Testing for social comparison in the field
encounters many difficulties (e.g., Charles Manski,
2000). For example, a positive correlation
between expectations about the mean behavior
of the reference group and one’s own behavior
is consistent with conditional cooperation, but
not conclusive, as causality is not clear. Behavior
may influence expectations, and not the
other way round. Only a few laboratory experiments
circumvent these problems and explicitly
test conditional cooperation (e.g., Urs
Fischbacher et al., 2001). These studies conclude