Segmentation approaches that differentiate consumption frequency from sensory preference

People can eat a food without having a strong preference for it, and people can prefer a food without eating it. Given this seeming disconnect between attitude and behavior, which type of measure or segment can best be used to profile or identify loyal consumer segments of a food, such as soy? This research compares a usage-based method (heavy-light-nonusers) with a new attitude-based method (seeker-neutral-avoider), and finds that the attitude-based method differentiates purchase-related intentions better than the usage-based method. Implications for profiling consumer taste patterns and consumer segments are provided.