Measurement issues with iterated, continuous/interval contingent valuation data

A useful contingent valuation technique for eliciting estimates of welfare during telephone interviews is the iterated willingness-to-pay question. However, this elicitation method raises at least two important measurement issues: choice of estimation technique and starting-point bias. We acknowledge the potential for starting-point bias and: (1) use a randomly chosen starting point which covers the possible range of willingness to pay; and (2) cease the iteration procedure after a range of willingness to pay is determined. With the resulting continuous/interval data we compare WTP estimates using different econometric techniques (least squares, Tobit and interval data regressions) in order to estimate bias-free WTP. We test for starting-point bias and suggest a correction procedure. We conclude by recommending that future telephone applications of the contingent valuation method should consider the use of the iterated, interval willingness-to-pay question.