Help using Base SAS procedures

comparing main effects in factorial model using NLMixed contrasts

Reply
N/A
Posts: 1

comparing main effects in factorial model using NLMixed contrasts

I was wondering what the proper syntax is for comparing main effects in a fully factorial model in NLMixed?   The model (with generic variables substituted for the real ones) is below.  In this case, x1a and x1b are two levels of a 3-level categorical variable (study site), and same goes for x2a and x2b (temporal seasons).   The reference levels are x1c and x2c.

eta = b0 + b1*x1a + b2*x1b + b3*x2a + b4*x2b + b5*x1a*x2a + b6*x1a*x2b + b7*x1b*x2a + b8*x1b*x2b;

mu = exp(eta);

model y ~ poisson(mu);

Is the right way to use the contrast statement for a "global test":

contrast 'global' b1,b2,b3,b4,b5,b6,b7,b8

How do I code contrasts for the main effects tests given that there are interactions in the model?  These are the tests I'm interested in:

    x1a = x1b = x1c   (site effect)

    x2a = x2b =  x2c (season effect)

And how do I code paired contrasts, for example

  x1a + x1a*x2a = x1b + x1b*x2a (difference between sites 1 and site 2 within season 1)

 

thanks

Jonathan

Ask a Question
Discussion stats
  • 0 replies
  • 121 views
  • 0 likes
  • 1 in conversation