Influence of Spatial Scale on Predicting Hunting License Sales

Previous studies showed that many factors influence hunting license sales. Developing descriptive models helps to predict sales of hunting licenses. However, what is not known is if these same predictors behavior in similar... [ view full abstract ]

Previous studies showed that many factors influence hunting license sales. Developing descriptive models helps to predict sales of hunting licenses. However, what is not known is if these same predictors behavior in similar manner when used in models across differing spatial scales. We examined the influence of different spatial scales on models predicting hunting license sales in Illinois by creating models using two spatial scales: 1) counties and 2) census tracts. All possible predictors (both socioeconomic and biophysical) were collected according methods employed in previous empirical studies. Geocoded socioeconomic data were used as independent variables and hunting license sales served as our dependent variable. Influential predictors were selected by random forest and step-wise regression methods, and then taken into simple linear regression with logit and Poisson transformations made according to the diagnostic plots of residuals. Cluster maps (“heat maps”) were developed to depict spatial correlations between the predictors and the license sales. We propose that county and census tract models have different predictors, and the same variables exert different influence in different models. The result may help to guide the future hierarchical linear modeling (HLM). Further, understanding the influence of spatial scales on models predicting license sales may assist state fish and wildlife agencies develop focused strategies to promote license sales.