# Thread: Three way anova for repeated measure

1. ## Three way anova for repeated measure

Hi all, I am struggling with how to properly fit a three way anova using STATA. It keeps giving me this error message "could not determine between-subject basic unit; use bseunit() option" and the solutions online do not seem to work for me.

Summarizing my data:

id = person identification
bloodpressure = outcome measure, it is continuous and normally distributed
gender = binary
arm = binary (right or left arm)
time = categorical (T1 T2 T3 T4)

basically after a stimulation only on the right arm (the left arm is a control) I am measuring the variation of perioheral blood pressure over time and assessing possible gender differences.
I have a total of 24 persons and for each person 4 measures for each arm ( 8 measures in total per person) for a total of 192 measures.

My problem is that the sample size is relatively low, I fitted a linear mixed model without many problems (but afraid of possible bias due to sample size) and therefore wanted to compare it with a three way anova with an interaction between time and arm

the code I used with STATA for the linear mixed model is the follow:

xtmixed bloodpressure i.arm i.time i.gender i.arm##i.time || id: , covariance(unstructured) mle var

followed by contrast time@arm, effect to study the effect size

The results seem quite clear to me but still afraid that the approach is not completely right that is why I would like to compare it with a three way anova for repeated measures

What would be the code in stata to do a three way anova with interaction between time and arm without having the error message? I have tried so many variations of the code but I keep having the error displayed above!

Many thanks in advance for the help!

2. ## Re: Three way anova for repeated measure

Maybe you'd better ask this in the STATA subforum?

With kind regards

K.

3. ## Re: Three way anova for repeated measure

Yes, you are probably right but I don't know how to move the topic (probably only the admins can)

4. ## Re: Three way anova for repeated measure

Simply post this again in the STATA forum and mention there
that you had initially posted here.

With kind regards

K.

5. ## The Following User Says Thank You to Karabiner For This Useful Post:

Ralph84 (03-25-2015)

6. ## Re: Three way anova for repeated measure

Why do you think a RM ANOVA will deal better with small samples than a multi-level model will?