Structural congruence of tasks and organizations

This dissertation presents a novel analytical method for designing optimal distributed architectures for decisionmakers (DMs) acting in uncertain, complex decision environments. Key issues addressed are: (i) the optimal information access structure (who should measure what); (ii) the optimal data aggregation procedure (who should compute what); (iii) the optimal communication structure among DMs (who should communicate what and with whom); and (iv) the necessary minimum level of internal communication; in short, the optimal organizational design.^ Extending the single event - fixed organizational structure formulation of the traditional distributed hypothesis testing paradigm, the internal structure of a task-organization system is described via a joint influence diagram. The optimal decentralized solution is obtained in the form of a process tree that contains the set of elementary subtasks with well-defined precedence constraints among them. Then, a structurally congruent organizational design, one that achieves centralized decision accuracy with minimal level of internal communication, is constructed by implementing the process tree. The method also provides the optimal subtask-DM assignment and a corresponding capacity provision for the communication channels in a candidate organization. Finally, results of a computer-mediated laboratory experiment to validate the model\u27s normative predictions with human teams are discussed.