The Perils of Benchmarking Good Behavior Using Self-Report
Are alcohol consumption guidelines undermined by social desirability bias?
Posted January 30, 2016
The UK government has published new guidelines on healthy alcohol consumption and, yes, as might be predicted, they are controversial. According to the new advice, adults -- male or female -- should drink no more than 14 units of alcohol per week. That approximates to around 6 glasses of beer or 7 glasses of wine. Beyond this, the UK Department of Health argue that the risk of alcohol-related death, especially cancer, is significantly increased.
To say that folks were unhappy about the new guidelines would be putting it mildly. One political leader called for an immediate protest: namely, for everyone to visit their local bar and drink something alcoholic. He himself said he would do so at noon that day.
In an interesting turn of phrase, the new guidelines were described as having "come into effect" on Friday 8 January 2016. This almost sounds as though excessive drinking only became unsafe on this particular date. However, what it means is that UK doctors are now required to refer to the new benchmarks when giving advice to their patients.
The reason for the update is new research. Put simply, the previous guidelines had not been updated since 1995 (to put this in perspective, people born in 1995 are now old enough to drink legally in the US). The new guidelines are inferred from the latest data linking various levels of alcohol consumption to eventual illness. According to these projections, drinking 14 units per week is sufficiently associated with poor health as to be declared dangerous.
However, the way such conclusions are drawn may be problematic. While it is fairly straightforward to measure disease outcomes per capita within the population, how exactly are we to measure alcohol consumption? Usually this is done by asking people to report their drinking habits to researchers. In other words, the alcohol consumption data are typically based on self-reports.
The problem with this is that research shows people drastically under-report their alcohol consumption. One prominent British consumer psychologist has pointed out that UK citizens typically under-report their alcohol consumption by as much as 50%. Although the average British person consumes over 20 alcohol units per week, when it comes to telling researchers about it, they admit to only 10.
As noted by UK consumer advocate Christopher Snowdon, this means that the new benchmarks may be particularly misleading. The new limit is based on data showing that people who reported drinking 14 weekly units were liable to suffer adverse health outcomes. However, in reality, these people probably consumed far more than 14 units per week (and quite possibly up to 28 units per week). And, of course, their poor health resulted from the actual amount they drank, not the amount they said they drank .
It is unclear whether the UK Department of Health have found ways to correct for self-report bias when benchmarking alcohol consumption to health consequences. They certainly make no obvious claim to having resolved this difficulty.
The problem reduces to an age-old conundrum that social and health psychologists will be well familiar with -- how can we rely on people to own up to bad habits with the accuracy needed for reliable and valid research?
Such concerns may appear somewhat technical. But they have substantive and practical implications. In the context of advice to the public regarding alcohol consumption, the consequences may even be of life-or-death proportions.
After all, if people find health promotion advice unconvincing, are they really going to follow it?