ABSTRACT
As discussed in Sects. 2.4 and 3.4, although paired comparisons form the basis for determination of JNDs, they have a limited dynamic range, and so cannot be used directly to measure sample differences or to characterize quality ranges that exceed a few JNDs. To characterize the distribution of quality produced by practical imaging systems, much wider ranges of quality must be quantified. As discussed in the next chapter, an effective approach to this problem is to establish a standard numerical rating scale anchored to physical samples, calibrate the scale once through selected paired comparisons of the standard samples, and subsequently rate samples directly against the standard scale (how this might be done is described in Chs. 7 and 8). The current chapter describes the properties of two types of numerical rating scales that are particularly useful for this purpose, namely, interval and ratio scales.
