repeated mixed model

[Laura Hänninen]

Hello!

This could be very basic question, but nevertheless, if someone more
clever person could answer;

We have a dataset of individual animals, which got either treatment A or
B, over 5 consecutive days. We are not interested in the overall day
effect, but the treatment*time interaction is our main question.

Is it OK to write the syntax as below, or we need to allocate the day as
fixed effect as well

MIXED freq BY animal day treatment
  /FIXED= treatment day*treatment | SSTYPE(3)
  /METHOD=REML
  /RANDOM=INTERCEPT individual|
  /REPEATED=day | SUBJECT(calf) COVTYPE(AR1)
  /SAVE=RESID

Best,
Laura

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bruce Weaver]

Laura Hänninen wrote

Hello!

This could be very basic question, but nevertheless, if someone more
clever person could answer;

We have a dataset of individual animals, which got either treatment A or
B, over 5 consecutive days. We are not interested in the overall day
effect, but the treatment*time interaction is our main question.

Is it OK to write the syntax as below, or we need to allocate the day as
fixed effect as well

MIXED freq BY animal day treatment
  /FIXED= treatment day*treatment | SSTYPE(3)
  /METHOD=REML
  /RANDOM=INTERCEPT individual|
  /REPEATED=day | SUBJECT(calf) COVTYPE(AR1)
  /SAVE=RESID

Best,
Laura

When one includes an interaction term in a regression model, it is standard to also include all lower order interactions involving the same variables, and the main effect terms.  So I would include Day as a fixed effect.