Authors: Wei Tu*, Department of Geology and Geography, Georgia Southern University, Weifeng Wang, Department of Statistics, University of Georgia, Liang Liu, Department of Statistics, University of Georgia, Hoehun Ha , Department of Biology and Environmental Science, Auburn University at Montgomery
Topics: Medical and Health Geography, Quantitative Methods, Social Geography
Keywords: suicide rates, gun ownership, spatial dependence, regression models, United States
Session Type: Paper
Start / End Time: 8:00 AM / 9:40 AM
Room: Tyler, Marriott, Mezzanine Level
Presentation File: No File Uploaded
An extensive body of research has demonstrated the association between gun ownership and suicide rates in the United States and in most of these studies, such a relationship was established using Ordinary Least Square (OLS) regression models. However, OLS is not an optimal method because it cannot take account into the vertical and horizontal spatial dependence in the suicide rates data. We aimed in this study to improve the estimation of the association between statewide gun ownership and county-level suicide rates (all, firearm, and nonfirearm) using spatial autoregressive (SAR) models and hierarchical spatial autoregressive (HSAR) models. We controlled demographic, geographic, religious, psychopathological, and suicide-related covariates in our models and used age-adjusted and smooth suicide rates (2008-2014) as outcome variables. Our main findings are: 1) Compared with OLS, Spatial error, and HSAR, Spatial lag (k=5) model was a better choice for our suicide data; 2) gun ownership was significantly associated with both all and firearm suicide rates but not non-firearm suicide rates; 3) the three final models explained 82.6%, 84.9%, and 76.1% of the variance in the all, firearm, and nonfirearm suicide rates. Findings from this research provide stronger evidence to support the conclusion in the literature that access to and familiarity with firearms is a significant risk factor for suicide.