BBC makes piss-poor job of JRF alcohol report

If you’ve caught the BBC’s news output this morning on television, radio or teh interwebs, you’ll have seen this:

Parents’ behaviour ‘can influence teen drinking’

Children who see their parents drunk are twice as likely to regularly get drunk themselves, a survey of young teenagers has suggested.

Poor parental supervision also raises the likelihood of teenage drinking, said the Joseph Rowntree Foundation.

On the other hand, if you read the report, then on page 71, you’ll see Table 37, which provides the evidence – in the form of odds ratios (OR) – for ‘Family Influence’, the base for which is a 2009 survey of students who drink.

Q. Thinking about a normal week, how often, if at all, do the following members of your family usually drink alcohol?

Family does not drink – OR 1

At least 1 member drinks every day – OR 2.13

At least 1 member drinks 3-6 days/wk – OR 2.20

At least 1 member drinks 1-2 days/wk – OR 1.65

From this sketchy information, the BBC concludes that:

But the study also looked at what influences excessive teen drinking – and the habits of parents seem to be particularly powerful.

The odds of a teenager getting drunk repeatedly is twice as great if they have seen their parents under the influence, even if only a few times.

Bullshit.

First, the base appears to be all students who drink alcohol and not all students who drink to excess, unless the assumption is that all students who drink alcohol necessarily drink to excess. It also excludes from consideration any students who don’t drink, irrespective of whether anyone in their family home does.

Second, the odd ratios double only where its reported that a family member drinks alcohol either every day or 3-6 days a week. It doesn’t indicate whether any of these parents actually get drunk – some may well only have a single glass of wine or a can of beer when drinking in the home rather than a skin-full – and it also doesn’t tell us anything about the frequency with which those parents who do drink to excess, actually get drunk.

Third, although we’re given odds ratios, were not given any confidence intervals or p values, so we have no way of assessing whether any of these odds ratios are actually statistically significant.

The statistics don’t appear to come close to supporting anything that the BBC has reported in its coverage. JRF’s own evaluation of this data says only that:

The model shows that young people who believe that obtaining alcohol is easy are more likely than those who say that it is not to have had an alcoholic drink in the week preceding the survey. It being ‘very easy’ appears to be the key threshold for ease of accessing alcohol impacting on frequency of consumption; apart from those who say that it is easy to access alcohol, there are no significant access differences once all other factors are accounted for in the model.

So this data isn’t even about parental influence, its relates to the availability of alcohol and its only where respondents have indicated that alcohol is very easy to access than statistically significant results were found once other potentially confounding factors were controlled for.

A little further on, on page 72, the JRF report lists the 10 strongest predictors of excessive drinking in teenagers as follows:

Significant predictors

The ten strongest predictors of excessive drinking are presented below – they are ordered largely in terms of the strength of their relationship towards the behaviour, in this case whether a young person has consumed larger volumes of alcohol in the seven days before the survey.

1. A young person’s attitude towards certain drinking behaviours (acceptability of getting drunk and frequency of consumption).

2. The drinking behaviour of a young person’s friends.

3. The region in which a young person lives.

4. Accessibility: ease of getting alcohol.

5. The frequency with which a young person spends their evenings with friends.

6. A young person’s school year group.

7. The importance of religious belief to a young person.

8. Who a young person was with last time they were drinking alcohol.

9. Where alcohol was sourced the last time a young person was drinking.

10. Their gender.

Parents may conceivably be a factor in items 4 (accessibility), 8 (who they were with when drinking) and 9 (where the alcohol was sourced) but only in cases where it was the teenagers parents who gave them alcohol. Any parental contribution to these factors is, however, not specified in terms of either prevalence or extent. However, neither parental influence nor parental drinking habits actually makes the top ten list of predictive factors for excessive drinking by teenagers.

On an admittedly, very quick and dirty reading of the report, nothing appears to directly support the line taken by the BBC about parental influence nor the prominence that it has been given in its coverage. It actually looks very much like nothing more or less than a bunch of scaremongering bullshit, although I will need to look at the report more closely to see whether and what extent this apparent misreporting of the evidence may be down to bad journalism by the BBC or – worryingly – to the JRF perhaps overplaying its findings to secure as much publicity as possible.