Outcome of academic seminar to review recent research on food promotion and children

• Prof Nicholas Mackintosh, Professor Emeritus and formally Head of Developmental Psychology, Cambridge University
• Dr Charles Crook, Reader in Psychology at Loughborough University
• Prof Uta Frith, Professor of Cognitive Development at University College London
• Prof Marvin E. Goldberg, Irving and Irene Bard Professor of Marketing, and Chair, Department of Marketing, Pennsylvania State University
• Prof Stephen Platt, Director, Research Unit in Health, Behaviour and Change, Edinburgh University
• Prof David Skuse, Professor of Behavioural and Brain Sciences at the Institute of Child Health

Authors/Contributors to the reports

• Prof Gerard Hastings, Centre for Social Marketing, University of Strathclyde
• Dr Mike Rayner, University of Oxford
• Dr Martine Stead, Centre for Social Marketing, University of Strathclyde
• Dr Brian Young, School of Psychology, University of Exeter

After presenting and discussing their work, the authors and contributors left the meeting.

Further discussion followed, and the following conclusions were drawn.

The seminar was organised to consider two recent reviews on food promotion and children ('Hastings' and 'Young').

In particular it sought to understand the extent of agreement between the reviews and, where there was disagreement, to identify the reasons why this occurred.

It was agreed that it is important to separate consideration of the evidence for the effect food promotion has on children's diets from how this might be used to formulate future Government policy.

The purpose of the seminar was not to consider policy options, but to consider the relative merits of the two reviews.

It was noted that social science by its very nature does not claim to provide definitive proof, but seeks to establish the relative strength of the evidence.

It was noted that the two reviews are fundamentally different in approach, in two areas:

Methodology

Hastings employed a systematic review methodology, seeking to ensure a comprehensive, transparent and as far as possible objective examination of the evidence, and also a narrative review which matched well with the evidence.

In contrast, Young provided only a narrative review based on selected literature.

Theoretical context

Hastings balances evidence from observational and experimental studies, taking account of the quality and relevance of individual studies.

Young seeks to place his research within a theoretical framework that takes account of the broader context in which advertising operates.

He highlights other cultural influences that bear on children's dietary behaviour, and stresses the significance of children's cognitive development, as it relates to their understanding of advertising.

Despite these differences of approach, there is a degree of overlap between the reviews and a number of areas where the conclusions reached are generally compatible.

However, the reviews disagree in one key area: whether the evidence establishes a link between promotional activity and children's food knowledge, preferences and behaviours, and the nature of that link.

An apparent reason for the difference of opinion is that in reaching conclusions in this area, Hastings took account of more studies than Young (of the 54 studies on which Hastings' conclusions are based, 21 are referenced in either Young's 1996 or 2003 reports).

Whilst recognising that the Hastings review was undertaken by a research team, as opposed to the Young report being a 'solo' effort, clearly it is preferable to take account of all the available evidence.

It was noted that there tended to be agreement between Hastings and Young as to the best quality research, but that there was a divergence of views as to the significance of the conclusions of these studies.

Essentially, this focused on the fact that Young more frequently questioned the generalizability of research.

His aim of considering the effect of advertising in its social context led him to conclude that the artificial nature of experimental studies disqualified their findings from consideration, because they could not be applied to the general population of children.

He also criticised observational studies that simply report correlations between two variables, because they could not establish a causal relationship between them.

It was also noted that the relationship between Young's 1996 report and the 2003 review was unclear.

The 2003 work did not appear to take account of studies indicating a link between advertising and children's behaviour, that had been cited in 1996.

This might explain why the 2003 research appears to arrive at different conclusions from the 1996 report.

While accepting that there was a need to take care in considering the results of experimental studies, it was agreed that the short-term influences measured by such studies were analogous to real life situations.

Indeed, it was concluded that both experimental and observational studies had a role to play.

It was noted that Hastings had provided convergent evidence from both types of studies, strengthening the case that advertising exerted significant influences on children's food preferences and selections.

Furthermore, advertising can work both directly and indirectly, and any direct effect was likely to be amplified, for example by the influence of peers.

The meeting noted that Young appeared content to accept the conclusions of experimental studies indicating a brand-switching effect, yet reluctant to accept studies of a similar design that indicated an effect on category switching.

The meeting accepted the argument made by Hastings that the evidence indicated effects at both category and brand level.

The meeting considered that Young's view that advertising plays a comparatively minor role in influencing children as compared to, for example, parents and school was not supported by evidence.

The studies cited in support of this view relied on self-report.

Such studies, in which respondents purport to know the causes of their or others' behaviour, are an unreliable method for assessing causal influences.

It was noted that practical constraints and the complexity of the issue militated against seeking a definitive experiment to test a direct causal effect of food promotion on diet and obesity.

It was therefore unrealistic to expect the studies examined in the reviews to be capable of providing a definitive answer to this question.

However, the systematic review provided by Hastings was seen to be an appropriate response to this problem, and participants agreed that his review assessed the available evidence in a consistent way and that the conclusions were both balanced and valid.

In contrast, participants noted that Young had considered fewer studies and that there were inconsistencies in the way that he had assessed the results of these studies.

As such, there were concerns expressed as to whether the conclusions reached by Young could be fully justified.

Participants also agreed that if further research were to be undertaken in this area it would be helpful to consider an alternative methodological approach, which took account of the social, economic and cultural context within which advertising operates.

That said, it was not felt that further research was necessarily required as, on the balance of evidence, the Hastings review had provided sufficient evidence to indicate a causal link between promotional activity and children's food knowledge, preferences and behaviours.