Exploring spatial variations in the relationships between residents' recreation demand and associated factors: A case study in Texas

Abstract

In the field of outdoor recreation, past research has attempted to identify the relationships between the socioeconomic status of recreation participants and demand. This approach has used global regression models that focus on local or site-level recreation demand prediction. No studies have applied spatial statistical approaches to examine recreation demand at the state or regional level. The objectives of this study are twofold: 1) to identify socio-demographic and economic determinants of the spatial disparities of residents' demand for national park visitation; 2) to demonstrate the spatial variation in the relationship between residents' demand and its associated factors through a local regression modeling technique (GWR). The analysis was conducted at the county level for Texas, using public and private levels of secondary data. A particular version of OLS (ordinary least squares), stepwise regression, was used to select significant explanatory variables for developing the model. Six explanatory variables were selected for the modeling procedure: 1) older adults; 2) family structure (traditional vs. nonfamily structure); 3) recreation-related spending; 4) education, and 5) poverty rate.The study results found that the Moran's I value for national park visitation showed a positive spatial autocorrelation across the state which means that spatial distribution disparities of residents' recreation demand exist in Texas. The variable relationships differed by specific counties and spatially in predicting residents' demand. The GWR model made improvements in model performance over the OLS model. Overall, the strongest positive influence on national park visitation was found for those who spent $250 or more to purchase recreation-related equipment. In conducting future research, the GWR model is a useful statistical technique to examine and compare spatial relationships across regional study areas and can be complementary to global statistical analyses.