STATS MAKE ME CRY

Hi there,

I am planning to do a mixed model analysis for one of my PhD research questions.

My outcome variable is continuous (movement of the bone) which has been collected repeatedly in five incremental test positions (i.e 10 degrees, 20 degrees, 30 degrees, 40 degrees and 50 degrees) for two different clinical groups. Thus, my independent variables are clinical group (0 , 1) and test position (10, 20, 30, 40, 50).

I would like to see if there is any between group difference in the trend of the movement of the bone (outcome) from 10degrees to 50 degrees. As my data are collected from the same participants in 5 test positions, they are correlated. Hence to account for the correlation in the data, I was thinking of using mixed models.

However, I am not sure if we could analyze for the trend between two clinical groups using mixed models? Further, the graph of my data are all over the place; and hence I do not think that there is a linear relationship between the outcome variable and the test position. Would this mean that I have to include a polynomial term in my model (such as: test position* test position). Further, to compare this polynomial trend between groups, would I have to define interaction between clinical group and the polynomial (such as: test position* test position * clinical group)?

I am all confused, reading about it and cant find a way.

Any suggestion or help would be truly appreciated.

Thanks and regards,
Divya.