Robert and I discovered a bug in his plotting routing, so the degrees of freedom plotted can't necessarily be trusted. We've re-exported and replotted his data using the a set of scripts I've been using to plot and save all of the data,
/ligo/svncommon/SusSVN/sus/trunk/QUAD/Common/MatlabTools/plotquad_dtttfs.m
/ligo/svncommon/SusSVN/sus/trunk/QUAD/Common/MatlabTools/plotallquad_dtttfs.m

The results are the first attachment.

The second attachment shows this measurement of H2SUSITMY (the QUAD formerly know as X1SUSQUAD02), compared against three other measurements and a model. One should not expect the model to match the measurements, as it is one of the test mass chain, and the reaction mass chain has subtle different parameters due to the thinner reaction masses; it serves merely to guide the eye. The measurements, however, are of X1 QUAD02 BUILD03 (H2SUSITMY when it was on the MTS in the assembly area), X2 QUAD11 BUILD01, and X2 QUAD12 BUILD01, all of which should be ITM / TCP configurations of the reaction chain, and therefore have identical dynamics.

We see several things:
(1) Not only has H2SUSITMY name changed, but her dynamics have as well (Look at PINK vs. FOREST GREEN). 
     (a) Obviously the measurement magnitude is noisy, distorted and smaller across the measurement band, but more importantly 
     (b) the resonances have changed. For example, look at the typically-benign YAW transfer function.
(2) Curiously, the PITCH transfer function, which typically varies the most and difficult to measure, is quite clean.

First guesses:
- To explain the changed dynamics, there may actually be physical rubbing.
- To explain the broadband drop in magnitude, perhaps we're not driving as hard as we think?