Table 3 presents the results of this analysis. Residuals were examined for spatial autocorrelation by calculating Moran’s I statistic. The resulting Moran’s I estimate, 0.0015, was not significant (p > 0.05), so we did not correct for spatial autocorrelation in this estimation. We tested for heteroskedasticity using a Breusch-Pagan test. The resulting test statistic was significant at the 1% level. We corrected for this by generating heteroskedasticity-consistent stan- dard errors using White’s method (White, 1980).
All structural and neighborhood variables were statistically sig- nificant at the 5% level. All signs for structural variables and most signs for neighborhood variables were as predicted. The signs for two neighborhood variables were not as expected, those for distance to the closest central business district and distance to the closest shopping center. We had expected these variables to indicate the ease with which homeowners could access shopping and places of business. However, our results indicate that proximity to these features has other negative consequences that reduce property values. It thus seems likely that these features may have nuisance values, perhaps as a result of increased noise levels, congestion, and crime rates in their vicinities that outweigh the amenity value.
Our results clearly indicate that both open space proximity and view attributes influence home sale prices. The coefficients for most view variables were significant and positive, among them those for view area and view percent composition of water and grassy areas. The marginal implicit price of increasing the area of a home’s view- shed by 100 m2 evaluated at the mean home sale price ($255,955) and initial area of 1000m2 is $386. The marginal implicit prices of increasing the percentage of a home’s view composed of grassy surfaces or water by 10%, evaluated at the mean home sale price, are $5517 and $7417, respectively. This illustrates a preference on the part of single-family homeowners for homes with large views including these land cover types.
Surprisingly, although the sign of the coefficient for the percent- age of a view composed of forest was positive, this variable did not significantly impact home sales values, indicating that forested areas are not particularly desirable in residential views. This may be a result of the tendency for trees to restrict views. It is still possible that views that include a high proportion of tree cover, for instance, views of heavily treed residential streets, might positively influence home sale prices. However, this study examined only the influence