data; making it freely available to researchers would facilitate research that enables us to both ‘reflect marketing complexity’ and determine ‘how different population groups respond to marketing’. The provision of more complete data—such as data on below-the-line expenditure—would inform debate on whether a reduction in alcohol advertising results in changes to consumption by enabling researchers to control for the re-direction of marketing expenditure from one channel to another.
Other, more controversial, issues must be considered if we want to address more fully these gaps in the evidence base—issues that do not have an easy ‘right or wrong’ answer. I will limit this to two examples: journals and ethics committees.
A common argument by governments in refusing to take action on alcohol marketing is the lack of published evidence from their own country. For example, in their response to the House of Commons Health Select Committee report  the UK Government discounted much of the cited evidence with parenthetical comments such as ‘this related to mainly US based studies’ (p. 26) . Of the 13 studies reported in the 2009 systematic review, 10 were conducted in the United States, and one each in Belgium, Germany and New Zealand. This creates a tension between what governments and journal editors see as ‘new’ research, with researchers perceiving that it is difficult to have local studies accepted in journals on the basis that the evidence is redundant (given the existing literature).
To untangle the impact of different forms of alcohol marketing we would need to be able to collect real-time data on both exposure to marketing messages and drinking attitudes and behaviours. Recent technological developments make this technically feasible [e.g. short message service (SMS), social networking sites], as do more traditional and resource-intensive methods (e.g. participant observation, diaries). However, researchers report that it is increasingly difficult to obtain ethics approval for complex studies, particularly those with young people, due to the legitimate concerns of ethics committees with respondent burden and the invasive nature of the research methodologies that would be needed to fill the ‘gaps’ identified.
In summary, I think Meier has conducted an excellent job of summarizing gaps in the evidence base, and setting out a research agenda. However, in order to achieve this agenda we need more than just willing researchers; we need to work with those who provide the infrastructure and set the parameters within which our research is conducted.