Table 3.
Regression analyses of changes in ABR thresholds with age.
| Air Conduction | Bone Conduction | Air - Bone | |||
|---|---|---|---|---|---|
| frequency (kHz) |
N | slope±SE (dB/day) |
N | Slope±SE (dB/day) |
Differences in slope |
| 5.6 | 56 | 0.028±0.008 (p =0.0016) |
53 |
0.007±0.007
(p=0.3149)# |
0.021 |
| 8.0 | 56 | 0.052±0.008 (p <0.0001) |
53 | 0.015±0.005 (p=0.0036) |
0.037 |
| 11.3 | 56 | 0.077±0.010 (p <0.0001) |
53 | 0.063±0.006 (p <0.0001) |
0.014 |
| 16 | 56 | 0.173±0.010 (p <0.0001) |
53 | 0.116±0.007 (p <0.0001) |
0.057 |
| 52 | 0.153±0.009 (p <0.0001) |
40 | 0.066±0.012 (p <0.0001) |
0.087 | |
| 22.6 | 56 | 0.205±0.006 (p <0.0001) |
55 | 0.126±0.006 (p <0.0001) |
0.079 |
| 45 | 0.186±0.007 (p <0.0001) |
35 | 0.063±0.019 (p <0.0024) |
0.123 | |
| 32.0 | 56 | 0.262±0.010 (p <0.0001) |
55 | 0.126±0.009 (p <0.0001) |
0.136 |
| 29 | 0.331±0.055 (p <0.0001) |
27 | 0.085±0.051 (p=0.108)# |
0.246 | |
Entries include all frequency and age conditions that contain measurements of both AC and BC sensitivity. The first column is the frequency, the second and third columns describe the number of points and the computed slope, standard error slope and uncorrected level of significance for the AC data, the fourth and fifth columns describe the same for the BC data, and the sixth column describes the difference between the slopes computed for the Air and Bone data. The Ns associated with the entries at frequencies of 5.6, 8, and 11.3 kHz and the second entries at the higher frequencies describe the number of the AC and BC thresholds determined below the maximum stimulus ceiling. The first entries at 16, 22.6 and 32 kHz describe the Ns and the slopes estimated when we include threshold estimates at 5 dB above the ceiling. As illustrated in Figure 9, all the individual age and threshold data (including the 37-day-old data) were normalized to the mean age and mean thresholds in the 37-day-old group. The regression coefficients were computed using least-squares techniques assuming a linear intercept of 0 dB at a relative age of 0 days. The calculated slopes have units of dB/day. Application of a Benjamini - Hochberg compensation for multiple tests suggest that all uncompensated p values ≤ 0.0294 are significantly different from 0 at a level equal to or better than p=0.05 with a 5% chance of a false positive. Non-significance is marked by a #.