Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Fwd: Smoothed ROC Curves


From   Svend Juul <[email protected]>
To   <[email protected]>
Subject   Re: st: Fwd: Smoothed ROC Curves
Date   Mon, 19 Jan 2009 09:38:46 +0100

John wrote:

I am working on a diagnostic test for an infectious disease.  I am
assessing optimal cut off points within the continuous measure (which
is strongly skewed) using ROC curves, but am unclear as to whether or
not one must smooth the ROC curve prior to selecting cut off points
(my sample size is approximately 800 with 118 events, not sure if this
is important). Also, I am pretty new to Stata and am not sure how I
would go about doing this (I think this would involve log transforming
the measure and using rocfit, but beyond that I am uncertain).

===================================================================

Some partial advice:

1.
ROC analysis utilize the ranks, not the absolute values of the 
predictor variable, so transforming it has no consequence.

2.
ROC analysis can not tell you the "optimal cut off points" - but it 
can help you decide. What is "optimal" also depends on the consequences
of false positive and false negative conclusions.

3.
You might benefit from Roger Newson's -senspec- command. Find it by:

   findit senspec

Hope this helps
Svend


__________________________________________

Svend Juul
Institut for Folkesundhed, Afdeling for Epidemiologi
(Institute of Public Health, Department of Epidemiology)
Bartholins Allé 2
DK-8000  Aarhus C, Denmark
Phone:  +45 8942 6090
Home:   +45 8693 7796
Email:  [email protected]
__________________________________________ 

*
*   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/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index