Skip to main content
. 2012 Jun;16(2):67–82. doi: 10.1177/1084713812451159

Table 2.

Accuracy in Task 1 and Estimated Coefficients in the Logistic Modelsa Fitted to Individual Data (for CI Listeners) and Group Data (for Both CI and NH Listeners) in Task 2

Listener/Group/Condition Accuracy in Task 1 (% (SD) F0 Height F0 Contour Peak Intensity Ratiob Duration Ratio
Full spectrum
CI-1 78.33 0.0632 2.2657** 0.0610* 0.3557
CI-2 94.17 0.2506 6.1637** 0.1251** 1.5800**
CI-3 98.33 0.0306 6.1041** −0.0565** −0.2227
CI-4 98.33 1.9140** 4.5714** 0.1580** −2.6335**
CI-5 63.33 0.2162 1.7730** 0.1786** 0.8243**
CI-6 95.83 0.6709** 3.2154** 0.0069 −1.0834**
CI-7 96.67 −0.8591** 7.1140** 0.0448* −0.2085
CI-8 65.83 0.5551* 2.4198** 0.0194 0.5151*
CI-9 95.83 3.7448** 4.8787** 0.1713** −2.4352**
CI-10 93.33 2.1612** 2.6967** 0.2812** 1.2774**
CI-11 58.00 1.4831** 1.1909** 0.2747** 1.2549**
CI-12 85.00 2.7837** 4.3230** 0.0068 −0.7333*
CI-13 75.00 1.9817** 2.2823** 0.1977** 1.8613**
CI group 84.46 (14.68 0.8141** 2.3968** 0.0841** 0.0952
NH-1 100.00 0.5218 5.9949** 0.0191 −0.8503*
NH-2 95.00 1.4088** 5.0620** 0.0095 −0.0733
NH-3 96.67 0.8809** 5.4048** −0.0065 1.0125*
NH-4 99.17 0.2879 9.2178** −0.0144 0.6219
NH group 97.71 (2.29 0.7123** 5.2470** 0.0025 0.1294
Acoustic CI simulations (group data only)c
16-channel 91.25 (1.98 NA NA NA NA
8-channel 84.79 (9.75 1.5452** 2.5262** 0.1317** 1.5280**
4-channel 80.63 (10.77 0.6000* 1.9564** 0.1170** 1.4540**
1-channel 69.38 (12.70 NA NA NA NA

Note: CI = cochlear implant; NH = normal hearing; NA = not applicable/not tested.

a

The equation used to derive the estimated coefficients is as follows: Expected logarithm of odds of question response = Intercept ( (A) × F0 height (=1 if 220 Hz; =0 if 120 Hz) ( (B) × F0 contour (in octaves) ( (C) × peak intensity ratio (in dB) ( (D) × log (duration ratio), where (A), (B), (C), and (D) in this equation refer to the estimated coefficient listed in column entitled F0 height, F0 contour, peak intensity ratio, and duration ratio, respectively.

b

Rather than using duration ratio, which was restricted to positive values, logarithm of duration ratio was adopted since it theoretically took values on all real numbers, and thus, it was more appropriate in a logistic model setting.

c

Only group data are shown for acoustic CI simulations, as estimated coefficients were quite consistent across NH listeners.

*

p < .05. **p < .001.