Note: while d0 values can also be estimated using a preference test, as shown in Table 3, their meaning will be different from d0 values obtained from difference tests in the presence of preference segmentation. For instance, even if a large difference exists (e.g., d0 = 2.0) and is picked up by a difference test, a preference test could result in a smaller d0 value being measured and no significant preference. This would be driven by preference segmentation in the population. An extreme situation will be where 50% of the pop- ulation prefers A, 50% of the population prefers B, i.e., preference d0 = 0, but products A and B can be clearly sensorily different. Gen- erally, one would expect a preference d0 value to be equivalent to or lower than, but not greater than, a d0 between the same products obtained in a discrimination test.