Geographic spatial autocorrelation and United States suicide patterns |
| |
Authors: | Ira M Wasserma Steven Stack |
| |
Institution: | 1. Department of Sociology , Eastern Michigan University , Ypsilanti, MI, 48197, USA;2. Department of Sociology , Wayne State University , Detroit, MI, 78076, USA |
| |
Abstract: | Research on aggregated geographical data in such fields as genetics and epidemiology have produced inefficient findings when spatial autocorrelation is present. This inefficiency results from the bias in the standard errors of the regression coefficients for multi-variate studies. It is unclear as to whether the numerous suicide studies that utilize geographical units of analysis am also flawed. The present study employs Odland's test for spatial autocorrelation to ascertain whether this error is present in geographical suicide studies. Two typical studies that employ state and county (parish) data art examined to decide whether spatial autocorrelation limits the findings of these studies. Using a conservative randomization test with Moran's I, it is concluded that for these studies spatial autocorrelation is not a serious problem. However, other weight matrices, and less conservative tests are needed to further support the findings of this paper. |
| |
Keywords: | Moran's I parishes spatial autocorrelation states suicide weight matrix |
|
|