Statistical Properties of Surface-Recorded Electrocochleographic Responses

Abstract

If surface-recorded electrocochleographic responses are to be used as the basis of a diagnostic procedure, then the statistical properties of the responses obtained from normally-hearing subjects need to be known so that accurate comparisons with the responses obtained from patients can be made. The data were obtained from 6 normally-hearing subjects who were tested at four stimulus levels, each measurement being replicated twice in each of two test sessions. The distributions of the latency and amplitude values were tested for normality, and the restrictions required by the use of parametric statistical methods are defined. Analyses of variance were carried out to investigate subject, session and subject-session interaction effects. The relationships between response peak parameters, and the overall mean values of latency and amplitude were computed at each stimulus level. From these the joint distributions of latency and amplitude and their relationships were derived. The relationship between response amplitude and response variance was calculated, and the ways in which the results can form a baseline for clinical diagnoses are discussed.