What determines the differences in perceptual rating of dysphonia between experienced raters?

Although the perceptual GRBAS scale for pathological voice quality has been found to be sufficiently reliable in clinical practice, even experienced raters disagree to some extent, and the degree of disagreement depends on the perceptual characteristics. We looked for a possible link between the degree of disagreement (65 voices; 2 experienced raters) and objective acoustical measurements. No significant correlation appeared between any acoustical parameter and the degree of disagreement for G. By contrast, the difference in perceptual rating of R was related to the amount of shimmer, and the difference in perceptual rating of B was related to some extent to the amount of jitter. Thus the presence of a strong breathy component in a pathological voice disturbs the rating of the rough component, and reciprocally.