Hi friends!
I need your help, because I have got some doubts about my experimental design. Look, I’m unable to set a proper desgn/analysis to my research. Perhaps I got a solution, but I’m not sure and I want gather some experiences of you.
I have two groups (demented subjects and non demented subjects, both groups are properly balanced) and a researcher applied a test, composed by four different test dimensions, each one with a different number of item scores. And that is the main difficult. You’ll see…
I need a design with two groups (we can’t manipulate the mental condition so we don’t have two different treatments, but just two attributive conditions). This is the between-groups variable. Besides I considered each test dimension as different repeated measures. This is the within-groups variable. Given each test dimension had a different amount of item scores, I transformed each total score to proportions. So the total score could be compared among dimensions.
A raw graphical representation could be:
Groups measure 1 (test dim. 1) measure 2 (test dim. 2) measure 3 (test dim. 3)
Group 1 (dementia) Mean Y11 Mean Y12 Mean Y13
Group 2 (normal) Mean Y21 Mean Y22 Mean Y23
The “mean” that I could get is a mean of proportions.
After that, I applied my design: I chose an split-splot design.
The Question is: Are you agree with this procedure?
Thank you!