Reliability, sensitivity and validity of magnitude estimation, category scaling and paired-comparison judgments of speech intelligibility by older listeners

This study investigated the reliability, sensitivity, and validity of speech intelligibility judgments for hearing aid evaluation. Subjects aged 60-87 years judged the intelligibility of sentences using either magnitude estimation, category scaling or paired comparisons. The 60+ age group was chosen as representative of the majority of hearing aid wearers. Speech recognition scores for Central Institute for the Deaf (CID) sentences and Northwestern University Auditory Test No. 6 (NU-6) words were also obtained. The speech was bandpass filtered using filter settings that produce a monotonic increase in predicted intelligibility based on articulation index theory. Speech recognition scores and intelligibility judgments were obtained for each of eight filter conditions. Test-retest reliability was poorest for paired comparisons and CID sentence scores. There were no differences in sensitivity among the three psychophysical procedures. Intelligibility judgments and NU-6 scores were more sensitive than CID sentence scores to differences among conditions. The results indicated that intelligibility judgments are valid measures of speech recognition.