This result is consistent with the assumption that players neglect the strategic aspects of the SLPG games.
In the SLPG conditions (PGS and PGP) subjects appear to be overconfident in comparison with the naive Bayesian model.
So we find that subjects are overconfident in the SLPG games, but not in the related game against nature.
Given that people tend to perceive an SLPG game as a decision theoretic problem instead of as a strategic game, one can move on to investigate the way in which subjects handle this problem.
A remarkable result is the difference in updating speed between the condition without interaction (NT) and the SLPG conditions with interaction (PGS and PGP).
This article focused on the expectation formation process in various SLPG games.
The evidence suggests strongly that subjects neglect the strategic aspect of the SLPG games.
The SLPG game in this paper has multiple equilibria.
Note, however, that convergence to a Nash equilibrium was not yet observed in the 20 periods of the experimental SLPG games.
In the various SLPG games the introduction of a recency/primacy parameter in the model did not increase the likelihood of the data significantly.
The behavior in SLPG games should be seen as the result of a combination of expectation formation on one hand and individuals' decision rules on the other.
Optimal behavior in an SLPG game is to contribute if and only if the subjective probability that the own contribution is critical for the provision of the public good (p(s - 1)) is sufficiently high, that is, if and only if p(s - 1) [greater than] c/(f - c).