No more SliceLines output in Glimmix?

[dsuchoff1]

I recently upgraded to SAS 9.4 m6 and for some reason am unable to get the SliceLines output that I was able to prior.

Basic code is as follows:

proc glimmix data=data nobound plots=studentpanel;;
class block trt week date;
model aphidmean=trt|week/ddfm=kr2;
random int/subject=block;
random week/ subject=trt*block type=ar(1);
lsmeans trt*week / adjust=tukey lines;
slice trt*week/sliceby=week lines adjust=tukey;
run;

Prior to the upgrade I was able to get results sliced by week, with tukey lettering for individual week time points. This is not not available. Is there an error in my code or has this been changed in the upgrade?

Thank you,

David

[Reeza]

What SAS Stat version are you on now? This would show it.

proc product_status;run;

Slice is still listed in the documentation. Is there any issues in your log or anywhere else?

Did this exact code work before?

---

@dsuchoff1 wrote:

I recently upgraded to SAS 9.4 m6 and for some reason am unable to get the SliceLines output that I was able to prior.

Basic code is as follows:

proc glimmix data=data nobound plots=studentpanel;;
class block trt week date;
model aphidmean=trt|week/ddfm=kr2;
random int/subject=block;
random week/ subject=trt*block type=ar(1);
lsmeans trt*week / adjust=tukey lines;
slice trt*week/sliceby=week lines adjust=tukey;
run;

Prior to the upgrade I was able to get results sliced by week, with tukey lettering for individual week time points. This is not not available. Is there an error in my code or has this been changed in the upgrade?

Thank you,

David

---

[dsuchoff1]

Base SAS Software 9.4_M6

Image version information: 9.04.01M6P110718

And yes, this used to work prior. The output generates the Trt*week F test, Trt*week diffs, and trt*week diffogram.

David

[Reeza]

If you include some sample data or change the code to work with other data we can run and test it and help you out better.

You check if ODS GRAPHICS ON; is set properly, if you're not getting any graphs that could be why.

[dsuchoff1]

Data attached

I noticed this warning in the log:

WARNING: MIVQUE0 estimate of profiled variance is linearly related to other covariance parameters

Here's the code:

proc glimmix data=data nobound plots=studentpanel;;
class block trt date;
model aphidmean=trt|date/ddfm=kr2;
random int/subject=block;
random date/ subject=trt*block type=ar(1);
lsmeans trt*date / adjust=tukey lines;
slice trt*date/sliceby=date lines adjust=tukey;
run;

[Ohiopp20]

Try adding LINESTABLE into your slice statement.

[Rick_SAS]

In SAS 9.4M5, a change was made to the LINES plot. You can read about it in 

"Introducing the new SAS/STAT lines plot"

and

"Graphs for multiple comparisons of means: The lines plot"

As mentioned in the second article, the LINES plot can be misleading if you have unbalanced groups. In my opinion, the diffogram is a more faithful visualization of the test. The diffogram is explained in this article.

By the way, if you want the old "table with letters," you can get it by turning off ODS graphics. But I don't recommend that tabular display b/c it can be misleading.