Repeatability and comparison of visual analogue and numerical rating scales in the assessment of visual quality

The repeatabilities of both the visual analogue scale (VAS) and numerical rating scale (NRS) as methods of assessing visual quality (VQ) were estimated by obtaining three replicated responses from each of a group of nine normal subjects who viewed a standardized scene under conditions of controlled refractive blur. In both cases, repeatability was highest in the absence of blur but fell markedly as increasing amounts of visual decrement were induced. Coefficients of reproducibility ranged from 4.1 to 36.0 for VASs; and from 1.2 to 34.9 for NRSs. Except where VQ was high, both types showed generally poor repeatability. When the data were used to compare the two methods, no consistent bias was found and agreement was good for high VQ levels. The NRS offers a reasonable alternative to the VAS as a method of assessing subjective VQ.