"A new large study challenges the idea that mental illness alone is a leading cause of violence. Researchers instead blame a combination of factors, specifically substance abuse and a history of violent acts, that drives up the danger when combined with mental illness in what they call an 'intricate link,'" writes Carla Johnson of AP. See also this article by Kathleen Doheny in WebMD Health News. The abstract of the article by researchers Eric Elbogen and Sally Johnson is available here at the Archives of General Psychiatry (full text requires $).

This is a longitudinal study, which is better than a cross-sectional study. The latter type takes one snapshot at one point in time and typically finds a correlation: A and B tend to be found together.  Does that mean A causes B, B causes A, A and B are both caused by C, or some other more exotic chain of causation? The cross-section doesn't tell us. In a longitudinal study, there is more than one snapshot at more than one time. If A precedes B, we can rule out B causing A as a complete explanation of the correlation.

From a criminal law perspective, if there is no causal link between a particular defendant's person's mental illness and his crime, should that illness be considered mitigating? Persons of sense would not consider it mitigating that a bank robber had the flu. Casual connection between a mental condition and a crime cannot be assumed. In the absence of such a connection, the condition is simply not mitigating.