This is a plausible explanation, but it would not be for AA, it would simply be for acceptable FDA filing, AA or not. I wrote to IR for an explanation of the differing mean changes. It could be a change in subpop (eg mITT, which can change the LSM) the type of t-test, etc. If arithmetic mean (instead of LSM) was originally provided by the CRO, that would be a shocking error in judgement. The company should not have left the change unexplained; their disclosure quality needs improvement. But I'm sure Jin is now reviewing all stats, so he might have caught something.
It was good to get clarity around the p-vals. This time around, from the given CIs it's clearly an appropriate 2-tail z-test. All hail Dr Jin.
News 
Market Data 
Discover 
