Clustering methods are developed, critiqued, and improved out of theories and research and are then classified, subjectively, in relation to theories, and objectively, by clustering itself, to assist in selecting required methods. A superior method, which combines promising features of several methods (including error reduction in data as it analyzes), yields all, or nearly all, dominant clusters and is offered with techniques for selecting clusters for various purposes. A useful volume for graduate students and scholars of psychological statistics research.
