Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Tue, 5 Aug 2003 10:51:30 -0400
You might want to use the survplot function in the Design library (builtin to S-Plus but you should get updates from our web site). survplot will automatically label the KM curves where they are most
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Mon, 25 Aug 2003 11:09:56 -0400
The rcorr.cens function in the Hmisc library will compute Somers' Dxy rank correlation and its standard error. Dxy = 2(C-.5) where C=ROC area (nonparametric estimator from Mann-Whitney U statistic).
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Mon, 25 Aug 2003 14:20:46 -0400
New versions of Hmisc and Design may be obtained from http://hesweb1.med.virginia.edu/biostat/s/Hmisc.html http://hesweb1.med.virginia.edu/biostat/s/Design.html for S-Plus 6.x for Windows and for S-P
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Wed, 27 Aug 2003 12:00:14 -0400
The nonparametric ROC area is essentially the Wilcoxon-Mann-Whitney linear rank statistic. A book on linear rank statistics will give you all you need. FE Harrell -- Frank E Harrell Jr Prof. of Biost
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Thu, 3 Jul 2003 17:12:34 -0400
You might look at the simple james.stein function in the Hmisc library. There is no documentation for it but it implements what is discussed in @ARTICLE{efr77sti, author = {Efron, Bradley and Morris,
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Tue, 8 Jul 2003 15:04:10 -0400
Your note should have been sent to r-help. is.missing in S-Plus just checks if the vector is empty and if so returns TRUE, otherwise it's the same as is.na. -- Frank E Harrell Jr Prof. of Biostatisti
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Fri, 6 Jun 2003 11:03:11 -0400
lsmeans is an unfortunate construction of SAS Institute that is poorly named and uses poor default contrasts. See the contrast function in the Design library (type ?contrast.Design) for a more natura
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Fri, 6 Jun 2003 11:50:26 -0400
Do library(Hmisc,T) ?wtd.stats -- Frank E Harrell Jr Prof. of Biostatistics & Statistics Div. of Biostatistics & Epidem. Dept. of Health Evaluation Sciences U. Virginia School of Medicine http://hesw
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Mon, 23 Jun 2003 10:53:33 -0400
How does one get the following to work properly in S-Plus 6? [1] 01/01/1960 01/02/1960 01/03/1960 01/04/1960 01/05/1960 01/06/1960 [7] 01/07/1960 01/08/1960 01/09/1960 01/10/1960 01/11/1960 [[1]]: [[
Author: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Mon, 30 Jun 2003 11:38:18 -0400
Those are quantities that one very seldom tests hypotheses about but more typically makes estimates (point and interval) about, using standard binomial methods (even then, sensitivity and specificity