Table 8. Comparison of the discrimination ability of the predictive models with DeLong-Test (N = 121).
| Types of correction mechanisms compared in motor performance models |
AUC [95% CI]corrected scores* |
∆ = AUC [95% CI]corrected scores - AUC [95% CI]raw scores† |
z |
p |
| Mirwald |
.73 [.63; .82] |
.00 [–.05; .04] |
–0.10 |
.922 |
| Moore-1 |
.73 [.64; .82] |
.00 [–.04; .05] |
0.06 |
.950 |
| Moore-2 |
.73 [.64; .82] |
.00 [–.03; .03] |
0.15 |
.879 |
| Fransen-1 |
.73 [.64; .82] |
.00 [–.05; .05] |
–0.02 |
.981 |
| Fransen-2 |
.73 [.64; .82] |
.00 [–.05; .05] |
0.04 |
.972 |
| %PAH |
.71 [.61; .80] |
-.02 [–.07; .02] |
–0.92 |
.356 |
| %RAH |
.76 [.67; .84] |
.03 [–.01; .07] |
1.33 |
.182 |
%PAH = attained percentage of the predicted adult height, %RAH = attained percentage of the real adult height; AUC = area under the curve. Adjusted %-level (Bonferroni correction) = 0.007. *AUC [95% CI]
corrected scores = These statistics represent the AUC based on the regression model using corrected scores. † AUC [95% CI]
raw scores = This statistic represents the AUC based on the binary regression model using raw scores and corresponds to 0.73 [0.64; 0.82].
