Psychophysical and Psychometric Properties of Visual Analog Scales: An Auditory-Perceptual Study Applied to Hypokinetic Dysarthria.

PURPOSE: Despite the widespread use of visual analog scales (VASs), their psychophysical and psychometric properties for rating dysarthria remain largely untested. Using a scale whose measurement properties, reliability, and validity are unknown undermines the comparison of findings across studies and clinical contexts. Therefore, we investigated these properties of VAS using samples of hypokinetic dysarthria.

METHOD: Thirty-seven nonexperts and three dysarthria experts evaluated 85 speech samples from individuals with Parkinson's disease and healthy controls on overall severity, reduced loudness, articulatory imprecision, short rushes of speech, and monotony. Listeners used VAS and direct magnitude estimation (DME), which is an established ratio-level scale, to complete the ratings. To determine measurement level, we assessed the relationship between VAS and DME; for inter- and intrarater reliability, we calculated intraclass correlation coefficients for both scales, and for criterion validity for both scales, we applied Spearman's correlations to the nonexpert and expert ratings.

RESULTS

VAS ratings showed strong linear relationships with DME for overall severity, reduced loudness, short rushes of speech, and monotony. Inter- and intrarater reliability values were good to excellent for VAS and significantly higher than those for DME. Criterion validity was strong to very strong for both scales across all features. Compared to experts, nonexperts provided significantly higher VAS and DME scores; however, the effect sizes of the between-listeners group differences were small.

CONCLUSIONS

VAS is a reliable and valid ratio-level scale for assessing prothetic features of hypokinetic dysarthria including overall severity, reduced loudness, and short rushes of speech. For metathetic dimensions such as monotony, a linear relation between VAS and DME does not indicate that the VAS functions as a ratio scale since linearity is expected for both interval- and ratio-scaled judgments. More work is needed to address the feature-specific variations and establish additional psychometric properties of VAS, such as sensitivity and specificity.