In a split-sample contingent valuation study of willingness to pay (WT
P) for food safety improvements, the dichotomous choice (DC) elicitati
on method consistently generated much larger estimates of WTP than did
a continuous method. Little or none of these differences was due to b
ias introduced by the statistical techniques used with the DC data. Mo
st or all of the difference was due to differences in respondent behav
ior. In addition, the continuous WTP responses showed a significant sc
ope effect while the DC responses did not. The observed difference in
behavior may be attributable in part to yea-saying by DC respondents.