Agent-based simulations can yield predictions untestable from individual-level data alone

Evidence

By examining multiagent interactions over time, agent-based modeling can reveal important emergent effects that could not be predicted only on the basis of knowledge of individual-level processes. At the same time, the ability to independently manipulate the nature of the agents’ individual and dyadic motives and behavior permits direct tests of the roles of these processes in producing aggregate outcomes. Finally, the ABM environment permits the manipulation of social context-level variables that are difficult to systematically vary in laboratory experiments. The ability to examine multiagent interactions over time also is a unique feature of agent-based modeling that cannot be accomplished with standard behavioral laboratory methods. — Roderick M. Kramer, Geoffrey J. Leonardelli, Robert W. Livingston, Social Cognition, Social Identity, and…, 859 (#corroborates, via linked claim)

I was selecting only five individual males and ten females to found each generation. This is a small sample, and it meant that some shift in the traits of my population could have occurred by chance. If, at the end of my experiment, the males in my artificially selected population had longer horns than before, I couldn’t rule out the possibility that this was a spurious result due to chance. — Emlen, Douglas J, Animal Weapons_ The Evolution of Battle, 291 (#complicates, via linked claim)

Categorization

• XREF: connects to: Agents following identical optimal-distinctiveness rules can be simulated to test group formation; Agent-based modeling reveals emergent group effects unpredictable from individual-level knowledge — Roderick M. Kramer, Geoffrey J. Leonardelli, Robert W. Livingston, Social Cognition, Social Identity, and…, 859; Agent-based models reveal aggregate consequences no individual agent intends or foresees. — Roderick M. Kramer, Geoffrey J. Leonardelli, Robert W. Livingston, Social Cognition, Social Identity, and…, 821 (+4 more)

Broader Connections

• Agent-based modeling reveals emergent group effects unpredictable from individual-level knowledge #corroborates (0.92) — Roderick M. Kramer, Geoffrey J. Leonardelli, Robert W. Livingston, Social Cognition, Social Identity, and…, 859
• Small founder samples make artificial selection results vulnerable to chance #complicates (0.72) — Emlen, Douglas J, Animal Weapons_ The Evolution of Battle, 291