"We can’t go out and lock up all the socially awkward young men in the world..."

"But we have to try to prevent the unpredicted."

Says a psychiatrist quoted in a WaPo article titled "Predicting violence is a work in progress."

The photos at the link — a line-up of Jared Lee Loughner, James Holmes and Adam Lanza — create the impression that you can tell by looking at them, especially at their eyes. But if you look long enough, you'll see the 3 men are quite different. Only Loughner is smiling. One may sense that madness radiates out of his face, but if you can exclude what you know — that's a photo taken after his shooting spree — he may seem like a fairly ordinary guy. Holmes's face, especially stuck between those other 2, looks open — even empathetic — and sad. Lanza looks abnormal, like an alien. The oddness is enhanced by knowing that this isn't a mug shot like the other 2. Is that his driver's license pic? It's hard to believe — in this age of digital photography — that a picture that came out that bad wouldn't be trashed. If I were diagnosing Lanza from that photograph, I'd say his problem was anorexia. What that boy needs is cheeseburger*... and a better haircut.

An analysis of 20 studies published three years ago found that schizophrenia increased the risk of acting violently fourfold in men and even more in women. The risk of schizophrenics committing homicide was 0.3 percent — more than 10 times greater than the average citizen.

What is the risk of serious violence — not just homicide — for schizophrenics who are also young and male? What is the risk for young, male schizophrenics of the paranoid subtype? If we're going to reason from statistics, we need to be able to look at the numbers in different ways. I suspect that the 0.3 figure — which screams you can't just institutionalize schizophrenics — is massively diluted by including large numbers of females, over 30s, and the nonparanoid subtypes (disorganized, catatonic, undifferentiated, and residual).

Back to the WaPo article:

John Monahan, a University of Virginia psychologist... and many others came up with a constellation of “risk factors” and “protective factors” for violent behavior.... [that] the presence of a mental disorder as only a small contributor to risk, outweighed by other factors such as age, previous violent acts, alcohol use, impulsivity, gang membership and lack of family support.

Gang membership? Now, you've got a list of factors that's off the subject of mental illness and more about a young male's social context.

“From our research, we could quickly distinguish between a patient whose chance of being violent was 1-in-10 from one whose was 1-in-2,” [Monahan] said.

Some statistics skepticism: he's saying "violent," not homicide or even serious violence.

[A British Medical Journal analysis found that of] the people predicted to “violently offend,” 41 percent did. Of those predicted to be nonviolent, 91 percent were. In practical terms, that meant that if authorities used the tools for the purposes of public health, they’d have to detain two people to prevent one from becoming violent.

That is patently defective reasoning. Where did they draw the line in scoring individuals using their set of factors? Show me how the factors were scored and who ended up being violent. Why have you simply divided people into 2 groups? Look at different subsets within the predicted-to-be-violent group with the highest scores. For example, if you break out 10% of them, the predicted-to-be-violent ones with the highest scores, what percentage of them went on to commit acts of serious violence? If that approaches 100%, then the "practical terms" about the fairness of detention would look entirely different.
________________________________

*Adam Lanza was a vegan — "He didn't want to hurt animals."