However, in these studies, there is scope for attracting donors
with poor characteristics over the long run. In contrast, in our treatment, there is much less
scope for attracting new low-quality" donors, since all our subjects are previous donors, and
no incentives were used prior to the experiment. Further, the incentives are much smaller than
in the settings analyzed in these other studies. There may be more subtle selection eects due
to a dierential attractiveness of the lottery ticket oered as the incentive. For example, donors
who are more risk-seeking in many areas of life have a higher willingness to pay for a lottery