Dealing with Complexity in Information Systems Modeling: Development and Empirical Validation of a Method for Representing Large Data Models

One of the most serious practical and theoretical limitations of the entity-relationship (E-R) model is its inability to cope with complexity. Once E-R models exceed a certain threshold of size, they become difficult to understand, document and maintain. This paper describes the development and empirical validation of a method for representing large E-R models called leveled data modeling (LDM). A combination of research methods were used to validate the method. Action research was first used to test and refine the method in a real-world setting. Eight action research studies were conducted in eight different organizations. Once the method had become stable, two laboratory experiments were conducted to evaluate its effectiveness compared to the standard E-R model and methods previously proposed in the literature. Finally, a field experiment was conducted using experienced practitioners to evaluate the likelihood of the method being accepted in practice. The resulting method defines a general approach for managing complexity which could be applied to any information systems modeling technique. The research findings thus have general implications for developing more effective IS design techniques. Another contribution of the paper is that it illustrates a systematic, multimethod approach to empirically validating an IS design method.