American Choice Modeling

Agent Based Models

Solutions for decentralized problems

Much like any competitive market for goods and/or services, an Agent-Based Model lacks central control. While there are markets forces that do affect market outcomes (i.e. invisible hand/invisible foot), no single person or authority has complete domain control over what is deemed a free, but regulated, market. In view of this, rather than attempting to gauge market participant behavior through a top-down approach via surveys, interviews or other traditional market research techniques, Agent-Based Models used in marketing create an artificial replica of the market from the ground up.

A pharmaceutical company, for example, may want to know how patients and doctors will react to a new treatment being considered for introduction into the market. In this situation the treatment is only going to be as successful as the compliance of the doctor and the patient to the treatment plan. The doctor must prescribe the treatment properly and provide appropriate follow up with the patient. The patient must visit the doctor in the first place, carry through with the prescription and take the recommended dosing. You can begin to imagine how the different actions taken by each party will affect the ultimate success or failure of the new treatment in the marketplace. Add even more patients, doctors and market alternatives, and a whole web of complex actions and behaviors is created.

Agent Based Models

An Agent-Based Model can be developed with as little or as much information as deemed practical. For situations where there is little opportunity to conduct robust, large-scale quantitative surveying, Agent-Based Modeling and simulation offer an ideal solution for visualizing hypothesized behaviors and outcomes with a rich level of detail. Complex behaviors, unexpected outcomes and hypothesized scenarios can all be explored visually and interactively through simulation that can be made intuitive and accessible to almost anyone. Resulting simulation data can, however, be analyzed more in-depth using a variety of standard statistical techniques to further establish reliability.