Modeling Environmental Concern for Urban Tree Protection Using Biophysical and Social Psychological Indicators

Abstract

Abstract Urban forest management is being increasingly recognized as a viable policy vehicle for improving community health, sustainable development, and well-being for city dwellers. To improve our understanding of public concern for the local urban forest, a theoretical framework employing concepts traditionally explored in social–psychological and environmental concern studies was tested using data from a public opinion survey of 800 homeowners living in a major urban area in Southern Appalachia, USA. Publicly available geographic information system (GIS) data were analyzed to measure tree canopy density around survey respondents' properties. Structural equation modeling was used to test the validity of the hypothesized relationships among the survey-derived constructs and the spatial data as predictors of public support for urban forest protection policy. The results suggest that homeowners living around urban trees place greater importance on various attributes of trees, and that tree knowledge and experience are indirectly and positively related to support for urban forest protection. Keywords: environmental sociologyGISSEMspatial analysissurvey methodsurban forests Acknowledgments A special thanks goes to Dr. J. Mark Fly and Becky Stephens of the UT Human Dimensions Laboratory, who provided assistance with organizing the survey responses. We also acknowledge Garrett McKinley and Chris Halcomb of Knoxville, Knox County, KUB Geographic Information System, for their GIS technical support. Notes Note. aUnst., unstandardized; bSE, standard error; cStd., standardized; dthis reference variable was not tested for statistical significance; all other unstandardized estimates are statistically significant at p < .001. Note. Suggested levels taken from Bollen (Citation1989), Byrne (Citation2010), Kline (Citation2011), Marsh and Hocevar (Citation1985), and McDonald and Ho (Citation2002). Note. SE is standardized error and was not estimated for indirect effects by Amos; significant differences indicated by ***p < .001, **p < .01, and *p < .05. "Urban forest" is defined as trees in populated areas, such as cities, golf courses, urban parks, and subdivisions (Tyrväinen et al. Citation2003). Survey Sampling International of Fairfield, CT, supplied the randomly selected sample of Knox County homeowners. Homeowners were selected as a study group because they regularly make decisions that affect the urban tree canopy through management of their own homes, yards, and sneighborhoods. Also, homeowners have a vested interest in public expenditures, unlike more transient renters (Youngentob and Hostetler Citation2005). Amos reports a 90% interval around the RMSEA value, in addition to testing for the closeness of fit (PCLOSE). That is, it tests the hypothesis that the RMSEA is "good" in the population (specifically, that it is <.08). Jöreskog and Sörbom (Citation1993) have suggested that the p value for this test should be >.50. The fact that χ2 is statistically significant results in the rejection of H0 that states that the model is an exact fit. Since this test is based on the assumption that the model holds exactly in the population, employing it may be unrealistic in behavioral sciences research. This is because social scientists generally expect that a model should closely approximate some phenomenon, but not perfectly reproduce it. Fit could have also been compromised by the sensitivity of this test to multivariate nonnormality, correlations of observed variables, variables with high proportions of unique variance, and sample sizes greater than 300 cases. Thus, the other fit indices were felt to be more useful in assessing model fit (Byrne Citation2010; Kline Citation2011). A structural equation model is "identified" if it is theoretically possible to solve the covariance structure equations for the unknown parameters (Kline Citation2011). For the four radii, DEN → TV standardized regression coefficients were: β = 0.115, p < .01 for 100 ft; β = 0.087, p < .05 for 250 ft; β = 0.060, p > .05 for 500 ft; and β = 0.047, p > .05 for 1000 ft.