Dear all,

I have data from a choice experiment with 120 individuals who answered from 6 sets of situations each. Each set of the situation requires from individuals to choose between two options and "opt-out " option. In each option are seven attributes, besides the distance (converted in cost).

I’m interested into obtaining the willingness to pay for each attribute. To do so, I have used the command mixlogit (available in Stata Journal Volume 7, Number 3) to fit mixed logit models. In my model, I used the Cost and Cost squared ( their coefficients are negative and positive, respectively).

The willingness-to-pay for each attribute has traditionally been calculated by dividing the attribute coefficient of the selected attribute by the coefficient associated with a price (or a cost). However, given that the utility is nonlinear with respect to price, I cannot use the traditional expression to calculate WTP values because the slope is not constant.

I calculated the Price (cost) Threshold by: CoefficientCost / (2*CoeffcientCostSquared)

My question is: How can I calculate the WTP for each attribute, given this nonlinear Function??


Thanks in advance


Fathallah Kerouaz.