Drinking is bad for Teens, National Public Radio 25 Jan

NB: after I wrote this, I went back and checked the 1st link. It seemed to me that the content at NPRs website had changed; if so, this story is outdated. If the link is correct, this blog post is somehwat beside the point, although technically correct
The change is in the prominence given to the psychological study, discussed below, in the updated npr story, the psychological study is clearly 2nd to the imaging, although to my memory, in the original version of the npr story, the imaging was secondary to the psychological study. In any event, I stand by my criticisms of the psychological study.


The story:
On Jan 25, NPR had a story entitled "Teen Drinking May Cause Irreversible Brain Damage", based on a scientific publication from the Tapert Laboratory at UCSD.

Technical flaws:
The NPR print version does not contain a link to the publication (here ) but rather to the journal which carried the publication; the interested reader has to then search through the journal website to find the actual article.
Also, although this research is almost certainly financed, at least in part, by Tax Payer Dollars, only the abstract of the publication is freely available; you have to pay $11.95 if you want to read the paper.

Science Problems:
This study looked at 76 teenagers, who, at the start of the study, had little or no alcohol use. The researchers subjected the teens to various tests of mental performance, and followed the teens for several years, administering the tests at later points.
After 5+years, the researchers compared the test performance of those teens who never drank (teetotalers) to those who became drinkers.


A careful examination of the data provided in the publication shows that the difference in test scores between the drinkers and non drinkers was small, and only weakly statistically (p < 0.05) on ONE test.

The authors then used complex statistics to try and find a meaningful difference between teetotalers and drinkers. Even with sophisticated data analyses (Table 3and Figure 2) there is not much of a difference between the drinkers and nondrinkers (there is an outlier in Figure 2 that skews the data).

Beyond the very small effect that was actually found, there is a problem in linking these results to teens in general. The authors found their 76 teens by solicitations at middle schools; of the people they interviewed, only 13% were accepted into the study, this means that they started off with a highly select group of people, who may not represent the general population.
Further, one wonders if the format of the test affected performance; for instance, what if the teens were tested in a manner that would interest the drinkers ? One might test the teens spatial reasoning by showing them two glasses, of different shape, filled with a clear liquid, and asking them which glass held more liquid; perhaps teens who drink would do better on such a test.

Conclusion:
I do not believe that the data in the this publication support NPR's reporting.