Lecture 3

Reliability coefficient of >0.8 Kendall’s coefficient of concordance of at least 0.7 Agreement >0.8

• Ratings that are in the MIDDLE of a scale are less reliable than those at the extremes of a scale • ORDER of presentation: voices were rated as more severe if presented later • The type and resolution of the SCALE • The CONTEXT in which the voices are heard - i.e. a severe voice will be rated more severely if heard amongst normal voices • ANCHORING provides a more reliable rating • PATHOLOGICAL voices have a lower level of agreement than normal voices (in terms of severity) • EXPERIENCE of the judge - different internal standards

Grade Roughness Breathiness Asthenia (weakness) Strain

On a 4-point Likert scale (ordinal data)

It does not provide clear guidelines for how to (a) administer the assessment (b) train listeners to make the judgments (c) evaluate reliability

Consensus Auditory-Perceptual Evaluation of Voice

Uses a visual analogue scale, thereby permitting parametric analyses of the ratings Clear administration protocol

MOSCAPE Middle vs. extreme part of the scale Order in which voices are presented Scale & resolution Context (i.e. with other severe voices or with normal voices) Anchor Pathological vs normal Experience of the rater

It classifies dysarthria into 5 different types

• Good overall - Varied among perceptual dimensions - i.e. hoarseness, harshness and strain were more variable.