For quite a while, stated preferences have been a major tool to measure consumer preferences for new products and services. Revealed preference methods, in particular experimental economics, have gained popularity recently because they have been shown to be more incentive compatible, and therefore more accurate. However, this advantage comes at the expense of higher survey costs. In the developing countries with limited funding for research, it is important to determine whether the extra cost can be justified by the extra gain in accuracy. A survey of 100 farmers was carried out in Western Kenya to determine consumer preference for yellow maize using the contingent valuation, choice experiments and experimental auction methods. Experimental auctions produced the most realistic results for mean willingness to pay. They are also the most accurate at all budget levels, but also the most expensive. Considering their accuracy and realistic results, we conclude that they should be the recommended method in measuring consumer preference in developing countries, since the extra cost is more than recovered by the gain in accuracy.