Ls means - "Non-Estimated"

[Mathis1]

Hello 🙂

I lanched this code to perform a TUKEY test :

ods graphics on;
proc glm data=HAVE order=internal;
class X1-X15;
model TI_NUM = X1-X15 ;
lsmeans X2 / diff adjust=tukey;
run;
quit;
title;

And the Tukey test display this to me :

 I do have the matrix for difference of the LS-MEANS but it says the ls means are not estimated. 

I precise that when performing the proc glm without the statement "lsmeans" i do get all my coefficients... 

Thank you by advance for your help.

[PaigeMiller · Accepted solution]

"Non-est" usually happens because the model cannot fit all of the requested model terms, in this case X1-X15, often because of collinearity or missing cells.