Prev: SAS BI DI ETL BASE Clinical Financial + software
Next: 2 PART MODEL
From: tamuheyii on 21 Nov 2009 14:57
Hello All, I got this panel data with individual purchase, advertising
and promotion to individuals.

I can setup regular regression as purchase=intercept+ad+promotion. I
want to test fixed effect by adjusting intercept to change across
individual.. But I am not sure how to do this when the purchase is
distributed in Poisson.

Thanks for any suggestion. I did the normal poisson regression with
nlmixed procedure, so if you can help me to integrate the fixed effect
into nlmixed procedure, that would be very nice.

From: Dale McLerran on 24 Nov 2009 01:27
Either the NLMIXED procedure or the GLIMMIX procedure can
fit a Poisson model with random intercept. Perhaps you can
give us the code which you employed for fitting the fixed
effect Poisson using NLMIXED and also identify the variable
which indicates which panel a response belongs to.

Also, it would be good if you would provide a little more
information about the data which you have collected. Often,
with panel data there can be a random intercept as well
along with residual covariance structure within the panel.
When observations are collected over time, observations
which are close in time may be more like one another than
observations that are distant in time, even within the
panel.

If all you need to do is account for a random intercept,
then the NLMIXED procedure or the GLIMMIX procedure may
both be appropriate. If you need to account for covariance
within the panel, then you will almost surely have to use
the GLIMMIX procedure. Write with more information and
we can provide a better response.

Dale

---------------------------------------
Dale McLerran
Fred Hutchinson Cancer Research Center
mailto: dmclerra(a)NO_SPAMfhcrc.org
Ph: (206) 667-2926
Fax: (206) 667-5977
---------------------------------------


--- On Sat, 11/21/09, tamuheyii <tamu(a)HEYII.COM> wrote:

> From: tamuheyii <tamu(a)HEYII.COM>
> Subject: fixed effect in Poisson regression of panel data
> To: SAS-L(a)LISTSERV.UGA.EDU
> Date: Saturday, November 21, 2009, 11:57 AM
> Hello All, I got this panel data with
> individual purchase, advertising
> and promotion to individuals.
>
> I can setup regular regression as
> purchase=intercept+ad+promotion. I
> want to test fixed effect by adjusting intercept to change
> across
> individual.. But I am not sure how to do this when the
> purchase is
> distributed in Poisson.
>
> Thanks for any suggestion. I did the normal poisson
> regression with
> nlmixed procedure, so if you can help me to integrate the
> fixed effect
> into nlmixed procedure, that would be very nice.
>
 | 
Pages: 1
Prev: SAS BI DI ETL BASE Clinical Financial + software
Next: 2 PART MODEL