Researchers find tests meant to predict future violence by psychopaths is less accurate than chance

[Jim__]

The article indicates that the tests are actually as accurate as chance for predicting future violence by psychopaths.

From Medical Xpress:

(Medical Xpress)—A trio of British researchers has conducted a study that has revealed that tests given to jailed psychopaths to predict the likelihood of engaging in future violence, are less accurate than chance. In their paper published in the British Journal of Psychiatry, Jeremy Coid, Simone Ullrich and Constantinos Kallis describe how they interviewed and gave tests to inmates in British prisons and then followed up later to see if they engaged in violent activities after release—they found that tests given to predict such behavior in psychopaths were no better than 50 percent accurate.

...

In analyzing the results, the researchers found that the tests did reasonably well in predicting behavior in people with no discernible mental illness—they proved to be approximately 75 percent right in predicting whether they would be jailed again for violent behavior. The tests were less accurate for those with mental ailments such as schizophrenia, with a success rate of just 60 percent. Predicting whether a person diagnosed as a psychopath would re-offend, sadly, was no better than 50 percent, which, the researchers point out, is no better than flipping a coin. For this reason, they suggest that courts stop using such tests when considering early release of such prisoners.

[Scuba]

1. The universe of our ignorance is vast to the point of being unmeasurable.

[Xipe Totec]

2. I can believe it's no better than chance, but don't believe worse than chance.

Because if the test is worse than chance at predicting an outcome, all you have to do is flip the results to get a test that is better than chance as predictor.

[Jim__]

3. Ouch! I would hope not.

If it's worse than chance what you have is a seriously incorrect theory. A seriously incorrect theory shouldn't be put to any use that has a major impact on people's lives. The problems need to be understood and the theory needs to be fixed.

[Xipe Totec]

4. It's a mathematical model, not a theory.

And as the article states in the body, contradicting the headline, it is no better than chance. That refutes whatever theory may have been attached to the model.

If the model had in fact been worse than chance, that would be an area for further investigation, a source perhaps of new theories and corresponding models.