Dear R users, 

   my question concerns my interest in comparing the beta coefficients between 
two identical regression models in two  (actually 3) groups. Disclaimer: I am 
quite new to R, so I might be missing some terminology that I have not come 
across. 

  I am trying to find out if I can easily implement a Wald test in R for this, 
but the only relevant thing that I came across is this link 
(http://hosho.ees.hokudai.ac.jp/~kubo/Rdoc/library/aod/html/wald.test.html), 
which seems to be 

referring to comparing the coefficients of separate IVs within the same 
regression. What I am interested is to compare the coefficients of the same IVs 
across three regressions and across three groups.


 Some basic information about my model/design:
My model includes 2 fixed factors (task, group) and 5 more IVs of interest.

Choice ~ Task * Group + IV1 + IV2 + IV3 + IV4 + IV5

I established that the Task * Group interaction is significant (by running two 
models of a regression, one comparing a model with and one without the 
interaction term) and I am interested to explore group differences within each 
task 

in a separate logistic regression. Task has 3 levels, therefore I do 3 of the 
following regressions: Choice ~ Group + IV1 + IV2 + IV3 + IV4 + IV5, one for 
each level of Task. At this point I want to compare betas for the 5 IVs across 
my 3 groups 

(actually I don't even need to compare all 5, based on the theory behind it, if 
that is a concern for some in terms of groupwise errors).

PS: I did a z-test across my critical IVs between groups, which is quite 
similar to Wald. I just want to do the Wald as well to doublecheck my results.

PS2: I also did another thing that some people might suggest, which is to 
recode Group in 2 dummy variables and include those and the interaction terms 
within each of my 3 regressions. I could go one 

step back and do it in the original regression (Choice ~ Task * Group + IV1 + 
IV2 + IV3) but I want to avoid having to include the complicated 3-way 
interactions. Also, I don't think it is much of a problem
to analyze different tasks individually, especially after I have already 
established that the Task * Group interaction is significant.

   So, can someone provide the code or point me to a place that a similar issue 
might already have been discussed, for the case of doing a Wald test to compare 
the beta coefficients of specific IVs
of interest  between different regressions/groups (same model used for all 
thesegroups)?


 Thanks in advance,
Kostas

        [[alternative HTML version deleted]]

______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Reply via email to