The importance of stressing the realism or policy relevance of a conti
ngent valuation (CV) survey is commonly recognized by CV researchers.
This paper focuses on the relationship between the probability that a
subject's response will result in ''real'' consequences and the subjec
t's accurate reporting of willingness to pay. Results from experiments
support the argument that efforts to make contingent valuation survey
s ''real'' are important. However, our, results also suggest that the
connection between the survey and the likelihood of a payment must be
rather powerful before surveys generate accurate reporting of willingn
ess to pay for a public good.