EFR test–retest repeatability

The accuracy of the fitted slope to the EFR magnitude-level function will depend on the test–retest variability of each individual EFR data point. In order to have an estimate of the measurement variability, the repeatability of individual EFR magnitudes at three stimulus levels was assessed using a Bland–Altman analysis⁶⁵. Data was also analysed using a one-way, random-effects, single-rater intra-class correlation (ICC) coefficient^66,67, interpreted using the guidelines provided by⁶⁸. Normality was ensured by means of a visual inspection of the corresponding quantile–quantile (Q-Q) plots⁶⁹ and by computing a Shapiro–Wilk normality test⁷⁰ (not shown). In the Bland–Altman analysis, the test–retest difference values (i.e., the value of the retest subtracted to the value of the test) were plotted against the mean response amplitude between two test runs. This method defines the limits of agreement (LoA) as 1.96 times the standard deviation of the differences. The method proposed by⁷¹ was used to compute 95% confidence interval (CI) for the mean of the differences and for the upper and lower LoA. The same repeatability analysis was performed on the EFR slopes. At frequencies 500, 1000 and 2000Hz, only the stimulus levels of 35 and 55dB SPL were used both for the test and retest, because the level of 75dB does not belong to the compressive part in the EFR magnitude-level functions. For these frequencies, the EFR slope was estimated as the difference between the EFR magnitudes at 55 and 35dB divided by 20. At 4kHz, all three repeated levels were used (35, 55 and 75dB), and the EFR slope was estimated by fitting a first order polynomial to the three data points in both the test and the retest. In any case, only statistically significant EFR data points (positive F-test) were considered in the repeatability analysis, discarding those test–retest pairs that contained missing data points.