Including all treatments in crossover design mixed model in SAS

[ph6]

Hi, I have a dataset effects of 3 treatments against ctrl on cell-lines collected from 4 donors.

```
Treatment donor value log_val
Ctrl D1 xx
T1 D2 xx
T2 D3 xx
T4 D4 xx
```
I would like to compare the effects of treatments against ctrl and also with each other using a mixed model. I use the following mixed model in SAS:
```
PROC MIXED data=data;
CLASS DONOR TREATMENT;
MODEL log_val= TREATMENT/ DDFM=SATTERTH;
random DONOR ;
LSMEANS TREATMENT / PDIFF CL E;
estimate 'T/R' TREATMENT -1 1 / cl ;
run;
```

When I include all the treatments in one model I am getting higher standard errors than the case if I include only two treatments which I want to compare for different combinations of treatments. Which approach should I choose? Including all the treatments as once or including two treatments and repeating for different combinations?

[blueskyxyz]

adjust your model

1. add visit information if the dataset has it, like week information

2. include week in part of repeated statement

3.treat week, treatment and treatment-week interaction, baseline of val... as fixed effects

4. use cs, ar or un matrix to converge

.....