Notice: On March 31, it was **announced** that Statalist is moving from an email list to a **forum**. The old list will shut down on April 23, and its replacement, **statalist.org** is already up and running.

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

From |
"Tan, Ji (GRS, Wuhan)" <Ji.Tan@towerswatson.com> |

To |
"statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu> |

Subject |
RE: st: How is confidence interval calculated in weighted GLM in Stata? |

Date |
Sat, 1 May 2010 09:46:16 +0800 |

Sorry I did not know the convention here, and Thanks for the reference, while I don't have immediate access to. Let me rewrite in plain text: Say in a simple Poisson regression, E(Yi)=ui=Oi*exp(b), Log likelihood function is {sum (Yi*b-ui)}, when we add weighting wi, it became {sum wi(Yi*b-ui)}, The estimation of b is {(sum wi*Yi)/(sum wi*Oi)}, and the Fisher Score algorithm gives its distribution as N(b, (sum wi^2*ui)/(sum wi*ui)^2). However, this calculation only agree with Stata when we have equal weights, i.e. all the wi are the same. So would anyone kindly explain this to me? Best regards, Ji -----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Maarten buis Sent: Friday, April 30, 2010 8:02 PM To: statalist@hsphsun2.harvard.edu Subject: Re: st: How is confidence interval calculated in weighted GLM in Stata? --- On Fri, 30/4/10, Tan, Ji (GRS, Wuhan) wrote: > Say in a simple Poisson regression, EYi= i=Oi*exp() > Log likelihood function is So And Fisher Score > algorithm gives: This is in line with Stata. > > However, in the weighted case, log likelihood > function changes to , , > I find Stata does not follow this numerically. So > would anyone kindly explain this to me? You can only sent plain text and no attachments to the statalist, so your message got hopelessly garbled. So I can't answer you question directly, I can only give you some literature recomendations: James W. Hardin and Joseph M. Hilbe "Generalized Linear Models and Extensions, second edition. College Station, TX: Stata Press. <http://www.stata-press.com/books/glmext.html> Hope this helps, Maarten -------------------------- Maarten L. Buis Institut fuer Soziologie Universitaet Tuebingen Wilhelmstrasse 36 72074 Tuebingen Germany http://www.maartenbuis.nl -------------------------- * * 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/ Notice of Confidentiality This transmission contains information that may be confidential. It has been prepared for the sole and exclusive use of the intended recipient and on the basis agreed with that person. If you are not the intended recipient of the message (or authorized to receive it for the intended recipient), you should notify us immediately; you should delete it from your system and may not disclose its contents to anyone else. This e-mail has come to you from the Wuhan branch of Watson Wyatt Consultancy (Shanghai) Limited, a Towers Watson company. * * 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/

**References**:**st: How is confidence interval calculated in weighted GLM in Stata?***From:*"Tan, Ji (GRS, Wuhan)" <Ji.Tan@towerswatson.com>

**Re: st: How is confidence interval calculated in weighted GLM in Stata?***From:*Maarten buis <maartenbuis@yahoo.co.uk>

- Prev by Date:
**Re: st: RE: RE: Parplot with value labels** - Next by Date:
**st: using Stata to detect interviewer fraud** - Previous by thread:
**Re: st: How is confidence interval calculated in weighted GLM in Stata?** - Next by thread:
**st: corrected OR/LR** - Index(es):