Finding Optimal Targets for Change Agents: A Computer Simulation of Innovation Diffusion

We introduce a diffusion of innovation model based on a network threshold approach. Realistic network and threshold data were gathered regarding the diffusion of new software tools within part of a large organization. Novel model features are a second threshold for innovation rejection and a memory that allows actors to take trends into account. Computer simulations produce expected outcomes, such as the S-shaped diffusion curve, but also diffusion breakdown and oscillations. We define and compute the quality of change agent targets in terms of the impact targeted actors have on the diffusion process. Our simulations reveal considerable variance in the quality of actors as change agent targets. Certain actors can be singled out as especially important to the diffusion process. Small changes in the distribution of thresholds and changes in some parameters, such as the sensitivity for trends, lead to significant changes in the target quality measure. To illustrate these interdependencies we outline how the impact of an actor targeted by a change agent spreads through the network. We thus can explain why a good change agent target does not necessarily need to be an opinion leader. Simulations comparing the effectiveness of randomly selected targets versus a group of good change agent targets indicate that the selection of good targets can accelerate innovation diffusion.