Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

st: RE: loglink and normality, mixed model


From   "Feiveson, Alan H. (JSC-SK311)" <[email protected]>
To   "[email protected]" <[email protected]>
Subject   st: RE: loglink and normality, mixed model
Date   Mon, 24 Feb 2014 13:56:49 +0000

Ruud - Do you mean the error term eps[i,j] to be additive? If so log(EY) is not what you have, and in fact would be undefined if Y can be 0 or negative. Perhaps you mean the error term is multiplied, not added. Assuming that a >0, Y then, cannot be zero. You need to treat zeros as a special case- not arising from this model.

Al Feiveson

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Ruud van Lieshout
Sent: Monday, February 24, 2014 6:49 AM
To: [email protected]
Subject: st: loglink and normality, mixed model

Dear all,

a question about a (simple) generalized linear mixed model. Say I have a model describing a response over time for different panels: 
Y[i,j]=a[i]*time[j]^b[i]+eps[i,j], with i= panel index and j time index. 


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index