You might want to use the -senspec- package, downloadable from SSC using
the -ssc- command, which computes sensitivity and specificity for the
cutpoint represented by each observation. You can then plot or tabulate
these sensitivities and specificities as you wish.
I hope this helps.
Best wishes
Roger
Roger B Newson BSc MSc DPhil
Lecturer in Medical Statistics
Respiratory Epidemiology and Public Health Group
National Heart and Lung Institute
Imperial College London
Royal Brompton Campus
Room 33, Emmanuel Kaye Building
1B Manresa Road
London SW3 6LR
UNITED KINGDOM
Tel: +44 (0)20 7352 8121 ext 3381
Fax: +44 (0)20 7351 8322
Email: [email protected]
Web page: http://www.imperial.ac.uk/nhli/r.newson/
Departmental Web page:
http://www1.imperial.ac.uk/medicine/about/divisions/nhli/respiration/pop
genetics/reph/
Opinions expressed are those of the author, not of the institution.
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Jacob Wegelin
Sent: 10 January 2009 17:29
To: [email protected]
Subject: st: roctab sensitivity specificity for each cutpoint
- roctab - with the detail option reports the sensitivity and
specificity values which one could use to create one's own ROC plot.
But it only spits these numbers out as text, not as a dataset. Am I
correct in thinking that, if I want a dataset with these values, "I'm
on my own" as they say, I will not be able to obtain them from roctab
or its sister functions?
One reason it would be nice to have these numbers in a dataset:
roccomp makes a nice plot, but the legend is limited: it uses the
variable names, not the variable labels. Suppose, for instance, that
our continuous predictor of "spontaneous survival" is MELD, but that
one has to
gen minusMELD=-MELD
to get a ROC curve that is above the reference line. Then the roccomp
plot has a funny label for the MELD ROC curve, and a clinician with
whom one collaborates will be confused.
If one had the actual sensitivity and specificity for each cutpoint,
then one could produce one's own program that, for instance, extracts
the variable labels, like this:
. local junk: variable label log_BILin
. di "`junk'"
log(Bilirubin level at admission)
and thereby produce a more accessible plot legend.
Of course it would be fairly straightforward to write a program that
computes the sensitivity and specificity for each cutpoint; I'm just
curious whether that would constitute "reinventing the wheel."
Thanks for any comments
Jacob A. Wegelin
Assistant Professor
Department of Biostatistics
Virginia Commonwealth University
730 East Broad Street Room 3006
P. O. Box 980032
Richmond VA 23298-0032
U.S.A.
E-mail: [email protected]
URL: http://www.people.vcu.edu/~jwegelin
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
