https://communities.sas.com/t5/Statistical-Procedures/Proc-mixed/m-p/455878#M23790 <P>It is okay. Pick the model you think is the best.</P> Fri, 20 Apr 2018 10:43:14 GMT PaigeMiller 2018-04-20T10:43:14Z https://communities.sas.com/t5/Statistical-Procedures/Proc-mixed/m-p/455814#M23789 Hello,<BR />I wanted to do proc mixed and I have 3 factors- treat group day. But while writing the model, I wrote e = treat group day treat*group group*day treat*group*day.<BR />I forgot to type treat*time. Because of this I got an significant interaction for treat*group*day. And the corrected by model I lost the significant difference for TGD interaction.<BR />Is it ok to use the model without treat*time? I already know that I have time effect. So for my results time is not a important criteria. So is that ok to leave out treat*time?<BR /> Fri, 20 Apr 2018 03:12:58 GMT https://communities.sas.com/t5/Statistical-Procedures/Proc-mixed/m-p/455814#M23789 Sri12 2018-04-20T03:12:58Z https://communities.sas.com/t5/Statistical-Procedures/Proc-mixed/m-p/455878#M23790 <P>It is okay. Pick the model you think is the best.</P> Fri, 20 Apr 2018 10:43:14 GMT https://communities.sas.com/t5/Statistical-Procedures/Proc-mixed/m-p/455878#M23790 PaigeMiller 2018-04-20T10:43:14Z https://communities.sas.com/t5/Statistical-Procedures/Proc-mixed/m-p/455882#M23791 Oh cool. Thanks for the reply. Fri, 20 Apr 2018 10:54:04 GMT https://communities.sas.com/t5/Statistical-Procedures/Proc-mixed/m-p/455882#M23791 Sri12 2018-04-20T10:54:04Z