I appreciate the thoughtful and thought-provoking contributions on our paper [1]. The commentators agree that the paper identifies valid research directions, but Jones [2] rightly points out that research agendas are not shaped only by scientists, and that we need more than ‘willing researchers’. Scientific progress is constrained by the lack of longer-term financial support, industry not being required to provide access to data, and some avenues of research being considered intrusive by ethics committees [2]. I also agree with Casswell [3] that to maximize the impact of new evidence we need to strive for a better understanding of how evidence can be used to inform the policy process, given a powerful pro-marketing lobby. Related to this, Casswell points out that a broad definition of marketing should include industry sponsorship of political events, lobbying and corporate social responsibility activity and that research is needed on the influences of such marketing on policy making. My paper focuses on the effects of alcohol marketing on consumption, but these are clearly important and complementary areas of research. The commentators also raised a number of points that I feel warrant further discussion.

While we have long known about some of the weaknesses in the econometric literature [3], such papers are still being published, reviewed and cited by industry stake-holders and they continue to shape policy in many countries. I would argue that the main reason why this literature continues to be so influential is because there is an evidence need that scientists have not been able to meet: there are currently no accepted alternative methods to quantify population-wide marketing impact. This is why I tend to disagree with Casswell's optimistic assessment that we have now moved beyond the limitations of the econometric approach [3]. There are indeed now good-quality longitudinal studies showing the size and cumulative impacts of certain types of advertising on youth populations. The studies on regional advertising budgets that were also mentioned are again youth-specific and the proportion that was spent on advertising relevant to this population remains unknown. As argued in the original paper, a proper consideration of targeting might result in substantially larger effect sizes, and thus at best such studies may provide information on the probable lower bounds of effect sizes for young people.

Saffer [4] points to the recent neuro-economic model of addiction as an explanatory framework applicable to advertising effects. Like many behavioural economic theories, it does not seem all that different to existing cognitive psychology theories. For example, both neuro-economic and the older dual processing models of decision-making assume a duality of slower and more rational cognitive processes versus faster impulse-driven processes. Which type of process is active is determined by individual factors (e.g. prior learning or cognitive capacity in psychology or ‘addictive stock’ in neuro-economics) and environmental cues, the response to which may be formed by conditioning and social learning. A problem I see with both models is the lack of evidence that these processes are indeed two distinct states between which individuals switch, as apposed to moving along a continuum between ‘rationality and irrationality’ (economics) or ‘analytical and heuristic’ (psychology). When determining the proportions of consumption which should be subject to standard versus modified demand estimation, this becomes of paramount importance.

Neuro-economic theory understands price as purely rational information as opposed to the irrational cues of advertising. According to the theory, heavy consumers are less rational and thus more sensitive to advertising and less sensitive to price. There is no good evidence that this is the case and I remain unconvinced, at least when the focus is on price as experienced in a real-life purchasing situation. In consumer psychology it has long been known that buyers rate the same product as more desirable when more expensive (price heuristics) [5] and that their perceptions and interpretations of a product's price can be variable and determined by situational factors (e.g. purchasing context, the way prices are shown, time pressure) [6,7]. In a recent signal processing experiment consumers accepted low price signals in the form of price match promises as an indicator of actual low price, unless they were in a condition that promoted scepticism and full cognitive elaboration[8]. Consumers may therefore make different—including ‘irrational’—purchasing decisions based on price, even when faced with the same price for the same product in comparable settings.

Engels & Koordeman [9] outline further findings of studies that have become available since I wrote my paper. The authors' findings were that alcohol marketing might be far more powerful when embedded contextually in a manner that makes sense to the viewer, or where attention to content lends empirical support to the applicability of dual processing models in advertising research, and they stress the importance of attending to message targeting and context. In addition to continuing this strand of experimental research, I agree with Saffer [4] that we must seek to find ways of studying susceptibility to micro-environmental factors across populations and in non-experimental settings.

Finally, I am cautiously optimistic that the time-scales involved in detecting first influences of policies such as marketing bans may be shorter than Jones [2] fears. If there is a substantial immediate effect of marketing cues that prompt people to drink beyond what they initially intended to consume, then some population-level consumption reductions might well be observed shortly after policy implementation.

Declaration of interests

P.M. has presented research findings to an industry-sponsored self-regulatory body and has received reimbursement of travel expenses.