The manner in which “no preference” responses have been treated in the E1958 Standard Guide for Sensory Claim Substantiation has evolved between 1998 and the current (2016) guide. For example, the 1998 guide proposes that an interviewer present consumers with a forced-choice preference question, and accept a “no preference” response from consumers who indicate that they do not have a preference.
By contrast, the 2016 guide advocates that the consumer be presented with a ballot in which a “no preference” option is provided. A forced-choice preference question is easily modelled using a binomial distribution, but what is the meaning of a preference response in a ballot without a “no preference” option for those consumers who perhaps have no preference? And what is the meaning of a “no preference” response in the context of claims related to superiority, unsurpassed, and equivalence? In each type of claim, current and historical guides have redistributed the trinomial data (prefer A, prefer B, no preference) to form binomial data that enables a binomial analysis. The assumptions that underlie these redistribution strategies are discussed.
Castura, J. C. (2018). Historical developments in the ASTM Guide for Sensory Claim Substantiation. In: Symposium 1: Statistical Issues in Claims Support (Ugurlayan, A., Castura, J. C., Basehoar, A., Ennis, D. M.). Current Topics in Sensory and Consumer Science Symposia. 8 May. Williamsburg, VA, USA. (Invited.)