The longstanding dispute over the accuracy of stated preference methods in eliciting the true valuations of individuals has stimulated interest in analyzing preference inconsistencies between revealed and stated preference mechanisms. This paper uses preference orderings to provide a more robust comparison between revealed and stated preferences and assess the validity of the latter. This is done by comparing an incentive compatible auction experiment (recoded as implied ranks) with a ranking procedure. Partial ranking models are constructed to examine consumer preferences under the two valuation mechanisms for the most preferred and the least preferred alternatives in order to provide a more detailed analysis. The stability and symmetry of parameters was tested and systematic differences between the models were analyzed in order to measure the extent of preference inconsistencies between the auction exercise and ranking procedure. Furthermore, the predictive power of the models was calculated to evaluate the relative reliability of each mechanism. The results provide robust evidence that individuals often employ different behavioral rules under the two elicitation mechanisms, especially when expressing mild feelings about certain alternatives. Compared to the more accurate auctions mechanism, the ranking exercise seems to perform fairly well only when eliciting preferences over the best ranked alternative.