Iain Murray argues here that alcohol abuse guesstimates aren’t statistics. Therefore, we shouldn’t treat such guestimates as if they were.
Alcohol Abuse “Guesstimates”
If every night is party night on college campuses, a new study warns of serious problems for students. Researchers at Boston University charge that their alcohol use kills over 1,400 students annually. Their report also claims that over 600,000 students a year are assaulted by other students who have been drinking. Additionally, over 70,000 are the victims of sexual assaults in similar circumstances. These are worrying figures, indeed. But they do not stand up to scrutiny.
Estimate of Health & Safety Problems
The researchers were attempting to estimate the number of health and safety problems related to alcohol. To estimate deaths, they based their calculations on the total number of traffic fatalities. Specifically, for the age group 18-24 that involving alcohol. In doing so, they inflated the numbers.
That’s because they included crashes where the BAC of drivers was often much less than the legal limit. Thus we have no way of knowing whether alcohol was in fact the cause of crashes. Driver may have had one drink or less and still be perfectly fit to drive when involved in fatal crashes. However, the researchers have attributed all deaths to alcohol alone.
Moreover, the methodology used to establish how many of the total crashes involved college students assumes too much. The study claims that as college students make up 31 percent of the age group nationwide. Therefore, it assumes that they must also account for 31 percent of alcohol-related deaths in that age group. However, there’s no evidence that students have traffic crashes at the same rate as other people their age. Many students after all do not own cars, or they live on campuses where walking is the norm. The researchers report that a higher proportion of students drank and drove at least once in the last year. That’s in comparison to others of their age group. But this says nothing about how often the respondents did so, which is the thing we need to know.
For non-traffic deaths, the story is similar. The researchers took 31 percent of the overall number of deaths for this age group. Again with no basis for this assumption. They also used another calculation to assess how many were alcohol-related. This calculation was based on a survey of accidental deaths over ten years. It found that 38 percent of people who died in accidents had a positive BAC.
This is problematic for two reasons. First, it assumes that any blood alcohol level at all was a contributory factor in the accident. Second, the data cover a broad span of time over which per capita alcohol consumption dropped greatly.
The figures for assaults are demonstrably inaccurate, because they conflict with what we know about crimes in general. Standard data used in this field tell us that there were about 735,000 alcohol-related assaults on 18-24 year olds nationwide. To suggest that 86 percent of these occurred on campuses is clearly implausible. That’s because we know from other standard data that violent crime victimization correlates closely with less education. A similar argument applies to the issue of sex assaults and date rapes. The researchers suggest that 65 percent of these assaults on the age group nationwide occur on campuses. Furthermore, that they are exclusively caused by alcohol.
This is a common problem with studies such as this one. Academic surveys that ask people general questions about their experiences with crime often produce higher figures. That’s in comparison with the tighter government surveys.
Researchers refine those over time to tighten definitions and reduce problematic gray areas. One-off surveys like the one the researchers used in this case do not have that luxury. Therefore, the numbers they produce tend to suffer inflation to a greater or lesser degree.
Surveys also suffer from confusion over what they really discovered. The survey in question asked respondents whether they had been “assaulted, hit or pushed” by someone who had been drinking. The press release reduced this to “assaulted or hit.” Dr. Mark Goldman was co-chair of the Task Force on College Drinking. When he appeared on The Early Show, he mentioned only “half a million assaults.”
Garbage in, Garbage Out
Therefore, we cannot rely on the numbers produced by the Boston University team. They use questionable assumptions, overly broad definitions and data that conflict with more reliable and established figures. There is no doubt that alcohol abuse is a serious problem that impacts significantly on many lives each year. These figures, however, do nothing to prove that the young people are in severe peril from the demon drink.
Iain Murray was the first Director of Research at the Statistical Assessment Sevice. He is currently a Competitive Enterprise Institute vice president and Senior Fellow. Mr. Murray holds master’s degrees from Oxford and the University of London. He specializes in the analysis of crime statistics. Reprinted with permission from www.techcentralstation.com and slightly edited.
Resources: Alcohol Abuse Guesstimates Aren’t Statistics
- Brignell, J. Sorry, Wrong Number! The Abuse of Measurement. London: Brigne, 2000.
- Huff, D. How To Lie With Statistics. NY: Norton, 2010. (Helps avoid falling for alcohol abuse guesstimates.)
- Milloy, S. Junk Science Judo. Self-defense against Health Scares & Scams. Washington: Cato, 2001.
This website is informational only. Thus it gives no advice on alcohol abuse guesstimates aren’t statistics.