Friday, 8 May 2015

An open peer-review of the report "Understanding behaviour in the Australian and New Zealand night-time economies"

One of the areas where my research has gained a lot of traction in recent years has been in the area of alcohol policy (in particular, alcohol outlet density). My research (with Bill Cochrane and others) was cited by the Law Commission in their review of the Sale of Liquor Act, and our more recent research has been used by local councils in the North Island in developing Local Alcohol Policies. Our research consistently shows that bars and night clubs are significantly associated with violence, property damage, and police activity more generally (see here as well, gated).

So, it was with some interest that I read about a new anthropological study on behaviour in the night-time economies (PDF), by Anne Fox (an anthropologist and consultant from the UK). This study was pointed out to me by Peter Miller (Deakin University), who is a collaborator on a research project (along with Matt Roskruge) looking at how well (or not) people can estimate their own breath-alcohol content (more on that in a later post). Peter has addressed the study in two pieces online (see here and here), but I thought I would take a slightly different angle.

One of the problems with the Fox report is that it hasn't undergone a thorough peer review. I thought I would fill that gap here. I have a fair amount of experience in peer review - I have been a reviewer for over twenty journals over the last several years, and not all in economics (they include Australian and New Zealand Journal of Public Health; PLoS ONE; Psychology, Health and Medicine; and Sociological Methods and Research). And of course I have plenty of experience with alcohol research. So here goes.


The report "Understanding behaviour in the Australian and New Zealand night-time economies: An anthropological study" looks at an important issue - the problem of drunken behaviour by (predominantly young) drinkers in the night-time economy, and how to address the problem. As signalled in the title, the report takes an anthropological approach and much of it is quite fascinating to read and consider. The author uses a variety of methods including participant observation, interviews, and focus groups, and concludes that "it is the wider culture that determines the drinking behaviour, not the drinking. You can't change a culture by simply changing drinking" (p.95). The report finishes with 25 recommendations on how to "tackle the true, underlying cultural causes of violence and anti-social behaviour" (p.95).

I see several major issues with the report, and a number of more minor issues. The major issues could be addressed by the author and the report would still provide a useful narrative on the night-time economy in Australia and New Zealand. With two exceptions (that I will address later), the recommendations are fine, and are in most cases complementary to existing public health initiatives. However, it could be argued that cultural change is an extremely long-run solution to a problem that has immediate consequences so more immediate solutions (while not addressing culture) should remain a preferred approach.

The first major issue with the report is that it is unnecessarily constructs straw man arguments to score cheap points against non-existent arguments and support the central conclusion that culture is the dominant driver of behaviour.

This is exemplified by the following quote from page 15 of the report: "Drunkenness and drunken comportment are most often regarded as being directly proportional to the amount of alcohol consumed". The author provides no citation to back up this claim, and if it is indeed "most often regarded" to be the case, then there should be ample potential citations available. I'm unaware of who is saying that drunkenness is "directly proportional to the amount of alcohol consumed", but it isn't in any of the literature I have read.

Also on page 15 the author writes: "First, the very same person on the same dose of alcohol can react in myriad different ways on different occasions and in different settings. This simply would not happen if we were talking about a purely physiological response." Again, I don't know anyone who is claiming that the relationship between alcohol and behaviour is based on a purely physiological response. Moreover, you could re-write those two sentences equally to argue against the reports central theme of culture being the primary driver of behaviour, i.e. "First, the very same person in the culture can react in myriad different ways on different occasions and having consumed different doses of alcohol. This simply would not happen if we were talking about a purely cultural response."

Similarly, on page 43 we find "There is a relationship between alcohol and violence, but it not such a straightforward one as many would have us believe." Again, I'm not sure who is saying the relationship is straightforward, and the assertion is not supported by citations. Again, on page 45 we find "If alcohol alone makes people violent..." Now I'm fairly sure no one says that alcohol alone makes people violent, and again there is no citation to support that anyone is saying this. Moreover, the author writes (again on page 45) that "If alcohol alone makes people violent... We would also expect to find an equal incidence of violence among drinkers in all societies, but we don't". Again, these sentences could be re-written to argue against culture as the sole driver of violence: "If culture alone makes people violent... We would also expect to find an equal incidence of violence among all people in the culture, but we don't".

