GENMOD AND MULTIPLE COMPARISON

[elizeusf210]

Hi all,

I am testing 24 compounds and accessing their mortality (%) under a pest. I am using the proc GENMOD with dist binomial and link logit. I would like to know if it is possible to make multiple comparisons, similar to Tukey in ANOVA, to compare these insecticides. Which PROC should I use? Thanks in advance.

Elizeu (beginner user)

[Rick_SAS]

I'm not an expert in multiple comparisons, but I think you can do this with the LSMEANS statement, which supports the ADJUST= option.

For more complicated comparisons, you can read about the LSMESTIMATE statement in Kiernan et al (2011) "CONTRAST and ESTIMATE Statements Made Easy: The LSMESTIMATE Statement "  The paper contains many examples. Maybe you will find one that looks like your data.

[elizeusf210]

Rick_SAS thank you so much. I will read this material. 

[elizeusf210]

"The option is not supported by the procedures that perform chi-square-based inference (GENMOD, LOGISTIC, PHREG, and SURVEYLOGISTIC)."

http://support.sas.com/documentation/cdl/en/statug/68162/HTML/default/viewer.htm#statug_intro

[Rick_SAS · Accepted solution]

You didn't say waht version of SAS you are using nor what your model looks like. That might make a difference.

Here is some sample data and code. Perhaps you can modify it for your needs:

data data1;
input Compound Success Trials Explanatory;
datalines;
 1 20 30 10
 2 10 18  8
 3 15 30  5
 4 20 55  5
 5  5 60  3
11 20 28  9
12 11 16  8
13 14 25  6
14 18 45  4
15  7 54  2
;

proc genmod data= data1;
class Compound;
model Success/Trials = compound /link=logit dist=binomial;
lsmeans Compound / adjust=tukey pdiff plots=diff;
run;

[Rick_SAS]

A search of the doc revelas that the phrase "The option is not supported by..." appears in the ADJDFE= option. Don't use that option. Use ADJUST=TUKEY  (or whatever other method you want).

[elizeusf210]

That is right. My mistake. Thank you for your patience!