Replication Study of Elder Suicide and Elder Gun Suicide Models Using 2010 Population Data

Background: Adults age 65 and over are the age group most likely to use guns to take their own lives.  Rates of firearm suicide began increasing with the economic downturn in 2007-2008, and have continued to increase despite the subsequent economic recovery. Although studies have examined individual risk factors for suicide among older adults, less is known about macro variables leading to higher rates of elder suicide. Building on Durkeim’s model and previous work by the authors, a cross-state retrospective replication study tested models of elder suicide and elder gun suicide rates for adults 65 and over in the U.S.   

Methods: The predictor variables for the models included the following: elder males, divorce, political climate, economic climate, gun access, and violence climate. AMOS 25.0 (Arbuckle, 1997) was used to estimate elder suicide and elder gun suicide models and estimate the fit of these data for the theoretical and measurement models. Secondary data were collected from each state (N=50). Political climate, economic climate, gun access, and violence climate were measured using the same indices as a previous study by the authors.

Results: The elder suicide model explained 58 percent of the variance in state elder suicide rates and was a statistically significant predictor of suicide rates (F (6, 43) = 9.86, p< .001). Six of nine direct paths were statistically significant predictors of elder suicide and political climate was responsible for 71% of the gun access—which was responsible for 21% of the elder suicides when controlling for the other variables. The elder gun model explained 71 percent of the variance in state elder gun suicide rates (F (6,43) = 19.31, p<.001). Eight of the nine path coefficients were statistically significant predictors of elder gun suicide.  Gun access was responsible for 40 percent of the differences in suicide rate and political climate was responsible for 68 percent of the variation in gun access.  The fit indices for both models were mixed—many indicating a good fit and several fit indices showed some issues to explore further as we refine the models. The calculated GFI and AGFI were .928 and .777, well within the expected range of 0 to 1.00 and indicative of an excellent fit.  The NFI and CFI values were also indicative of a good fit at .922 and .975, respectively. The Elder Suicide Model CMIN index was not ideal however (χ2(9, N=50)=12.57, p=.183.) 

Conclusions and Implications: The present study added to the evidence supporting the use of macro-level variables in the prediction and explanation of elder suicide and elder gun suicide. Although these data showed more explained variance than our first study, the fit indices demonstrated some problems with goodness of fit. The economic recession of 2008 had large effects on the economy across the country and this is one possible explanation for the differences between the 2000 and 2010 data. Through a deeper understanding of the ways our gun culture influences violent self-harm, we can outline evidence-based practice, design protective policies, and design adequate prevention programs.