From the peak-to-peak value of RF error signal of MICH (REFLAIR_A_RF17_Q_ERR), we estimated the modulation index.

1. DC power of reflection light (from one end) entering the RF PD
60 - 10 = 50 cnts (see the first attachment)
--> 50 * (10 / 2^15) = 15 mV
--> 15 / 100 = 0.15 mA  (transimpedance 100 ohm; https://gwdoc.icrr.u-tokyo.ac.jp/cgi-bin/private/DocDB/ShowDocument?docid=1280)
--> 0.15 / 0.7 = 0.20 mW (assuming the responsibility of the PD to be 0.7 A/W)

2. RF sideband power
amplitude = 10000 cnts (see the second attachment)
--> 10000 / 16 = 625 cnts (Whitening gain of 24 dB)
--> 625 * (10 / 2^15) = 190 mV
--> 190 / 11 = 17 mV (demodulator gain of 11; alog 10213)
--> 17 / 290 = 0.060 mA (transimpedance ~ 290 ohm; 4872)
--> 0.060 / 0.7 = 0.085 mW (assuming the responsibility of the PD to be 0.7 A/W)

On the other hand, laser power at f1 can be written as: 2 * beta * R_ETM * P_eff * sin (theta_asym) * sin (fringe) * cos (2pi*f1*t).
Considering we now  know R_ETM * P_eff / 4 = 0.20 W from 1. and sin (theta_asym) = 0.92, beta = 0.058

Note: JGW-T1605611 says EOM eff. is beta = 1.1 @ 27 dBm for f1, and now 13 dBm is injected to the EOM, so the modulation index should be beta = 0.2 at PSL room.
I have no idea for the reason of this factor of ~ 3 discrepancy.