Dear all,

I am currently struggling with my masters-thesis on the subject of control variables or covariates in a repeated measures analysis.

I have a rather strait-forward 2x2 data-matrix:

2 dichotomous independent variables. (I will call them condition & treatment)
1 dependent interval variable

In order to enlarge response, i developed a mixed design:

Group 1 condition [1] : Treatment 1 & Treatment 2
Group 2 condition [1]: Treatment 2 & Treatment 1
Group 3 condition [0]: Treatment 1 & Treatment 2
Group 4 condition [0]: Treatment 2 & Treatment 1

I tested for carryover effects, and found none, so finally my data-set looks like this:

l Treatment [1] l Treatment [2]
Condition [0] l Score X l Score y l
Condition [1] l Score A l Score B l


Now, i want to conduct an Repeated Measures ANOVA (Treament as Within Variable, and Condition as Between Variable), and everything is going well, until i came up with the idea to control for gender (again dichotomous variable).

Various people all over the internet struggle with the difference in covariates / control variables. Moreover, some people suggest that 'covariates' in SPSS can only be used for continuous variables, and that gender therefore must filled in the 'fixed factors' (together with my between variable).

On top of this all suggested my supervisor that for control variables, the model i had to run shouldn't include interaction effects, but that i had to run only direct effects.

I am a bit lost now, can anyone answer the following questions:

1) What is the difference between control variables and covariates? And what implications does this have in SPSS?
2) Is excluding the interaction effect a viable option in this case

Bonusquestion: When i tried to exclude the interactions from the model, it seemed that SPSS did not allow me to remove the interaction effect in the within-variable (it did allow me to remove it in the between-variable though, via 'custom model')

Estimated Karmapoints for answering this question: 315