L'Herrou, Todd Alan

Three studies use interaction in groups to test macro-level phenomena of the type predicted by computer simulations of dynamic social impact theory, an interactive, reciprocal, and recursive extension of Latane's (1981) theory of social impact, which specifies how individuals are affected by their social environment. A key prediction of dynamic social impact theory is that such phenomena as clustering and polarization of attitudes will emerge from social interactions among spatially distributed people. Study One, using a group of 24 people corresponding with their neighbors about an attitudinal topic, found both substantial polarization and a suggestive tendency toward clustering. Studies Two and Three, using 192 people organized into 48 24-person groups, provide strong empirical support for the emergence of clustering and incomplete polarization, and extend the findings to two alternative spatial structures. As predicted, a control geometry, involving random connections without any spatial organization, failed to produce clustering. Directions for future research are discussed.