We conducted in-the-field choice experiments in China to investigate farmers’ willingness to pay for crop insurance and to determine how objective and subjective beliefs affect Willingness to Pay (WTP). We deploy three variants of the choice experiment using a priming mechanism on objective and subjective beliefs plus a control. We find that the cuing frame matters in that there are distinct differences in WTP within five attributes and across variants. In terms of policy practicality our results suggest that farmers’ frame of reference can affect insurance demand. We then examine whether WTP choices were determined by affective choice, dual processes and/or expected utility. In a follow-on survey we asked farmers to state minimum, typical or most likely, and maximum yields from history (objective risks) and the next-harvest future (subjective risks). These were used to construct objective and subjective probability distributions for each respondent farmers using the beta-PERT distribution. The first approach interacts choices with mean, standard deviation, and skewness about a generalized Taylor’s series expansion about utility. The second interacts the certainty equivalents drawn from a logarithmic utility function. The third involves a scoring function based on the subjective-objective spread of PERT cumulative distribution functions. Results were mixed. We find that the direct expected utility approach was unsatisfactory because of over-fitting. However, with strong multicollinearity between primal choices and the certainty equivalent measures we conclude evidence of a utility -centric dual process model. However, when the dual-process analytical component was based on the cumulative distribution scoring rule, the dual process hypothesis is weakened considerably.