Bid design in dichotomous-choice questions is an issue of considerable
concern and debate. This paper investigates the effect of bid structu
res on welfare estimates using two pretest distributions (from open-en
ded and dichotomous-choice questions) and three bid structures (two-bi
d, five-bid, and multi-bid designs). Both Monte Carlo simulations and
responses from a field experiment are used. Results support the growin
g evidence that ''yea saying'' occurs and the problem becomes worse wh
en bids are clustered at discrete bid levels in the upper tail of the
distribution. The systematic effect of bids on responses to dichotomou
s-choice questions reduces the effectiveness of optimal bid designs.