STATS ARTICLES 2007
Is There a College Substance Abuse Crisis?
Are things really getting worse or did Columbia’s National Center on Addiction and Substance Abuse get the media’s attention through the selective use of statistics?
Are we “Wasting the Best and the Brightest?” as the latest report on use of alcohol and other drugs by college students by Columbia’s National Center on Addiction and Substance Abuse claims? Or is what Stats has dubbed the “Center for Abuse of Statistical Analysis” (and others have labeled the “Center for Alcohol Statistics Abuse”) up to its old tricks?
Unfortunately, while cliché watchers will have a field day with this report, so too will those who need examples of misleading uses of statistics for their Stats 101 classes. In this case, the problem is not CASA’s usual confusion of correlation and cause, but rather its selectively choosing a date with which to compare college students so as to make today’s kids look worse than previous generations and, by implication, worse than those who do not attend college.
Careful readers of the report – a category which excludes almost all the journalists who reported its findings verbatim – may have wondered why CASA chose to look at trends in alcohol and other drug use from 1993 to 2005, rather than from the more obvious ten-year starting point, 1995 – or from the 1970s, when statistics on the issue first began to be kept. Informed readers might also be curious about why the comparison point wasn’t the highest level of drug use measured in teens (which occurred in 1979-81, depending on the particular drug); and they also might wish to know why the students were compared with the general population rather than with others their own age who do not attend college.
If such readers go to the source for most of CASA’s data, the government’s Monitoring the Future (MTF) and National Household Survey on Drug Use and Health (NHSDUH) studies, they will rapidly discover why.
Take, for example, the startling claim that in 2005 “almost one in four college students (22.9%) met the medical criteria for substance abuse or dependence, almost triple the proportion (8.5%) in the general population.” The source of this statistic appears to be a re-analysis of data collected for MTF or NHSDUH done by CASA.
This makes it appear as though college students are more likely to be addicts or alcoholics than non-college students, which is something that confounds common sense when you consider that the most severe cases of addiction start young and often result in failure to complete high school, let alone attend college – and that addiction itself often causes college dropout. (Note: “substance dependence” is the medical term for addiction; “substance abuse” is the medical term for use of substances that is potentially harmful but is not characterized by compulsion or long-term problems).
When you look at the NHSDUH figures for the general population age 18-25, you find a rate of substance abuse or dependence for 2005 of 21.8%. While this sounds equally as horrifying, the reason the rate is probably slightly higher for college attendees is that college binge drinking can often result in a “substance abuse” diagnosis: in other words, it’s potentially dangerous but is not necessarily indicative of a long-term problem.
What CASA fails to point out (but is hinted at in the 8.5% substance abuse/dependence rate for the general population) is that the vast majority of college binge-drinking ends when graduates realize it is not compatible with employment that requires 9am cognitive clarity at work. The substance/abuse dependence rate for those over 26 is just 7.1%. In other words, stop the presses: many young people experiment with drugs and alcohol before they settle down and grow up.
CASA was forced to admit that college binge drinking itself has been steady between 1993 and 2005, so they problematized this as being an instance of “no significant decline” and pointed out minor changes in subcategories like a 16% increase in binge drinking “three or more times in the past two weeks.”
In the field, this is known as “data dredging:” Your main finding is not really that significant, so you parse enough subcategories in order to find something that looks scary or important. (Which, naturally enough, media reports in the vein of the Associated Press’s “Binge Drinking Rises at Colleges,” via Forbes.com, repeat without qualification or analysis).
Now we come to the curious choice of 1993 as the comparison year for the findings taken from Monitoring the Future . CASA notes that during that period the number of daily marijuana users more than doubled, going from 1.9% to 4%. If they had chosen 1995, however, the figure would have been 3.7%, which would have made the increase far less impressive. Even less scary would have been to note that in 1980, in the peak period of U.S. drug use, 7.2% of college students smoked marijuana daily and the nation did not collapse (in fact, this is the generation that produced the Internet boom).
CASA’s focus on a “tripling” of past-year heroin use is even more misleading: In 1993, 0.1% of college students reported past-year heroin use; in 2005, the figure was 0.3%. But this “tripling” had happened by 1995 and had bounced all the way up to 0.6% in 1998, before falling back to 0.1% again in 2002. In other words, if they’d used 1995 as a starting point, they’d have been forced to say that heroin use was unchanged. And this was probably true because most researchers believe that because heroin use is so rare, trends in these numbers among college students (who are less likely to use heroin than high school graduates, drop-outs and non-students) probably don’t mean much anyway.
CASA’s highlighting of a “52% increase” in use of “drugs other than marijuana” since 1993 shares the cut-off problem with the other figures. The 1993 statistic is 5.4% but by 1995 it’s already up to 6.3%, yielding only a 30% rise to hit 2005’s 8.2%. The comparable 1980 figure is a whopping 20.7%. The same problem is seen in the cocaine trend as well.
CASA did point out what appears to be a genuine recent increase in misuse of prescription drugs, but it buries the fact that this is far less common among college students than among those who do not attend college, failing to mention this in the press release or introductory letter. In fact, nearly twice as many of those who are not enrolled in college misused prescription opioids like Vicodin in the past month (5.6%) compared to 3.1% for college students.
CASA says it wants to take the “high” out of “higher education;” but because college graduates are far less likely to have long term substance misuse problems than those who do not attend or do not graduate, maybe it should focus more on keeping students in school rather than hyping fears about college drinking and other drug use.
That college graduates remain less likely to get or stay addicted to alcohol or other drugs despite vastly expanded enrolment suggests that while “stay in school” may be a boring message, it may be the best widely-applicable form of addiction prevention we have yet to develop.