Representing and Testing Organizational Theories: A Holistic Construal

TLDR

The authors present a holistic construal to represent and test organizational theories, illustrated with contemporary theory examples and data from a wholesale distribution study. Their methodology links theory construction and testing by explicitly modeling theoretical and empirical concepts, nonobservational hypotheses, and correspondence rules, while allowing researchers to quantify and control random and systematic measurement error. Comparisons with traditional procedures reveal that the latter can produce inaccurate conclusions about constructs, hypotheses, and measures. The research was funded by a Senior Fulbright Research Grant and a DREF grant, with acknowledgments to collaborators and reviewers.

Abstract

This research was supported by a Senior Fulbright Research Grant to R. P. Bagozzi and a grant to L. W. Phillips from the Distribution Research and Educational Foundation (DREF), National Association of Wholesale Distributors, Washington, DC. Special thanks go to Dirk Van Dongen, Executive Director of DREF, and Ron Schreibman, Director of Research, for their assistance throughout the project. The contributions made to the research by Louis W. Stern, Northwestern University, are gratefully acknowledged. Comments on an earlier draft by Claes Fornell, University of Michigan, Barbara Lawrence and Alvin Silk, MIT, and V. Srinivasan and Jeffrey Pfeffer, Stanford University, also contributed to ideas presented in this paper, as did suggestions made by the editors. A holistic construal is presented to represent and test organizational theories, using examples from contemporary organizational theory and data from a recent study of wholesale distribution companies. The methodology provides a mechanism for linking theory construction and theory testing in organization research by explicitly representing theoretical and empirical concepts, nonobservational hypotheses, and correspondence rules. Unlike traditional methods used for construct validation and hypothesis testing, the methodology permits the researcher to model the extent of random and systematic error in measures of theoretical concepts and to control these sources of error when testing substantive hypotheses. Comparisons made between the holistic construal and traditional procedures show how the latter can lead to inaccurate conclusions about the status of a theory's constructs, hypotheses, and measures.