Paired comparisons in visual perception studies using small sample sizes

The paired comparisons method appears to be a promising method in the field of display research. As experimental designs in this field are generally performed using a rather small sample of subjects, general methods described in literature are reconsidered resulting in a specialized analysis for visual perception experiments on displays. This overview describes this modified statistical method which deals with: (1) modelling and testing the significance of z-score differences between the ranked items, (2) testing the effect of an additional factor on z-scores, and (3) corrections of extreme fractions occurring more easily when using small sample sizes.