topic Re: Latin Square Design with two factors in Statistical Procedures

Latin Square Design with two factors

palolix — Mon, 29 Jan 2024 18:32:52 GMT

Dear SAS Community,

I have a Latin Square Design with two factors (4 treatments and 5 depths). I am not sure if my code is correct, so I would greatly appreciate if I could get some feedback. Thank you!

proc glm;
class row col trt depth;
model lgRing=row col trt depth trt*depth;
lsmeans trt*depth / adjust=tukey lines;
contrast '1 v 2 when depth=1' trt 1 -1 0 0;
contrast '1 v 3 when depth=1' trt 1 0 -1 0;
contrast '1 v 4 when depth=1' trt 1 0 0 -1;
contrast '2 v 3 when depth=1' trt 0 1 -1 0;
contrast '2 v 4 when depth=1' trt 0 1 0 -1;
contrast '3 v 4 when depth=1' trt 0 0 1 -1;
run;

Re: Latin Square Design with two factors

StatsMan — Tue, 30 Jan 2024 13:31:55 GMT

You may need to specialize some error terms here. This example in the GLM doc will get you started. PROC MIXED is a good alternative to GLM in these situations, too.

Re: Latin Square Design with two factors

palolix — Tue, 30 Jan 2024 17:01:06 GMT

Thank you very much for your feedback StatsMan,

The example was very helpful. In my case I have a latin square split block design with depth as the strip factor.

I tried using proc mixed but I didn't get any F values. Any idea of what it is wrong?

proc mixed;
class row col trt depth;
model lgRing= row col trt col*trt row*col*trt depth row*depth col*depth trt*depth col*trt*depth row*col*trt*depth/ddfm=satterthwaite;
run;

Thanks a lot!

Re: Latin Square Design with two factors

StatsMan — Wed, 31 Jan 2024 13:17:57 GMT

Posting your results and providing more detail about your data (replications, specifically) will help us diagnose the problem

Re: Latin Square Design with two factors

palolix — Thu, 01 Feb 2024 17:33:05 GMT

Thank you StatsMan for your reply. I attached the SAS code I am using. I have four replications (blocks) per treatment. I have also attached a file of the map.

Thank you very much for your help!