Reliability of broadband middle-ear power reflectance in younger and older adults: application of generalizability theory.

PURPOSE To assess the reliability of broadband middle-ear power reflectance (BMEPR) and transmittance profiles for chirp and tonal stimuli using generalizability theory (GT). METHOD In adults without a history of middle-ear disease, the authors assessed the reliability of BMEPR to chirp and tonal stimuli using a multivariate approach based on an analysis of variance model (GT). For comparisons with other published studies, Pearson's product-moment correlation coefficients (Pearson's r) also were used. RESULTS Based on GT with chirp stimuli, overall BMEPR measures had good reliability; however, the reliability of individual profiles across frequencies and ears was less than optimal. Lower generalizability coefficients were found when transmittance was evaluated. Test-retest reliability using Pearson's r was better for right versus left ears, and mid-frequencies were generally more reliable than those at either extreme of the frequency range. In contrast, tonal stimuli had higher generalizability coefficients and Pearson's r values than chirps for all frequencies tested; Pearson's r values were also higher for right versus left ears. CONCLUSION The authors extended the use of GT as a preferred way to evaluate reliability of BMEPR and transmittance profiles for chirps and tones because it allows for a more comprehensive evaluation compared with unidimensional pairwise correlations.