Improving Inferences About Preferences In Choice Modeling