This paper introduces similarity among strategies in the payoff assessment
model of choice (Sarin and Vahid (1999, GEB)). The assessments of strategies
that are more similar to the chosen strategy are updated more similarly
to the chosen strategy. We use this model to explain a recent experiment.
The coordination game repeatedly played by the experimental subjects had
two symmetric, efficient and strict stage game Nash equilibria. In the
experiment, the subjects always converged to play one of these equilibria,
and converged to this equilibrium remarkably fast. The model we propose
converges to choose the same equilibrium, and does so in roughly the same
number of repetitions. Statistical tests cannot distinguish between the
choice distributions generated by the model and the observed choice distributions
in almost every period.
Keywords: Econometric analysis of experimental data,
Adaptive learning, Similarity, Coordination games, Payoff assessments