On the whole, the author really needs to moderate their language. Alcohol is not the only driver of violence, but as far as I am aware no one is asserting that alcohol is the only driver. Similarly, it would be misleading to claim that culture is the only driver of violence.

The second major issue is related to the first. The author makes a large number of (often bold) assertions that are completely unsupported by any evidence. Or at least, the author makes no formal reference to any evidence that supports the assertions that are being made. Some examples of this include:
- On page 13, "At Galahad, we tested this 'placebo effect' in an experiment..." If we are to believe this experiment, then have the results been peer reviewed and published somewhere?
- On page 55, "Violence-reinforcing cultures tend to share the following features..." Who says they share those features? What is the evidence to support this? Following on page 56 by "It is beyond the scope of this report to analyse fully the extent to which each of these features is embedded in Australian and New Zealand society, but we can be find fairly solid evidence for the presence, to a greater or lesser extent, of features 1-9". So the author provides no evidence, and then states that they're not providing any evidence, but want the reader to believe their assertion anyway?

Similarly, there are a number of places where the evidence provided is extremely dated. In a lot of cases the literature has moved on from the 1980s and 1990s, but the author is still citing old literature in support of their assertions. There is one particularly laughable example of this on page 87, where the author writes: "Many alcohol and drug education packages are based on false and outdated assumptions about the nature of peer pressure. Several studies have concluded that the influence of peers is not necessarily a factor in the adoption of unsafe or reckless drinking habits and that, in fact, the very opposite can occur: peers can exert a stabilizing and controlling influence on the drinking behaviour." Then cites a paper published in 1985. It's hard to make a convincing case that others are using outdated assumptions when your supporting citation is 30 years old.

The third major issue is that some parts of the report are simply misinformed. On page 90, the author notes that "Parents need reliable, non-judgmental, unbiased and scientifically accurate information" about a number of issues. There already exists resources for this, produced by the Health Promotion Agency. Similarly, Recommendation 15 states that "Programs should be developed for high-school and first-year university students". Clearly the author is unaware of life skills programmes that already operate in high schools in New Zealand. And again, Recommendation 18: "Educational materials on designing drinking environments should be developed to support hospitality operators in improving their establishments". These resources are already available from the Health Promotion Agency.

On a related note, the author might want to check her data. On page 55, she asserts that Australia and New Zealand rank in the world's top ten countries for income inequality. However, in the OECD report that is cited, New Zealand is 12th out of the 34 OECD countries included, so clearly not in the world's top ten countries.

Fourth, the methods are not described in enough detail. How were respondents selected for interviews or focus groups? How was the data analysed? Was there a framework for the data analysis? Was the data analysed thematically, or using some other method? If thematically, how were the themes selected? How was the literature review conducted? (Given the dated nature of many of the references, this would seem important). Did the field research undergo an institutional review process? If so, where was it approved? If not, then how can the reader be sure that the research was undertaken to a high standard of research ethics?

There are also a couple of more minor issues.
- On page 44, the author tries to convince us that the incidence of violence is so low that alcohol could not be a contributing factor. She writes "If alcohol were a prescribed medication, a side-effect that was reported in only 0.11% of cases would not be considered to have been caused by the drug." This is clearly false - side effects are classified as rare if they affect between 0.01% and 0.1% of patients, but they are still caused by the drug. Moreover, I would consider a 0.11% incidence is quite high. If violent incidents were evenly distributed among the night-time public, then a person who went out three nights a week for a year has about a 1 in 6 chance of being a victim of violence. How is that acceptable?
- On page 54, in comparing rates of general violence and 'alcohol-related' violence between countries, the author states that: "The difference of course is the culture". But of course, that isn't the only difference between countries.

My overall recommendation for this report is that it requires major revision to meet generally accepted academic quality standards.

No comments:

Post a Comment