HSC Section 8_April 2017

J. FINBOW ET AL.

detected in the unaided condition with noise to the poorer ear (W = 0.796, p = 0.026), but there was no significant skew or kurtosis in any condition, and so data were an- alyzed with a repeated measures ANOVA. A significant main effect of noise presentation ear was found (F(1,7) = 54.8, p = 0.0001, G 2 = 0.496), indicating that scores were better when noise was presented to the poorer ear. The interaction between noise presentation ear and de- vice condition was also significant (F(2,14) = 15.5, p = 0.0002, G 2 = 0.218). Post hoc pairwise t tests with Bonferroni corrections showed that scores were signifi- cantly better with noise presented to the poorer ear for the unaided condition (t(7) = 7.1, p = 0.003). There were no significant differences between device conditions with noise presented to the better ear, but scores with the CROS device were significantly poorer than in the unaided condition with noise presented to the poorer ear (t(7) = 8.96, p = 0.0007). Self-Assessment Questionnaires The BBSS and SSQ questionnaires were scored using the standard test procedures. As in Kompis et al. (27), individual totals for the BBSS were obtained by adding the ratings for each listening scenario. The average total score was 26.2 (range = 12.5 Y 39) for the BAHD and 25.4 (range = 11 Y 40) for the CROS. This finding corresponds to grand mean ratings of 2.6 for the BAHD and 2.5 for the CROS. Figure 4 displays mean BBSS ratings for all listeners and both devices using an equal-area violin plot, truncated at minimum and maximum values. These data were not distributed normally in 6 of the 20 sub- conditions, so a Friedman’s nonparametric ANOVA was used for subsequent analysis. No significant differences were found between ratings for the BAHD and CROS, but ratings varied significantly across test items ( W 2 (9) = 35.4, p = 0.00005). Friedman’s post hoc measures indicated that ratings for conversation in quiet were significantly higher

noise presentation ear and device condition was also sig- nificant (F(2,14) = 6.192, p = 0.01, G 2 = 0.165). Post hoc pairwise t tests showed that scores were significantly lower when noise was presented to the better ear than when noise was presented to the poorer ear, but only in the unaided condition (t(7) = 8.33, p = 0.001); the differences related to ear of noise presentation for the BAHD and CROS devices were not significant. Figure 2 shows that scores were slightly higher for both devices relative to the unaided condition with noise presented to the better ear (S0Nbe), and slightly lower with noise presented to the poorer ear (S0Npe), although these differences were also not found to be significant. QuickSIN Scores Figure 3 shows mean QuickSIN scores for all three device conditions with babble noise presented to the poorer and better ear, with error bars showing 95% con- fidence intervals. A slight departure from normality was FIG. 2. Mean word recognition scores for each device condition with noise presented to the better ear (S0Nbe), noise presented the poorer ear (S0Npe), and in quiet (Spe). Error bars denote 95% confidence intervals.

FIG. 3. Mean speech-to-noise ratio thresholds for the QuickSIN test in the S0Nbe and S0Npe conditions. Lower values corre- spond to better performance. Error bars denote 95% confidence intervals.

FIG. 4. Equal-area violin plot of BBSS ratings for all listeners and both devices, truncated at maximum and minimum values. Values of j 5 correspond to ‘‘Much easier without the device,’’ and values of +5 correspond to ‘‘Much easier with the device.’’

Otology & Neurotology, Vol. 36, No. 5, 2015

151