11.5 percentage points.4 This effect is more than
twice the behavioral change actually occurring
due to conditional cooperation. The correlation
between elicited expectations and behavior therefore
greatly overestimates the effect of conditional
cooperation. This can be explained by a
false consensus effect: Individuals’ preferences
regarding contribution may influence expectations
about the pro-social behavior of others.
Panel (C) of Table 2 controls for individual
heterogeneity by incorporating the coefficient
of past behavior in the probit model. The marginal
effect of a 1-percentage-point change in
expectations is 0.3. A change in expectations
from 46 percent to 64 percent corresponds to a
change in the probability of contributing by
around 5.3 percentage points. This effect is more
in line with the behavioral change due to induced
beliefs, because the coefficient of past behavior
captures part of the false consensus effect.