Thursday, August 6, 2009

Silliness

Two bits of nonsense that I just came across:

1. The Huffington Post reports that the husband of Diane Schuler, the woman who caused the horrific crash in New York that killed eight people, vehemently denies that she had an alcohol or substance abuse problem. Evidence of high levels of cannabis was found in her system, her alcohol level was 0.19 (more than twice the legal limit), and a vodka bottle was found at the scene. And the accident occurred in the early afternoon.

Her husband referred to medical problems she suffered from that may have "caused her" to drink excessively that day (although the fact that she was quite alert with that alcohol level suggests that her tolerance, and therefore her regular use, was quite high). If it could be shown that her alcoholism was secondary to diabetes, it would revolutionize psychiatry. It's an appalling event all around, but in this case denial (by proxy) knows no bounds.

2. A post by the usually astute KevinMD, in discussing the risks of physicians treating celebrity patients, alludes to a study that reportedly found that celebrities are 17% more narcissistic than the general population. Not 15% as previously thought. Presumably this was found either by hooking folks up to a new and improved narcissometer, or by a serum levels of self-regard.

A possible limitation of the study was that levels of narcissism were "pinpointed" by narcissistic researchers.

Are these lead-ins to the coming triumph of DSM-V?

6 comments:

Anonymous said...

A 2% difference could mean narcissism has actually evolved and branched off into an entirely new disorder.

Psychiatric evolution is the science/art of splitting hairs.

James said...

Sorry but an increase from 15% to 17% is not hair splitting.

People confuse the difference between percentage points and percentages.

An increase from 15 to 17 is an increase of 13.3 percent.

Just divide 17 by 15. Answer is 1.1333. Or divide 2 by 15 and the answer is 13.33%

Keep this in mind the next time a politician wants to increase taxes.
An increase of a 4% sales tax to 5% is not 1%. It's 25%.

Novalis said...

Thanks, but I guess I confused the issue by inserting the 15%, which I made up. The point of course was not the difference between the two, but the notion that narcissism can somehow be quantified (people can't agree on a definition, much less a quantification).

Dr X said...

That was sarcasm.

Dr X said...

The 2006 study of celebrity narcissism was co-authored by Drew Pinksy. What a shock.

Dr X said...

"Sorry but an increase from 15% to 17% is not hair splitting."

But even if N wasn't being a little snarky to make his point, a higher score on an objective measure of psychopathology doesn't mean a higher degree of psychopathology is present.

Let's say you've established norms for a depression scale. Using scaled scores with 50 set as the mean for a representative, random sample of adults, you generate norms for the adult population. Then you generate norms on the same scale for a criterion group that has been previously diagnosed depressed. Let's say that you determine that a scaled scored above 65 will predict 95% of the time that a person belongs to the depressed group. Thus, a score above 65 tells you that there is 95% probability that the person is depressed.

If someone obtains a score of 70, that doesn't mean that they are 9.3% more depressed than someone who scored 65. It only means that the probability that they are part of the depressed group is higher. Not 9.3% higher. 95% probability might move to 95.5% probability. It's really hard to get gains in predictive power at the higher end of normative scales.

Anyway, these scores typically represent probabilities of a diagnosis, not severity of the diagnosis.

Now if you're looking at scores below 65 on our hypothetical depression scale, a difference in scores might not tell you anything at all. If, for example, celebrities average 55 on our depression scale (10% higher than the standardization sample mean), that doesn't indicate that they are more depressed than the average adult or even that they are more likely to be depressed. When you're in the range below the cut-off score you're predictive ability falls sharply, quickly diminishing to zero. At scores below the cut-off, clinicians won't make attributions or generate clinical hypotheses about an examinee.