EXPECTATIONS AND STABILITY OF PREFERENCE CHOICE

When identical milk samples are presented, only 30% of participants respond with a “no preference” rating. Stability of the “no preference” rating was studied under a variety of conditions, having consumer panelists rate both identical and different pairs of milk samples with varying fat content. The proportion of participants choosing the “no preference” option, when the samples in the pair were identical, was largely consistent despite manipulation of pretest conditions and changes in test questions and answer formats. However, when the milk preference test was preceded by a same/different test, and those responding “same” were assumed to have no preference, the percent of “no preference” was two to three times larger for identical test samples (60–69%). Thus by branching the question, the “false preference” choice for identical milks was lowered. Among those responding “different” to identical milks, the false‐alarm rate increased to 91%, suggesting that perception of (spurious or momentary) differences is driving at least part of the preference choice.