DATA Step, Macro, Functions and more

Mixed Model Type 3 Estimates

Reply
New Contributor
Posts: 4

Mixed Model Type 3 Estimates

Hi all,

 

I am trying to develop code to obtain the ANOVA and break out the main effects and interaction SS using complete sets of orthogonal contrasts. However, my program is not generating estimates for the interaction contrasts (it returns "Non Est."). Followed is my code, if anyone could shed some light, I would greatly appreciate it!

 

data aa; infile '/folders/myfolders/sasuser.v94/lab8.txt'firstobs=2;
input trt temp week y;
proc print; run;
proc mixed method=type3; class temp week;
model y = temp week temp*week;
estimate 'temp_lin' temp -1 0 1;
estimate 'temp_quad' temp 1 -2 1;
estimate 'week_lin' week -3 -1 1 3;
estimate 'week_quad' week 1 -1 -1 1;
estimate 'week_cub' week -1 3 -3 1;
estimate 't_l*w_l' temp*week 3 0 -3 1 0 -1 -1 0 1 -3 0 3;
estimate 't_l*w_q' temp*week -1 0 1 1 0 -1 1 0 -1 -1 0 1;
estimate 't_l*w_c' temp*week 1 0 -1 -3 0 3 3 0 -3 -1 0 1;
estimate 't_q*w_l' temp*week -3 6 -3 -1 2 -1 1 -2 1 3 -6 3;
estimate 't_q*w_q' temp*week 1 -2 1 -1 2 -1 -1 2 -1 1 -2 1;
estimate 't_q*w_c' temp*week -1 2 -1 3 -6 3 -3 6 -3 1 -2 1;
lsmeans temp week temp*week;
run;

 

Thanks,

Cody

Respected Advisor
Posts: 3,279

Re: Mixed Model Type 3 Estimates

NON-EST usually happens when you have a cell of the interaction with no data in it.

--
Paige Miller
New Contributor
Posts: 4

Re: Mixed Model Type 3 Estimates

Posted in reply to PaigeMiller

Well the treatment structure is a 3x4 factorial (3 different temperatures, 4 different time periods). Could the error be a result of the factors not being of equal levels? 

Respected Advisor
Posts: 3,279

Re: Mixed Model Type 3 Estimates

I'll stick with my first answer. The most likely cause is that you have an empty cell.

--
Paige Miller
New Contributor
Posts: 4

Re: Mixed Model Type 3 Estimates

Posted in reply to PaigeMiller

Any idea as to where this empty cell might be?

Respected Advisor
Posts: 3,279

Re: Mixed Model Type 3 Estimates

[ Edited ]

You have the data, I don't. You can find it easily enough using PROC FREQ.

--
Paige Miller
Ask a Question
Discussion stats
  • 5 replies
  • 106 views
  • 0 likes
  • 2 in conversation