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: FW: Standard error estimates lincom vs. estimates from predict command


From   a b <andythezoologist@hotmail.com>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: FW: Standard error estimates lincom vs. estimates from predict command
Date   Mon, 14 Nov 2011 16:59:17 +0000

Dear Statalisters,

There is probably a simple explanation (so patience might be required!) for this but I could not find an answer yet.

I have a count data dataset which I modelled using a GLM with Poisson distribution and log link.

I wanted to get standard errors to derive 95% CI for the predicted outcome values (i.e. each record in my dataset) (as an aside, I wanted to calculate the % coverage of the model estimates and the observed counts)
So I used 'predict' with the xb option to get a linear prediction and the stdp option to get the standard error around the linear prediction.

I then tried some combinations of the predictors in lincom
lincom beta1*x + beta2*z ... + _cons
and compared the predicted values and standard errors from lincheck and predict.

While the predicted values were the same (as expected) from both, there was a very large difference between the standard errors (the standard errors being much larger from the predict stdp route).

Can anyone tell me why this may be the case? And if I am testing the model using % coverage, which standard error I should derive my 95% CI from 

Any help would be much appreciated! 		 	   		  
*
*   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