Abstract
Individual factors by no means completely account for
individual popularity within a group structure. To begin to
explain the majority of the variance, we must investigate the
hypothesis that popularity is strongly influenced by the
dynamics of group interactions. Here, we present a
computational model of peer interaction that allows us to
investigate the influence of different distributed factors. In
constructing the model, we discovered that certain elements
are vital for the simulation to produce data that matches the
observed patterns in real social groups. We found that the
internal representation of how much agents like each other
must be discrete, that judgements should be made relative to
behavioural expectations, and that models do not require
variation in the initial state of the agents to produce realistic
individual differences in popularity. Our result is a set of
models with psychologically realistic attributes. When
simulated, these models result in popularity data that cannot
be reliably distinguished from real life data. Since these
models capture the essential dynamics of the social group
interaction, they can form the basis for understanding how
interaction within the group influences individuals to become
popular or rejected.
Users
Please
log in to take part in the discussion (add own reviews or comments).