Over time psychologists have developed a number of models to predict which individuals are most at risk of dying by suicide, but for the most part these models have been tested only on civilian populations. Because there are a number of differences between civilians and servicemembers that could affect suicidal behavior, it cannot be taken for granted that models that are effective in predicting risk among civilians will work equally well for military populations. Thus Michael Anestis at the University of Southern Mississippi tested several of these models on a group of servicemembers to see which are most effective at predicting the likelihood of suicide. 

Over the first six months of the study, Anestis and his colleagues collected data from 1,000 Veterans at Camp Shelby outside Hattiesburg, Mississippi, an Army National Guard base which is one of the primary locations for demobilization, especially of servicemembers arriving home from Afghanistan. The data collected included information relevant to predicting suicidal behavior, such as the presence of depression and feelings of hopelessness. Anestis collected follow-up data from the subjects at 6, 12, and 18 months after the initial interview. The follow-up data, was collected online questionnaires, included information about suicidal thoughts and suicide attempts.

Dr. Anestis and his colleagues determined soldiers have a tendency to underreported suicide ideation when military personnel have access to their responses. They also discovered that, although personal firearm ownership does not appear to prompt the development of suicidal ideation, how soldiers store their personal firearms may impact their ability to act upon thoughts of suicide. They also gained greater clarity on the manner in which specific components of PTSD related to various aspects of suicide risk. Lastly, they expanded the research on the interaction effect between insomnia and agitation in the prediction of suicidal ideation. More details about their findings can be found in their published manuscripts.