home page links quotes statistics mission statement success stories resources Lighter Side Authors! Search Page
June 14, 2016

Reason and violence

By Bernie Glos



Having done much study in the field of violence risk assessment and done quite a few violence risk assessments myself, I would like to bring some rationality to the discussion in light of the attack in Orlando on Sunday. Predicting anyone's future behavior is sketchy at best; predicting violent behavior even moreso.

Violence is a very infrequent event. Very infrequent events are statiscally difficult if not impossible to predict accurately. Violence is not a single dimension; there are various dimensions of violence. Some individuals are more at risk for frequent small outbursts; a very few have little history until a serious event occurs. For some, the liklihood is obvious (e.g. Jared Loughner who injured Rep. Gabby Giffords and killed others or James Holmes, who shot up the movie theater in Colorado). Both of these events may have been prevented based on available information. For others, like Dylan Klebold of Columbine, there was no information that would have led to the predicted shootings. He kept things very tight to himself, in an era before social media was prominent.

Even in the area of violence risk assessment, we do not attempt to "predict" future violence or assign a probabillity function.. All we can do is assess the risk factors that accumulate to low, moderate or high risk of violence. But not everyone with an assessed high risk of violence will actually engage in violence. For youth, the risk assessment is more accurate. Using the right tools (some violence risk measures result in predictions that are no better than chance) and using all information available, of 100 actually violent kids, the assessment can at best accurately predict 87 of them. For adults, the numbers are not as good. (Most kids have a bunch of readily available and consistent information especially regarding school, etc. which adults lack). At best we can probably identify 81 of the 100 adults who actually commit a violent act.

Certainty in the prediction of future violent behavior will always elude us.