Organization taxonomy: searching for performance while solving the problems of polythetic classification

The lack of a comprehensive organizational classification system limits research design by calling into question the adequacy of sampling frames and therefore the generalizability of empirical results. Following the precepts of earlier organizational taxonomists (McKelvey, 1982; Miller & Friesen, 1984), this dissertation used rigorous taxonomic methods to demonstrate that developing polythetic organizational classification is possible and the results are useful.

The purpose of this dissertation was to see if:(a) the use of rigorous taxonomic methods (narrow sample of organizations and broad array of variables) is an appropriate method to discover organizational differences; (b) the results of taxonomy distinguish between populations with respect to the relationship between the internal systems of variables and performance; and (c) general taxonomic results are different, more descriptive, and broader in predictability than those from special purpose typological approaches.

Survey data from Texas apparel manufacturers used five pre-existing scales and other indicants to measure concepts such as performance, reasons for business ownership, management functions, negotiating, strategic posture, products, organizational characteristics, and individual demographics.

Clustering algorithms were used to determine the natural number of populations within the data and population membership. Regression analysis was used to determine the system of internal organizational variables that relate to performance.

Results indicated that four populations existed within the sample data — lifestyle firms, artisan firms, organic firms, and mechanistic firms. These populations follow characteristics of polythetic classification and are relatively stable.