Optimal Design Of Bid And Sample Size In Double Bounded Dichotomous Choice CVM

The contingent valuation method(CVM) is a stated preference method and it is widely used to evaluate the ecological environment and resources. The double bounded dichotomous choice CVM is the most popular and effective in the CVM thorey. Questionnaire design and dissemination is the key to the success of the CVM study, initial bid amount and sample size in questionnaire will impact assessment results of CVM. The number of initial bids is always obtained by pre-test, while the sample size is always got by field research or online survey. Less number of initial bids and sample size will influence the validity and reliability of WTP. On the other hand, more number of initial bids and sample size will increase the difficulties and costs of investigate and survey. There is not enough objective basis in the choice of initial bid numbers and sample size when the questionnaire designed, and the study on the effect of different initial bid numbers and sample sizes on the mean and median of WTP. Dichotomous choice issue is the core part in questionnaire design, and the bid of questionnaire is always obtained by pre-test and experice of designer. In questionnaire design including low bid value may will increase the deviation, and if questionnaire contains a high bid value will reduce the hypothetical bias. While double bounded dichotomous choice CVM questionnaire almost all existing contains smaller bid amount.This paper summarizes the double bounded dichotomous choice CVM model on the basis of the existing research, and analysis of the survey data from the parameter model and non-parameter model. Based on existing research, we can divide the double bounded dichotomous choice contingent valuation method parametric into the WTP function model and random utility models.The non-parametric methods which based on Turnbull method is called second nonparametric method. This paper studies on the dynamical effect of different initial bid content and sample sizes on the mean and median of WTP by using the willingness to pay(WTP) function model and posteriori Monte Carlo simulation, and using the quartile to design bids. We use Weibull model and Log-Logistic model, which base on two different source of the original datum as the distribution of WTP.The data of Weibull model is from the value evaluation of wetlands of the Sanjiang Plain, while the data of Log-Logistic model is from the value evaluation of Shanghai Zhangjiabang River. We research on the dynamical effect of different initial bid content and sample sizes for WTP, by discussing the value and astringency of mean squared error(MSE) on WTP. Following the steps of Monte Carlo simulation, we finally get minimum quantity of nitial bid content and sample sizes by the visualization of the simulated results. According to the problem of double bounded dichotomous choice CVM questionnaire contains low bids will produce the deviation, we put forward a new bid design based on the existing optimal design method, which including D-optimal design, C-optimal design, Fiducial method design. Taking wetlands in the Sanjiang plain as the evaluation object, assuming WTP obeys the Logistic distribution, we campare two bid scheme which one including low bids another not, and select the same sample size and confidence level.It is found that initial bid content and sample sizes in the two different distribution of WTP almost have the same influence on mean and median of WTP. When the initial bid content and sample size respectively exceeds 5 and 500, the influence of mean WTP is almost the same. And based on the result we can exploringly provide a constructive suggestion that at least 5 initial bids and 500 samples are needed in order to ensure the validity and reliability of the mean of WTP indicated by Monte Carlo simulation, while for the median of WTP at least 3 initial bids and 300 samples are needed. Based on containing low bid scheme design, the median of WTP is 164.679, the 95% confidence interval is [133.534, 195.824], and the width of confidence interval is 62.29, AIC is 1.81. While for the not including low bid scheme, the median of WTP is 187.635, the 95% confidence interval is [162.279, 212.991], and the width of confidence interval is 50.712,AIC is 1.24. Acrrowding the results, the not including low bid scheme have a narrower confidence intervals, and its AIC is smaller. Therefore, he not including low bid scheme collect more effective data.We enriched double bounded dichotomous choice CVM model and ensure the validity and reliability of the mean and median of WTP. And the evaluation results provide theoretical basis for the formulation of relevant policies and suggestions for the relevant decision-making departments.These results provide quantitative basis for questionnaire design of CVM and improve the validity and reliability of WTP.