Prohabilistic choice models which generalize the Bradley-Terry-Luce (BTL) model (Luce, 1959) and the elimination-by- aspects (EBA) model (Tversky, 1972) are presented. These models account for the effect of presentation order within a pair of stimuli, and thereby allow for quantifying judgmental biases for the first or second presentation interval. Data were collected in an experiment where 39 subjects made pairwise choices between short musical excerpts reproduced in eight different audio formats (mono, stereo and multichannel). Choice criteria were overall preference, and eight specific (spatial and timbral) auditory attributes. The results indicate that biases are well accounted for by the generalized modeling, and that for seven of the nine attributes, including preference, they significantly favored the second presentation interval.