Performing risk assessment is a complex, and some would say, volatile practice.  The rate of false positives is high, and this may result in some individuals being incarcerated or refused release when they in fact would not have committed another crime.  While these predictions are not as accurate or easy as often depicted, there is some evidence to suggest that under certain conditions such predictions could be accurate.  These conditions include a number of demographic and criminological factors, and in some cases the presence of certain psychiatric symptoms may provide indications, though these are not as accurate as factors such as drug and alcohol use.  The fact that such predictions are only moderately accurate should not be surprising.  No human behaviour that is significant, or in this case extreme, is the result of any one or two factors.  Usually, the presence of violence is a result of the interplay between numerous factors including psychological, social and situational determinants.  The ability to identify and acknowledge all of these is difficult, and so often assessments might be made on incomplete case data.  Although many tools have been explicitly designed for this purpose, it is unlikely that predictors in the immediate future will be any more accurate than a slightly better chance level, and subject to error.