| Abstract: | This paper presents a mathematical model to study the joint impact of organizational design and of the task environment on the decision performance of hierarchical organizations with limited internal communication. The problem context is a special class of distributed situation assessment problems, where possible patterns of binary variables are to be classified on the basis of partial and noise-corrupted information. Structural properties of tasks and organizations are described using a graph formalism, and optimal decision strategies at all decision makers are determined. Organizational expertise is characterized in the form of a Team Relative Operating Characteristic (TROC) curve, thereby replacing the organization by an equivalent single decision maker. Implementing the model, issues of task decomposition and the process of matching organizations with tasks are discussed. |
