topic Re: How different is the Proc Mixed code from a complete block design vs an incomplete block design? in Statistical Procedures

How different is the Proc Mixed code from a complete block design vs an incomplete block design?

marcioadg — Fri, 18 Jul 2014 17:29:21 GMT

I went through 2 different books and they are not too clear because they are testing other things (slopes,etc) at the same time. How different is the Proc Mixed code from a complete block design vs an incomplete block design?

Thanks,

Marcio

Re: How different is the Proc Mixed code from a complete block design vs an incomplete block design?

PaigeMiller — Fri, 18 Jul 2014 17:36:58 GMT

I suppose the answer depends on how you code the two different analyses ... this is a very vague and undefined question

The answer to "how different" also depends on whatever you choose as a scale of differences, and I am not familiar with such a scale

Re: How different is the Proc Mixed code from a complete block design vs an incomplete block design?

marcioadg — Fri, 18 Jul 2014 17:57:36 GMT

Thank you Paige;

Here is how I would code as a complete block design for this specific experiment.

PROC mixed data=growth ;

CLASS Trt block;

MODEL response = Trt;

Lsmeans Trt/diffs;

random block;

How would one code if it was an incomplete block design?

Thank you

Re: How different is the Proc Mixed code from a complete block design vs an incomplete block design?

lvm — Mon, 21 Jul 2014 13:40:04 GMT

The code is exactly the same. That is the beauty of mixed model procedures like MIXED. Anyone interested in mixed models (i.e., models with multiple random effects) should read SAS for Mixed Models, 2nd edition (2006).