ABSTRACT
In this chapter, the author summarizes what he knows about difference testing at Pepsi: how it was done, who it was done with, and what should have been done. A panelist may state which of two products has more of some carefully defined quality; when that quality is hedonic, this becomes a paired-preference test. Test procedures in signal detection are simple and straightforward, similar to those in other forms of testing. Most quality control (QC) testing uses an informal version of this test. Comparison tests are usually divided into attribute and hedonic tests—that is, tests which determine whether a product has more of some attribute than another and those which determine whether a product is liked more than another. Paired comparison attribute testing is inherently the most sensitive difference procedure available. A set of screening tests using moderately easy differences should be passed by any prospective panelist before that panelist is used.
