Just to take it a little further. I have noticed the MOS is directly proportional to full rate proportion. I used a statistical software to extract correlation between MOS and various other network parameters extracted from the drive. The correlation with full rate came out to be .4 which is huge. I have not taken care of pre conditions for the statistical test so it may be inaccurate.
So:
Half rate increase -> Poor RxQual
Half rate increase -> Better MOS
But RxQual should be directly proportional to MOS!
I think it is due to subjective nature of MOS; it depends on so many factors. It may be that at the time of MOS drive the utilization was low. Des anyone has a better insight?