i just ran rocfit on a dataset obtaining the following output. surely the
std error or the AUC can't be right (>1). would appreciate advice and
comments. fwiw, roctab showed standard error of 0.0724.
Binormal model of fbs on onefive Number of obs =
39
Goodness-of-fit chi2(-1) = 0.00
Prob > chi2 = .
Log likelihood = -23.115784
------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
intercept | 1.085096 4.579e+03 0.00 1.000 -8.973e+03
8.975e+03
slope (*) | 0.912109 5.161e+03 -0.00 1.000 -1.011e+04
1.012e+04
-------------+----------------------------------------------------------------
_cut1 | 0.887147 0.362530 2.45 0.014 0.176601
1.597692
------------------------------------------------------------------------------
------------------------------------------------------------------------------
| Indices from binormal fit
Index | Estimate Std. Err. [95% Conf.
Interval]
-------------+----------------------------------------------------------------
ROC area | 0.788637 382.665741 -7.492e+02
750.799708
delta(m) | 1.189655 1.712e+03 -3.354e+03
3.356e+03
d(e) | 1.134972 1.726e+03 -3.381e+03
3.383e+03
d(a) | 1.133775 1.871e+03 -3.665e+03
3.668e+03
------------------------------------------------------------------------------
(*) z test for slope==1
