Evidence of range bias in contingent valuation payment scales

The payment scale format has been widely used in willingness-to-pay studies in health care. Concerns have been expressed that the format is, in theory, prone to range bias, although this proposition has not been tested directly. We report the findings of a contingent valuation questionnaire study of colorectal cancer screening, wherein different subjects were provided with payment scales of two different lengths. Whilst the long-scale instrument included scale values up to pound 1000, the short-scale version extended only to pound 100. After controlling for inter-sample differences in, for example, income, education and health behaviour, it emerged that the long-scale instrument produced a mean willingness to pay more than 30% higher than that resulting from the short-scale version. We believe our findings to be strongly supportive of the likelihood of range bias in payment-scale instruments, with important consequences for the estimation of both average valuation and consumer surplus.