I looked into the PR3 YAW motion during the iKAGRA test run in April.

The pdf file shows the seismic signals and Oplev signals of PR3. There are coherence at around micro-seismic peak.

Considering that the calibration factor of the seismometer is 750V/(m/sec), the seismic level at atound 0.2 Hz is ~8 x 10^-6 m/sec ~ 1 x 10^-5 m.
Assuming the PR3 oplev beam path length is 1 m, the PR3 mirror YAW motion at 0.2 Hz is about 2 x 10^-7 rad.
Therefore, the coupling ratio at 0.2 Hz is 0.02 rad/m.

I implemented the shift of the suspension point into the iKAGRA PR3 simulation model: SUMCON and calculated the transferfunction from the seismic motion to YAW motion. (See jpg file.)
When I simply shift the suspension point of IM from the COM by 1mm, the IM tilt by 0.04 rad (2.3 deg), and the coupling rate is about 0.01 rad/m.

I am not sure that it is only because of this effect, but I think it can be a reference for the coupling estimation.