A stated choice model is used to estimate wetland mitigation preferences. In a split sample mail survey, a main effects design is compared to a randomized design. Although randomized designs estimate main effects less efficiently, several policy relevant interactions were found to be significant, suggesting some merits of randomized designs.