Dear Friends,
When I run the PROC MIXED procedure, in the SAS 9.4 result, I found the “Non-est” for one group (in my set of data, group 2). Please let me know how to solve this problem? Thank you very much in advance.
data chew_act_cov;
input number Name$ day group RUMINATETIME EATTIME TOTALCHEWINGTIME RUMINATECHEWSPERMINUTE;
cards;
2 Kiana 1 1 512.33 234.75 747.08 55.35
2 Kiana 2 1 557.6 238.55 796.15 54.74
2 Kiana 3 1 606.17 272.25 878.42 59.57
2 Kiana 4 1 573.43 185.9 759.33 55.16
4 Babsi 1 1 428.25 307.68 735.93 61.80
4 Babsi 2 1 567.3 397.55 964.85 70.08
4 Babsi 3 1 567.3 397.55 964.85 70.08
4 Babsi 4 1 592.58 414.98 1007.57 70.57
4 Babsi 5 1 538.13 409.6 947.73 60.41
4 Babsi 6 1 498.42 309.98 808.4 62.34
5 Utah 1 1 493.37 429.57 922.93 51.67
5 Utah 2 1 495.32 438.32 933.63 60.63
6 Uruguay 1 1 479.65 146.88 626.53 43.95
6 Uruguay 2 1 648.48 180.57 829.05 60.76
6 Uruguay 3 1 667.55 222.9 890.45 57.94
6 Uruguay 4 1 680.93 171.82 852.75 54.65
6 Uruguay 5 1 683.13 162.23 845.37 60.12
7 Polli 1 1 421.93 239.68 661.62 52.46
7 Polli 2 1 306.05 214 520.05 41.65
7 Polli 3 1 214.8 279.08 493.88 36.05
17 Prima 1 1 448.92 197.57 646.48 57.23
17 Prima 2 1 550.47 209.9 760.37 57.72
17 Prima 3 1 518.18 210.45 728.63 59.96
17 Prima 4 1 496.22 171.1 667.32 55.42
17 Prima 5 1 520.12 171.88 692 56.69
25 Zoey 1 1 88.1 180.3 268.4 19.34
25 Zoey 2 1 72.83 215.55 288.39 17.11
25 Zoey 3 1 78.57 192.08 270.65 19.26
26 Anna 1 1 592.82 328.95 921.77 74.45
26 Anna 2 1 530.55 345.15 875.7 68.13
10 Bella 1 2 539.95 455.317 995.27 55.27
10 Bella 2 2 565.1 435.9 1001 64.05
10 Bella 3 2 539.63 444.75 984.38 57.025
10 Bella 4 2 529.9 416.42 946.32 60.54
10 Bella 5 2 577.42 407.08 984.5 63.42
12 Mala 1 2 456.28 358.05 814.33 57.85
12 Mala 2 2 606.7 393.23 999.93 74.71
12 Mala 3 2 580.6 444.52 1025.12 75.04
12 Mala 4 2 538.93 372.2 911.13 65.26
14 Wendy 1 2 420.87 276.42 697.28 46.49
14 Wendy 2 2 586.95 349.73 936.68 68.89
14 Wendy 3 2 573.27 392.8 966.07 68.46
14 Wendy 4 2 534.65 363.15 897.8 62.34
18 Mona 1 2 430.42 300.08 730.5 56.42
18 Mona 2 2 577.92 402.03 979.95 70.58
18 Mona 3 2 633.47 332.7 966.17 63.35
18 Mona 4 2 635.35 378.65 1014 73.07
18 Mona 5 2 578.28 353.4 931.68 70.91
21 Peggy 1 2 453.73 347.95 801.68 42.48
21 Peggy 2 2 572.6 458.18 1030.78 66.99
21 Peggy 3 2 600.8 401.62 1002.42 64.48
21 Peggy 4 2 560.22 476.95 1037.16 61.47
21 Peggy 5 2 617.82 417.9 1035.72 58.71
22 Panama 1 2 355.95 410.68 766.63 46.36
22 Panama 2 2 508.4 414.63 923.03 1583.50
22 Panama 3 2 544.17 473.7 1017.87 59.46
22 Panama 4 2 578.25 391.25 969.5 74.88
22 Panama 5 2 580.02 418.43 998.45 68.03
24 Bianca 1 2 487.53 368.15 855.68 48.79
24 Bianca 2 2 514.52 480.73 995.25 51.77
24 Bianca 3 2 516.03 445.52 961.55 56.87
24 Bianca 4 2 514.88 387.35 902.23 56.00
;
proc mixed;
class Name day group;
model RUMINATETIME=group day group*day;
repeated day/type=ar(1) sub=Name(group);
lsmeans group;
run;
proc mixed;
class Name day group;
model EATTIME=group day group*day;
repeated day/type=ar(1) sub=Name(group);
lsmeans group;
run;
proc mixed;
class Name day group;
model TOTALCHEWINGTIME=group day group*day;
repeated day/type=ar(1) sub=Name(group);
lsmeans group;
run;
proc mixed;
class Name day group;
model RUMINATECHEWSPERMINUTE=group day group*day;
repeated day/type=ar(1) sub=Name(group);
lsmeans group;
run;
I am expand a bit on @PaigeMiller 's response. If you fit a model with main effects and an interaction, and you have a missing cell, the main effects will be non-estimable. However, if you fit a model that contains ONLY the interaction (a one-way model often called a "means model"), and use an LSMESTIMATE statement you can calculate main effect means and standard errors. Main effect F tests can also be calculated within LSMESTIMATE statements as well, using the JOINT and/or FTEST options.
SteveDenham
There is an empty cell, with 0 data points, where day is 6 and group is 2. Interactions cannot be estimated when there are cells with 0 data points; and in addition some of the LSMEANS are non-estimable.
Possible paths forward is to remove day 6 from the analysis, or remove the interaction from the model.
I am expand a bit on @PaigeMiller 's response. If you fit a model with main effects and an interaction, and you have a missing cell, the main effects will be non-estimable. However, if you fit a model that contains ONLY the interaction (a one-way model often called a "means model"), and use an LSMESTIMATE statement you can calculate main effect means and standard errors. Main effect F tests can also be calculated within LSMESTIMATE statements as well, using the JOINT and/or FTEST options.
SteveDenham
I am interested in producing main effect F Tests within LSMESTIMATE using the joint option. Could you please point us to some example code?
@Gimlet wrote:
I am interested in producing main effect F Tests within LSMESTIMATE using the joint option. Could you please point us to some example code?
@Gimlet please start a new thread and explain your question and explain the data, in detail, from the beginning. People will be happy to help you in a new thread which contains a clear explanation of your problem/question. Do not discuss this further in this thread.
Dear Paige Miller,
Thanks for your response. The first path didn't worked, after removing the day 6 for one of cow, but I also had non-estimated error.
Fortunately, the second path you suggested was worked. I removed the interaction in the model and then get the results. Thank you very much.
All the best,
Mansour
Registration is now open for SAS Innovate 2025 , our biggest and most exciting global event of the year! Join us in Orlando, FL, May 6-9.
Sign up by Dec. 31 to get the 2024 rate of just $495.
Register now!
Learn the difference between classical and Bayesian statistical approaches and see a few PROC examples to perform Bayesian analysis in this video.
Find more tutorials on the SAS Users YouTube channel.
Ready to level-up your skills? Choose your own adventure.