The researchers developed an agent-based simulation model of a social process, the Integrated Disability Evaluation System (IDES), that replicates every step of the system and simulates the associated human actions. Analysis of the model outputs shows that the performance metrics of individual agents in the social process simulation are similar to their real-world counterparts. The success of this agent-based social process simulation model allows for increased confidence in the predictive accuracy of what-if analysis conducted on human processes. In addition, process changes may be modeled to inform policy recommendations.
Agent-based simulation was mainly chosen because of the massive quantity of individual decisions required in the IDES.
The predictions made by the agent-based social process simulation model were similar to key IDES performance metrics, as indicated by a 96% mean timeliness accuracy and a 90% median timeliness accuracy. Additionally, the model’s aggregate timeliness distribution overlapped historical values by 88%.
Predicted and historical timeliness distributions