The Statistical Prediction of Environmental Preferences

This paper discusses a number of statistical and interpretive problems which arise in the analysis of complex, multivariate data common in the study of man-environment relations. The problems are encountered in the course of a detailed analysis of a multiple regression model of landscape preferences developed by Shafer and coworkers. Questions raised by the analysis suggest that the predictive validity of the model and its usefulness for making any resource management decisions have not yet been demonstrated.