Scenario:
Experimental Design: RCBD with 3 blocks.
Response Variable: Biomass Wt.
Independent Variables: Treatments (2 levels), Time (1...8), Samples (4 samples per treatment per collection time).
SAS Procedure: Repeated measures in PROC MIXED.
1. Why are the model estimated standard errors of the means (estimates) equal at respective time (e.g. at time 1, ... at time 8)? For example,
Effect Treat Time Estimate Standard Error DF tValue Pr>t
Treat*Time 1 1 266.14 79.7816
Treat*Time 2 1 198.33 79.7816
. . .
. . .
. . .
Treat*Time 1 8 986.49 88.2263
Treat*Time 2 8 703.17 88.2263
2. Why are the arithmetic mean (excel generated) and estimate of the mean very comparable, while model estimated standard errors and arithmetic standard errors (for lack of an appropriate word) not comparable?
Treat Time Arithmetic Mean Standard Error
1 1 266.14 43.00
2 1 198.33 39.00
. . . .
. . . .
. . . .
1 8 986.49 94.00
2 8 703.17 79.00
3. Which standard error values should be used when publishing results?
Thank you.
