We compare the experimental results of three stag-hunt games. In contrast
to Battalio et al. (2001), our design keeps the riskiness ratio of the
payoff-dominant and the risk-dominant strategies at a constant level as the
optimisation premium is increased. We define the riskiness ratio as the relative
payoff range of the two strategies. We find that decreasing the riskiness
ratio while keeping the optimization premium constant increases sharply the
frequency of the risk-dominant strategy. On the other hand an increase of
the optimization premium with a constant riskiness ratio has no effect on the
choice frequencies. Finally, we confirm the dynamic properties found by Battalio
et al. that increasing the optimization premium favours best-response
and sensitivity to the history of play.