Every discrete choice experiment needs one, but the impacts of a statistical design on the results are still not well understood. Comparative studies have found that efficient designs outperform especially orthogonal designs. What has been little studied is whether efficient designs come at a cost. Thus, we compare four designs with respect to their effect on outcomes, especially willingness to pay (WTP) estimates and predicted market shares. Regarding statistical efficiency all designs perform rather well but the WTP estimates differ significantly on several occasions. This indicates that results are not neutral to the design, even when a high number of observations are available.