Mems Gyro Mass Unbalance

i am interested to model the effects of acceleration on my mems gyro.

i'm using a kalman filter to estimate bias (using magnetometer observations).

i have modelled the bias drift as a random walk and this is working fine. however i understand the bias is also affected by linear acceleration (both directly and via a squared term). The data sheet quotes this as 0.1 deg/sec/g for the part i am using.

my questions are:

1) which axis affects which gyro (i.e. will the x axis gyro be affected by x axis acceleration only, or do all axis affect it to the same amount?

2) how do i interpret the 0.1 deg/sec/g figure? if i sample at 100 hz say and each sample i have a linear accel of 1 g on the gyro, then i get a figure of 0.1 deg/sec bias impact. Do i just add this to the bias i am expecting? so if its goes to -1g i then take 0.1 deg/sec off the bias etc. (in this scenario the gyro is essentially sensing the accel too). Or does this accel have an effect on the bias drift that persists after the accel force itself has gone?

Thanks for your help.