The Predictive Value of Three Ecological Models

Abstract

The degree to which three ecological models can predict the direction of neighborhood change is evaluated in this article. The life-cycle, arbitrage, and composition models commonly are found in the literature on neighborhood change, but each focuses on a different aspect of this process. To date, researchers have evaluated these models individually, but few have compared their predictive ability. This study uses discriminant analysis to evaluate the efficiency of each model in correctly classifying upward and downward succession of tracts, using census tract data from Cincinnati, Ohio. The composition model had the greatest predictive value, closely followed by the life-cycle model, but each identifies different and complementary aspects of the neighborhood change process.