Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


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

st: Inconsistent results with rocfit


From   Paul Seed <paul.seed@kcl.ac.uk>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   st: Inconsistent results with rocfit
Date   Thu, 25 Feb 2010 18:30:40 +0000

Dear Statalist,

An odd problem has come up.
I have two versions on the same predictor
(as measured & logged) , and one binary outcome.

When I use -roctab-, I get identical estimates of the ROC area.
when I use -rocfit-, I do not.

Is this to do with the way -rocfit-
divides the predictor into groups?
Is it a bug in -rocfit-?

I am using Stata version 10.1 on a Windows machine, in case it is relevant.
I can supply the data set if anyone is interested, but won't take up bandwidth by listing it here..

********************************************************
* Commands:
gen l_max_mcp_1 = log( max_mcp_1)
su   max_mcp_1 if  max_mcp_1 >0, mean
replace  l_max_mcp_1  = log(r(min)/2) if  max_mcp_1 ==0
corr l_max_mcp_1 max_mcp_1
spearman l_max_mcp_1 max_mcp_1

rocfit  del_34 max_mcp_1 , cont(5) nolog
rocfit  del_34 l_max_mcp_1 , cont(5) nolog


********************************************************
* Selected output

. version
version 10.1

. corr l_max_mcp_1 max_mcp_1
(obs=125)

            | l_max_~1 max_mc~1
-------------+------------------
l_max_mcp_1 |   1.0000
  max_mcp_1 |   0.6305   1.0000


.
. spearman l_max_mcp_1 max_mcp_1

Number of obs =     125
Spearman's rho =       1.0000

.

. rocfit  del_34 max_mcp_1 , cont(5) nolog

Binormal model of del_34 on max_mcp_1 Number of obs = 120
Goodness-of-fit chi2(1) =        4.86
Prob > chi2             =      0.0274
Log likelihood          =  -62.882551

------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
intercept | 0.736393 1.202242 0.61 0.540 -1.619957 3.092743 slope (*) | 1.303260 0.890019 0.34 0.733 -0.441145 3.047665
-------------+----------------------------------------------------------------
/cut1 | 1.199603 0.168230 7.13 0.000 0.869879 1.529326 /cut2 | 1.404337 0.184995 7.59 0.000 1.041753 1.766921 /cut3 | 1.563270 0.209553 7.46 0.000 1.152554 1.973987
------------------------------------------------------------------------------
------------------------------------------------------------------------------
            |                    Indices from binormal fit
Index | Estimate Std. Err. [95% Conf. Interval]
-------------+----------------------------------------------------------------
ROC area | 0.673024 0.198622 0.283733 1.062316 delta(m) | 0.565039 0.564877 -0.542100 1.672178 d(e) | 0.639435 0.810891 -0.949883 2.228753 d(a) | 0.633964 0.778517 -0.891903 2.159830
------------------------------------------------------------------------------
(*) z test for slope==1

. rocfit  del_34 l_max_mcp_1 , cont(5) nolog

Binormal model of del_34 on l_max_mcp_1 Number of obs = 120
Goodness-of-fit chi2(2) =        0.72
Prob > chi2             =      0.6961
Log likelihood          =  -149.19666

------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
intercept | 0.972725 0.318396 3.06 0.002 0.348680 1.596771 slope (*) | 1.127073 0.243452 0.52 0.602 0.649916 1.604230
-------------+----------------------------------------------------------------
/cut1 | -0.942573 0.150530 -6.26 0.000 -1.237607 -0.647539 /cut2 | -0.642605 0.137181 -4.68 0.000 -0.911475 -0.373735 /cut3 | 0.946408 0.148170 6.39 0.000 0.656001 1.236815 /cut4 | 1.565039 0.196433 7.97 0.000 1.180037 1.950041
------------------------------------------------------------------------------
------------------------------------------------------------------------------
            |                    Indices from binormal fit
Index | Estimate Std. Err. [95% Conf. Interval]
-------------+----------------------------------------------------------------
ROC area | 0.740724 0.058236 0.626584 0.854864 delta(m) | 0.863055 0.236093 0.400322 1.325788 d(e) | 0.914614 0.256090 0.412687 1.416541 d(a) | 0.912986 0.254270 0.414626 1.411346
------------------------------------------------------------------------------
(*) z test for slope==1


---------

paul.seed@kcl.ac.uk (work), paul.t.seed@gmail.com (personal)

King's College London, Division of Reproduction and Endocrinology
& Department of Primary Care and Public Health Sciences
St Thomas' Hospital, Westminster Bridge Road, London SE1 7EH

*
*   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 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index