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

Project: Research project

Description

Existing research on the development of organizational networks explains their development using mechanisms and factors like homophily, balance and exchange. Though this approach has advanced our understanding of the development of networks, if suffers from three limitations. First, it fails to account for specific activities of organizational members. Second, it lacks rich models of agency as it affects network development. Third, it overlooks the role of groups and other intermediate structures in network formation and change.
This project develops a theory of network structuring based on human agents influencing one another in group contexts. It will accomplish this by modeling network nodes as agents with their own individual motives and networking tendencies rather than treating actors as belonging to simple classes; modeling organizational activities as important influencers of networking behavior by agents; modeling the impacts of groups and work units on networks; and taking the temporal aspects of communication into account. We will encompass theory, data, and model by developing an integrative theory of network structuration, instantiating this theory in an agent-based model that simulates network dynamics, and evaluating this model and theory against unique data collected in a long-term study of a software development organization called the Software Factory (SF).
Our point of departure is existing work on organizational network structuration, which treats structuring processes as a black box. Corman and Scotts theory of network reticulation offers theoretical scaffolding capable of bringing in agency and the group context. They posit a recursive loop in which agents schemata of networks reflect their observations of situated communication behavior and in turn guide agents networking behavior. Agent behaviors operate within a tripartite set of modalities. Enactment describes the production and reproduction of system coding conventions in triggering events, and captures environmental inputs to the structuring process. Activation describes the production/reproduction of activity foci in organizational activities, and accounts for collective influence on behavior. Reticulation describes the production/reproduction of perceived network links in observable communication behavior.
These processes will be instantiated in an agent-based model in two phases. In Phase I, we will build a model with non-cognitive agents to use as a baseline. In Phase II this model will be supplemented with a SOAR cognitive model for decision making at the network nodes. To validate these models, we will test their output against a dataset collected from three years of observations at SF. Data include over 700 hours of time-synchronized audio recordings, from which we will extract observable communication data using advanced signal processing techniques. The dataset also includes weekly sociometric surveys completed by participants, bi-weekly interviews with participants, field notes from non-participant observation of SF activities, recordings of weekly staff meetings, and organizational activity reporting records. The data support measurement of perceived communication between organization members, observable communication between members, focused activity dynamics of the SF, and (d) triggering events from the environment. We will evaluate the theory and models in two ways. First, we will compare predictions of the agent-based models with observations of perceived and observable communication. Second, we will test the theory directly against empirical observations using Relational Event Modeling, to reveal potential flaws in the assumptions of the theory and/or improvements in the mechanics of the agent based models.

Description

Existing research on the development of organizational networks explains their development using mechanisms and factors like homophily, balance and exchange. Though this approach has advanced our understanding of the development of networks, if suffers from three limitations: It fails to account for specific activities of organizational members, it lacks rich models of agency as it affects network development, and it overlooks the role of groups and other intermediate structures in networkformation and change.
This project develops a theory of network structuring based on human agents influencing one another within group contexts. It will accomplish this by modeling network nodes as agents with their own individual motives and networking tendencies rather than treating actors as belonging to simple classes; modeling organizational activities as important influencers of networking behavior by agents; modeling the impacts of organizational structure on networks; and taking the temporal aspects of communication into account. We will encompass theory, data, and model by developing an integrative theory of network structuration, instantiating this theory in an agent-based model that simulates network dynamics, and evaluating this model and theory against unique data collected in a long-term study of a software development organization called the Software Factory (SF).
Our point of departure is existing work on organizational network structuration, which treats structuring processes as a black box.Corman and Scotts theory of network reticulation offers theoretical scaffolding capable of bringing in agency and the group context. They posit a recursive loop in which agents schemata of networks reflect their observations of situated communication behavior and in turn guide agents networking behavior. Agent behaviors operate within a tripartite set of modalities. Enactment describes the production and reproduction of system coding conventions in triggering events, and captures environmental inputs to the structuring process. Activation describes the production/reproduction of activity foci in organizational activities, and accounts for collective influence on behavior. Reticulation describes the production/reproduction of perceived network links in observable communication behavior.
These processes will be instantiated in an agent-based model in two phases. In Phase I, we will build a model with non-cognitive agents to use as a baseline. In Phase II this model will be supplemented with a SOAR cognitive agent model for decision making at the network nodes. To validate these models, we will test their output against a dataset collected from three years of daily observations at SF. Data include over 700 hours of time-synchronized audio recordings, from which we will extract observable communication data using advanced signal processing techniques. The dataset also includes weekly sociometric surveys completed by participants, bi-weekly interviews with participants, field notes from non-participant observation of SF activities, recordings of weekly staff meetings, and organizational activity reporting records. The data support measurement of perceived communication between organization members, observable communication between members, focused activity dynamics of the SF, and (d) triggering events from the environment. We will evaluate the theory and models in two ways. First, we will compare predictions of the agent-based models with observations of perceived and observable communication. Second, we will test the theory directly against empirical observations using
Relational Event Modeling, to reveal potential flaws in the assumptions of the theory and/or improvements in the mechanics of the agent based models.
Intellectual Merit. This project opens the black box of network structuring by building a sophisticated representation of organizational activities, and describing how these enable and constrain members communication choices. It incorporates individual human agency as an influence on network development, and allows for the influence of groups as intermediate structures in network formation and change.
Broader Impact. This project will suggest design principles to maximize organizations ability to cope with environmental demand, and diagnose organizational failures resulting from inability of the communication system to cope with dynamic activity load. The signal processing method has applications in emergency response and other applied contexts to diagnose dynamic networks and the emotional states of communicators.
StatusActive
Effective start/end date9/1/168/31/20

Funding

  • National Science Foundation (NSF): $205,060.00

Fingerprint

Empirical test
Collaborative research
Agent-based model
Communication
Factory
Software
Networking
Modeling
Structuration
Node
Organizational networks
Network development
Network dynamics
Black box
Homophily
Decision making
Organizational modeling
Enactment
Activation
Factors