See "blunders". See also here for OKN calibration error.
The eyes and the rotatory chair velocity are calibrated independently. Independent things can be mismatched. When there is a mismatch, there can be a peculiar pattern where mid-frequency VOR gain is too high. No human being can produce r-chair gain of 0.9 at 0.02 hz.
Note that the phase is normal. This means that there is a gain error, but timing is normal. If the phase were 0 degree at all frequencies, this could mean that there was instead a light leak.
This causes the eye to move more than it should for a given displacement of the chair. It is puzzling to us that the eye velocity does not exceed chair velocity.
Now the gain is low-normal, and phase is unchanged.