Collaborative Research: The reticulation/activation nexus in organizations: An agent-based model and empirical test using unique data

点击翻译按钮获取中文摘要

Understanding the development of communication networks in organizational contexts is a fundamental research problem in the science of organizations. An explosion of research in this area over the last 30 years has produced an understanding of how networks operate in equilibrium conditions based on dyadic features like balance, homophily, and exchange. Yet we still do not understand the effects of individual agency and contextual factors like organizational activity. The purpose of this project is to develop and test a theory of network structuring and development based on the activities of human agents influencing one another within group contexts. This project makes four contributions: (1) modeling network nodes as agents with their own motives and networking tendencies, (2) describing the influence of focused activities on network development, (3) accounting for the influence of organizational structures and work units on network development, and (4) explaining network dynamics as the product of a loop of structure and behavior which influences one another over time. Understanding these phenomena will enable better design of a wide range of organizations, as well as better prediction and mitigation of organizational failure.The theoretical context of this work is Network Reticulation Theory (NRT). This theory draws-together elements of structuration theory and activity focus theory to explain how perceived network relationships are activated by triggers in an organizational system, for example how a deadline in a software engineering firm activates certain elements of an organizational network to produce a hackathon, potentially strengthening the constituent network relationships in the process. We study this process using a unique dataset comprising ubiquitous observation of a software engineering shop, from which data about perceived relationships, observed communication, task activities, and member perceptions are recorded for 79 employees over a period of three years. To evaluate the theory, we employ three complimentary methods: (1) A direct test on the empirical data using statistical relational event modeling; (2) a multi-agent social-network simulation that faithfully models NRT by capturing activation, reticulation, and enactment processes; and (3) a cognitive multi-agent social-network simulation that adds architecture for decision making at the network nodes. The project also utilizes a novel method for identifying network connections from audio recordings that offers a useful supplement to sociometric badges.