A COMPARISON OF THE EFFECTIVENESS OF HEDONIC SCALES AND END‐ANCHOR COMPRESSION EFFECTS

Three experiments were conducted to compare the relative performance of hedonic scaling methods, including the labeled affective magnitude (LAM) scale. In the first study, three versions of the LAM were used to evaluate 20 phrases that described diverse sensory experiences. One scale was anchored to “greatest imaginable like/dislike for any experience” and another used the “greatest imaginable like” phrase of the LAM but with the interior phrases repositioned relative to “any experience.” The scale anchored to “any experience” showed a smaller range of scale usage and lower statistical differentiation, relative to the LAM scale, with the repositioned scale intermediate. Two further experiments compared the LAM to the nine-point hedonic scale, an 11-point category scale using the LAM phrases, and to a three-label line scale, a simplified version of the LAM with only the end phrases and the neutral center-point phrase. All scales showed similar differentiation of juices in the second study and sensory experience phrases in the third. A modest advantage for the LAM scale in the second experiment did not extend to the third study. Researchers should be careful in the choice of high end anchors for hedonic scales, as a compressed range of scale usage may result in lower product differentiation. PRACTICAL APPLICATIONS Hedonic scales for food acceptability are widely used in new product development for consumer testing and in food preference surveys. A desired goal of efficient sensory evaluation testing is the ability of tests to differentiate samples on the basis of scale data, in this case scales commonly used for food acceptability and preference testing. Scales which are able to differentiate products more effectively are less likely to lead to Type II error in experimentation, in which true differences between products are not detected. Such errors can lead to lost opportunities for product improvements or to enhanced chances for taking undetected risks in the case of false parity conclusions.