FW: GLM confidence intervals

["Coetzer, Roelof (RLJ)" <roelof.coetzer@sasol.com>]

A while ago we asked the s-news list regarding the calculation of the Wald asymptotic confidence intervals on Splus.  Hugh Jones replied that we should evaluate the pointwise() function itself in order to try and figure out the confidence intervals used.  However, this does not solve our problem. 

 

Does anyone have a program that produces the same type of Wald confidence intervals for GLM’s on Splus as what SAS PROC GENMOD produces?

 

Thank you

Roelof Coetzer
Statistical Consultant
Sasol Technology R&D Division
Sasolburg
South Africa
Tel: +27 16 9603888
Fax: +27 11 5223425
E-mail: roelof.coetzer@sasol.com

 

-----Original Message-----
From: Hugh Jones [mailto:hajones@dlwc.nsw.gov.au]
Sent: 10 February 2003 12:39 AM
To: Coetzer, Roelof (RLJ)
Subject: Re: [S] GLM confidence intervals

 

You can often determine the code behind S-Plus functions by typing the function name omitting the parentheses.

 

Hence, typing:

 

> pointwise

gives:

 

function(results.predict, coverage = 0.99)
{
 fit <- results.predict$fit
 limits <- qt(1. - (1. - coverage)/2., results.predict$df) *
  results.predict$se.fit
 list(upper = fit + limits, fit = fit, lower = fit - limits)
}

This should give you a better idea what is going on.

 

Hugh Jones
Senior Biometrician
Centre for Natural Resources
NSW Depart. Land & Water Conservation
PO Box 3720, Parramatta NSW 2124

 

Tel:    61-2- 9895 7628
Fax:   61-2- 9895 7867

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

NOTICE: Please note that this eMail, and the contents thereof,
is subject to the standard Sasol eMail disclaimer which may be found at:
http://www.sasol.com/disclaimer.htm

If you cannot access the disclaimer through the URL attached and
you wish to receive a copy thereof please send an eMail to
disclaimer@sasol.com. You will receive the disclaimer by return eMail.