Choice of test for detecting difference in methods for estimating muscle attachment

In my line of research, we are interested in determining how muscles drive body movements. To do so, we need to know where muscles attach on the bones for the subject we are studying. It is difficult to measure muscle attachment locations on live subjects, so we normally use a generic model of a human that has average muscle attachment locations (based on a cadaveric database). We determine the muscle attachment location of a subject by scaling the generic model to match the bone lengths of our subject. There are multiple generic models available (essentially the same, but based on different cadaveric datasets) and different ways to determine the scale factors (ie based on bone dimension from MRI or measurement through skin with tape measure).

I want to compare two generic models(a and b) and two scale measurement methods (1 and 2). I have 30 muscle attachment locations for 3 subjects measured experimentally through MRI, so I have computed the error between the experimentally measured and the estimated attachment locations for each combination (a1, a2, b1, b2), for all 30 muscles in all three subjects.

What test should I use to determine if there is a significant difference in the error between the scaling methods or generic models?

What test should I use to determine if I can predict one muscle's attachment location better (or worse) than the others?

I have taken a class on regression and ANOVA but it has been a long time, so I need some refreshing on what direction to take. I am thinking a two factor factorial with blocking by subject.

I have SAS available for statistical analysis.

Thanks in advance!
Last edited: