Dear Colleagues,
I am using SAS to generate general linear models.  I have a complicated 
model design, but I will simplify it for this question.

We compared the effects of two different forest harvest treatments (each 
treatment created small gaps in the forest) on insects by collecting insects 
in several forest stands (three for each harvest treatment).

My model terms are
harvest treatment
forest stand within harvest treatment
canopy condition (= harvest gap v. close canopy)
insect species

To make sure the “harvest treatment” term is testing the correct question, I 
need to include a line in my model that asks the question specifically by 
setting the correct mean square error:
“Mean Sq between treatments” /
“Mean Sq between ‘forest stand within harvest treatment’”

When using PROC GLM, I can use the following test statement after the model 
statement:
test: H = treatment e = stand(treatment)

[i.e. test: H = treatment  e = “forest stand within harvest treatment”]

However, when using PROC GENMOD (with a negative binomial log link, see 
below), I cannot use this same test statement.  (I get a syntax error 
statement).

**Do you know how to do this same test (or something similar) in PROC 
GENMOD?**

Thank you,
Shelly


PROC GENMOD with a negative binomial log link:
proc genmod data = x;
class treatment stand canopycondition species;
model count = treatment stand(treatment) canopycondition species
/dist = NEGBIN link = log type3 maxiter=500;

_________________________________________________________________
Check out all that glitters with the MSN Entertainment Guide to the Academy 
Awards®   http://movies.msn.com/movies/oscars2007/?icid=ncoscartagline2

Reply via email to