For the visual evaluation of X-ray images, we use a sensory test, Thurstone's method of paired comparisons, in which differences between samples are detected with high sensitivity by making a Yes or No judgement. However, with this method, large differences between paired samples can't be transformed into an interval scale because the evaluation for the pair becomes missing value. In an attempt to overcome this problem, we examined Scheffe's method of paired comparisons for the subjective evaluation of X-ray images, in which the scored data are examined by analysis of variance. Scheeffe's method was concluded to be useful in overcoming the above-mentioned problem and even in analyses of variance among observations.
