Matched and self-reported loudness of tinnitus: methods and sources of error

A consecutive series of 62 patients who reported tinnitus at the time of their first attendance at a neuro-otology clinic were studied. Loudness matches were obtained both at the frequency of the tinnitus and at 1 kHz. These matches were expressed in dB HL, dB SL and in units derived from individualised loudness functions (personal loudness units; PLUs). Self-reports of the loudness of tinnitus at the time of loudness matching were obtained on five different scales. Moderate correlations were found between self-reported loudness and some of the scales by which the loudness match was expressed. When subjects who had some difficulty with the test procedures were excluded, the correlations between PLU expressions of the matched loudness and certain of the self-report scales were found to be markedly improved. Correlations of traditional expressions of matched loudness with self-report improved to a limited extent. It was concluded that: measurement error can appreciably reduce the maximum correlation between the best self-report measure and loudness match measures; PLU conversions of matched loudness data produce the highest correlations with self-report measures of loudness, and explicitly labelled self-report scales (Guttman and adjectival) produce better correlations with loudness match values than other self-report scales.