Bolograph for those of us who aren't scientists, what do you think of the NYT article which 80Kgirl posted above? Is it interesting, well researched, unbiased?
It's biased and wrong-headed. And it's an interesting educational project to figure out why.
Firstly, it cites no references. So all the claims it makes you have no way to check the methodology, the numbers, the statistical treatment, the limitations. It reduces research studies to one-line statements.
Consider, for example, the paragraph:
The same may hold true for hard-drinking teenagers. In 1998, Sandra Brown and Susan Tapert, clinical psychologists at the University of California, San Diego, and at the Veterans Affairs Medical Center there, found that 15-to-16-year-olds who said they had been drunk at least 100 times performed significantly more poorly than their matched nondrinking peers on tests of verbal and nonverbal memory.
Now what is referring to? Well, an opening guess would be [1] except (at a quick scan) (a) I can't find any reference to tests of verbal memory and (b) this is a study about adolescents admitted to state-level substance abuse programmes, and arguing from there to the general population (especially at n=166) is simply wrong. Widening the search and looking in other years, I suspect it's really a reference to [2], from 2000, which does match the summary. Except it doesn't really, and says nothing about the substantial limitations of the work:
Alcohol-dependent adolescents (n= 33) with over 100 lifetime alcohol episodes and without dependence on other substances were recruited from alcohol/drug abuse treatment facilities. Comparison (n= 24) adolescents had no histories of alcohol or drug problems and were matched to alcohol-dependent participants on age (15 to 16 years), gender, socioeconomic status, education, and family history of alcohol dependence. NP tests and psychosocial measures were administered to alcohol-dependent participants following 3 weeks of detoxification.
Matching of participants in a study at n=33 to n=24 is a joke. I take 24 kids from naice homes, and find as if by magic I can match them on family history and education to 33 kids from a substance abuse programme? Sure I can.
But that methodological quibble aside, we learn that being sufficiently much of an alcoholic by the age of 16 that you are in an alcohol abuse programme and require three weeks' detoxification is bad for you. So the results of the paper are not intuitively surprising. And that tells you what about drinking a couple of beers, exactly? What's the shape of the dose-response curve for alcohol in teenagers? Being an alcoholic is bad for you: we knew that. What does it tell you about social drinking? And the answer is: nothing.
That's the flaw in all these papers. They find that, unsurprisingly, heavy drinkers have poor health outcomes, and being scientists who want to get published, say nothing else. University PR departments then unwisely make them interpolate out of sample, to "well, if drinking a litre of vodka a day is bad for you, drinking a small Buck's Fizz alternate Christmases is as well". Well no: we have endless examples were dose-response in humans is wildly non-linear, where you body copes with small amounts with no damage, while excess is severely dangerous. The fatuous example would be water, but better would be paracetamol: if you took a paracetamol for your headache ones a month, it would have no ill effects on you at all. Fifty paracetamol will kill you. The dose-response curve is not linear.
Stuff like this:
In a 2002 e-mail survey of 772 Duke undergraduates, Dr. White and Dr. Swartzwelder found that 51 percent of those who drank at all had had at least one blackout in their drinking lifetimes; they reported an average of three blackouts apiece.
is just a joke. You publish crap like that in the New York Times because even the weakest journal would laugh at you. And they aren't doing research on human subjects anyway, which is they are reduced to nonsense self-recruiting retrospective studies; they are in-vivo animal testers [5]. What does this mean? They canvassed the undergraduates (7000 at Duke), got 772 responses, and of those that said they drank, all had blacked out. Suppose they got back 772 responses, 770 of whom never drank, and 2 drank to excess and had blacked out. That fits the summary, right? Without the numbers, we're nowhere: all we've got is "in 7000 people, a few of them drink to the point of unconsciousness". What, in a university? You're surprised? And they didn't see fit (so far as I can tell) to publish this anywhere, so all we're left with is an anecdote in a newspaper, based on a "study" not even worthy of the name. Big cheer.
Then the NYT article goes off into fMRI woo, and we lose the will to live. Journalists love fMRI: coloured pictures, big machines, strong claims. [3] [4]
[1] Brown, Sandra A., et al. "Psychometric evaluation of the Customary Drinking and Drug Use Record (CDDR): a measure of adolescent alcohol and drug involvement." Journal of studies on alcohol 59.4 (1998): 427-438.
[2] Brown, S. A., Tapert, S. F., Granholm, E. and Delis, D. C. (2000), Neurocognitive Functioning of Adolescents: Effects of Protracted Alcohol Use. Alcoholism: Clinical and Experimental Research, 24: 164–171. doi: 10.1111/j.1530-0277.2000.tb04586.x
[3] Diener, Ed. "Neuroimaging: Voodoo, new phrenology, or scientific breakthrough? Introduction to special section on fMRI." Perspectives on Psychological Science 5.6 (2010): 714-715.
[4] Whiteley, Louise. "Resisting the revelatory scanner? Critical engagements with fMRI in popular media." BioSocieties 7.3 (2012): 245-272.
[5] Obernier, Jennifer A., Aaron M. White, H. Scott Swartzwelder, and Fulton T. Crews. "Cognitive deficits and CNS damage after a 4-day binge ethanol exposure in rats." Pharmacology Biochemistry and Behavior 72, no. 3 (2002): 521-532.