Influences of the absence of random assignment of bids on estimating willingness to pay using a discrete‐choice question

This study examined the influences of the interaction between a bid and the respondent's characteristics due to insufficient random assignment of bids on the estimation of willingness to pay (WTP) using data from a discrete‐choice question. A contingent valuation survey of 152 examinees undergoing X‐ray testing for gastric cancer screening was conducted, and the median and mean WTP for the serum pepsinogen test were estimated using a logistic regression model to which the interaction terms between the bid and the respondent's characteristics, which included gender, age, annual income, frequency of prior use of a gastric cancer screening program, and perceived health, were added. There were remarkable differences in the estimated WTP according to whether the interaction term of annual income, to which the bids had failed to be assigned randomly and which had been positively correlated with the bid, was added in the model. It is suggested that it may be necessary to check if the bids were randomly assigned to the respondent's characteristics and, when correlations with the bid are found, to adjust their interaction effects. Copyright © 2004 John Wiley & Sons, Ltd.