Date of Award

Summer 6-2023

Degree Type


Degree Name

Doctor of Philosophy (PhD)



Committee Chair

Daniel Capron

Committee Chair School


Committee Member 2

Kelsey Bonfils

Committee Member 2 School


Committee Member 3

Megan Renna

Committee Member 3 School


Committee Member 4

Richard Mohn

Committee Member 4 School



Nearly 50,000 individuals die by suicide each year and, despite only representing 7% of the population, military veterans represent nearly 14% of suicides. As a critical public health concern, research has been directed toward identifying risk factors for suicide. Over the last 50 years of research, nearly 3,000 separate effect sizes for suicide risk factors have been identified, but this research has made little progress toward identifying individual suicide risk and predictive capacity remains low. Although research from medicine and traumatic injury suggest that public health interventions that are implemented broadly can be highly effective at reducing risk for other causes of mortality, suicide research has largely failed to adopt these strategies. One method for developing public health interventions for suicide could include the identification of demographic factors associated with psychographic profiles of groups that appear to be at risk for suicide. The present study will address these goals through the following study aims: 1. To develop an atheoretical model of suicide risk in United States military veterans; 2. To isolate a population of individuals with similar psychographic profiles to individuals with prior suicide attempts; and 3. To identify demographic and psychological similarities between latent profiles within this group. Data from the Military Suicide Research Consortium’s Common Data Elements will be analyzed to assess these aims. First, linear regression was used to identify risk factors with the strongest predictive capacity for suicidal thoughts and behaviors. Second, a zero-inflated count regression was used to identify a group of “excess zeros” with similar psychographic profiles to individuals with a history of prior suicide attempts. Third, latent profile analysis was used to identify any latent profiles within this “excess zero” group to define potential risk groups. Zero-inflated negative binomial regression had superior model fit to uninflated models for the overall population, but not the military ideator subsample. Latent profile analysis identified a two-class solution and four-class solution that were then used to compare demographic profiles and suggest potential avenues for targeting at-risk or underreporting groups identified.



Available for download on Friday, June 13, 2025