topic Re: lsmeans proc mixed in Statistical Procedures

lsmeans proc mixed

jescam05 — Fri, 08 Jun 2018 17:27:25 GMT

Hi, I runed this invocation, and run very well, just I want to obtain the Tukey's results with literals and not like appear.

data CEN; 
DO PG=1 to 2;
DO PCH= 1 to 4;
DO BLOQ=1 to 4;
INPUT Y@@;
COMPGPCH= PG*10+PCH;
output;
end;
end;
end;
DATALINES; 
9.93	10.58	9.94	10.18
10.75	9.71	10.53	10.26
.	.	.	.
10.52	10.36	10.35	10.16
12.29	12.06	12.06	11.88
12.00	12.55	12.82	12.52
12.52	11.92	11.98	11.73
12.09	11.88	11.83	11.96
;
proc mixed data=CEN method=reml; 
class pg pch bloq; 
model y= bloq pg bloq*pg pch pg*pch/ ddfm=satterth;
random bloq; 
lsmeans pg pch pg*pch/ adjust=tukey; 
run;

Re: lsmeans proc mixed

sld — Sat, 09 Jun 2018 03:18:06 GMT

As I've suggested in another of your threads:

https://communities.sas.com/t5/SAS-Statistical-Procedures/LSMEANS-LINES-option-in-PROC-GLM/m-p/466947#M24285

"you really, really need to read SAS® for Mixed Models, Second Edition"

In GLM, random effects are included in the MODEL statement. In MIXED, random effects are not included in MODEL; instead random effects are specified in RANDOM or REPEATED statements. Even if your model runs apparently well in MIXED, your specification is not correct, and neither are the results. Mixed models are complex, and you must devote effort to learning enough to apply them correctly.