The parent comment says that sipping gives a different experience to drinking which would explain preference for Coke vs Pepsi in blind vs unblinded testing.

I argue that this doesn't make sense as an explanation, as blind vs unblinded testing is orthogonal to sipping vs drinking.

So, if this is the explanation, they'd have to set the testsa very particular way, for seemingly no reason (expept, perhaps, such manipulation of results).