Question About Default Means Separation in PROC GLIMMIX

[jbibb14]

Hello,

I am trying to figure out what the default separation method is when I run this PROC Glimmix code. Usually I specify it to run Tukey, but when I remove that specification from my code I wanted to know what method is the default that runs?

Code Example Below:

proc glimmix data=crapes;
class trt rep color;
model arl = color trt color*trt;
random rep;
lsmeans color trt color*trt/diff lines;
run;
quit;

Typically in the lsmeans line I have adjust=tukey, but as I stated I am curious as to what program runs the means separation when that statement is removed and it runs as typed above.

TIA!

J

[StatDave · Accepted solution]

This is covered in the description of the DIFF option in the documentation of the LSMEANS statement. By default, the DIFF option produces all possible pairwise comparisons of the LS-means without any adjustment using simple t-tests. You can specify optional values after DIFF= to request differences with the average LS-mean or differences with a control level. If you want mean comparisons to be adjusted, specify the ADJUST= option with one of the available methods.

[jbibb14]

I am still new to running everything with this software, in your opinion would there ever be a reason to run the code i typed without entering a diff or adjust parameter?

[PaigeMiller]

---

@jbibb14 wrote:

I am still new to running everything with this software, in your opinion would there ever be a reason to run the code i typed without entering a diff or adjust parameter?

---

Yes, of course you can run the program exactly as you wrote it, without other options. You can use different statistical methods by specifying other options, if you want. There are probably a dozen different ways to compare means that are useful in some situations, as explained in the documentation.

[SteveDenham]

Would there ever be a reason to not include DIFF=method>? Absolutely yes. All of the methods are designed to control type I error (false positives). However, suppose that it is far more important to control type ii error (false negative rate) in whatever field you are working. Then the unadjusted values give a greater ability to declare a difference significant. As an example, suppose you are working on a safety study for a new drug. In that study, it would be much more important to detect a true difference to protect you from declaring the drug "safe" when in fact it causes some change that may indicate a potential safety concern.

SteveDenham