This paper compares protest rates and willingness to pay (WTP) using a payment card versus single and double bounded voter referendum contingent valuation question formats. Using a chi-square test, the payment card had a significantly higher protest rate (6.7 per cent) than the voter referendum question format (2.2 per cent). The median WTP of the single bounded and double bounded referendum format exceeds the payment card by a factor of nine and seven, respectively. The median WTP from the referendum formats represent about 8 per cent of income, while the payment card results represents about 1 per cent of income. These large differences in WTP between question formats are double what have been found in past studies. We believe this result may be due to excessive yea saying at high bid amounts in the dichotomous choice question formats. This behavior may arise in our case study in rural China because citizens have not had a long history of open elections or voting on tax referenda.