I made some research on this and found http://arxiv.org/pdf/physics/0210022 which says basically that drop rates among 720 different carriers have a lognormal distribution with mean of about 2.4%.
iPhone 4 might have better RF and baseband chain to compensate, but it's clearly not enough when the phone seems to lose the signal altogether.
Edit: Jobs said clearly that the 4 drops ~1 calls more per 100 than the 3GS. 1 per 100 -> 2 per 100.
1 more is still a ton, an average user probably does more than 100 calls per month so... that's a drop call per month and for people in shitty area that could be multiple a week on the high end (some people never drop calls on the iPhone)
Jobs said: "One or less additional dropped call per hundred calls". I don't know how I should parse this to be "one percent more dropped calls than 3GS".
"It means that if the 3GS drops 1% of calls, the iPhone 4 drops 1.01% of calls."
"If the 3GS drops 1% of calls", it drops one call per hundred.
That means the iPhone 4 drops two calls per hundred, or 2% of one hundred calls. Not 1.01%.
Just to show I'm somewhat unbiased on the issue -- if we want to question Apple's data why did they show AT&T store returns and not Apple Store returns?