Improving Agent-Based Models Of Diffusion Effects

We argue that agent-based modeling, as currently practiced, suffers from two deficiencies. first, the assumptions are often unverified and, second, these assumptions may be insufficient. Deficiencies in this new method may be overlooked because researchers are affected by the heuristics and biases in judgment revealed by Kahneman (2011). Building aggregate effects from individual behavior will be assisted when more is known about the influence mechanisms that operate in the social network. we give examples of model deficiencies, evidence on assumptions that is now available, and list influence mechanisms that must be studied if this type of modeling is to advance.