Model May Predict Violence Among Soldiers

Close-up of U.S. Army uniformA computer model can successfully identify the United States Army soldiers who engaged in the most acts of workplace violence over a six-year period, according to a study published in Psychological Medicine. Researchers found that 5% of soldiers committed more than a third of all violent crimes.

Using a Model to Predict Workplace Violence

In 2009, Army psychiatrist Nidal Malik Hasan fatally shot 13 people at Fort Hood, Texas. The rampage, which killed more people than any other mass murder at a U.S. military base, ignited concerns about soldiers’ mental health and workplace violence.

Harvard Medical School postdoctoral fellow Anthony Rosellini and his team wanted to explore which factors might reliably predict military violence. Using Army and Department of Defense records, the researchers reviewed the records of nearly 1 million soldiers on active duty between 2004 and 2009.

Using a computer model, researchers assigned risk classifications to each soldier studied. The high-risk soldiers identified by the model committed a large portion of all crimes over the six-year study period. The 5% of soldiers who had the highest predicted risk accounted for 36% of crimes committed by men and 33% of crimes committed by women.

After testing the model on a more recent group of Army soldiers between 2011 and 2013, researchers found the 5% of soldiers predicted as highest risk accounted for 50% of all significant violent crimes.

Can Violence Be Predicted?

The study’s authors suggest their research could eventually be used to offer early interventions for soldiers at the highest risk of committing future crimes. However, they do not claim to be able to predict all violent crimes, and they stress that severe violent crime is still uncommon even in the high-risk group. It is also unclear whether early interventions would be effective in reducing violent crime, as they have not been tested yet.


  1. Dallas, M. E. (2015, October 8). Research may help spot soldiers at risk for workplace violence. Retrieved from
  2. Kenber, B. (2013, August 28). Nidal Hasan sentenced to death for Fort Hood shooting rampage. Retrieved from
  3. Predicting which soldiers will commit severe, violent crimes. (2015, October 6). Retrieved from
  4. Thompson, M. (2009, November 18). Fort Hood: Were Hasan’s warning signs ignored? Retrieved from,8599,1940011,00.html

© Copyright 2015 All rights reserved.

The preceding article was solely written by the author named above. Any views and opinions expressed are not necessarily shared by Questions or concerns about the preceding article can be directed to the author or posted as a comment below.

  • Leave a Comment
  • Mike


    October 13th, 2015 at 3:29 PM

    So my big question is this: how do you then put this info to good use and stop them before the violence is committed?

  • len


    October 14th, 2015 at 3:32 PM

    I do not believe this.
    If it is accurate then someone should have started using this model of prediction a long time ago!
    We now have a broken system and I am not sure that this is what can fix things or predict it fast enough.

  • Tabby


    October 16th, 2015 at 12:43 PM

    My issue with all of this is that we expect someone to change who they are instantly the moment they are no longer in war or action. How does anyone even do that? You are so accustomed to behaving one way and thinking one way and then you come home and are expected to be something totally different. There are very few of us who can turn on a dime like that. I am not taking up for the violence but I do know that this has to be a very difficult transition for so many of our soldiers that there needs to be even more resources out there and available to them.

Leave a Comment

By commenting you acknowledge acceptance of's Terms and Conditions of Use.

* Indicates required field.

GoodTherapy uses cookies to personalize content and ads to provide better services for our users and to analyze our traffic. By continuing to use this site you consent to our cookies.