That's Bill Schneider's take on it. Greenfield says it's probably more like 2%. So does Michael Beschloss. Of course it's mathematically possible to win 70% of the PV and still lose. But unlikely.
It's 83%. 
Mathematickky you can win with 11 votes (1-0 in the big 11 states ) and millions against you un all other states - 99.999%
Yes, but that's altering the number of voters. My example is based on turnout being the same which makes more sense.
Actually, turnout DOES vary a lot from state to state.
In 2000 Gore won New York by about 1.7 million votes, while Bush won Texas by about 1.3? million or so - making up most of Gore's margin in the PV, even though NY and Texas are about the same size (bush won texas by 21, Gore NY by 25)
Why the huge difference..? In 2000 there was a very hot senate race in NY (Clinton vs Lazio) while the Texas race was a yawner..
Here is a scenario.. On election day the Eastern seaboard gets clobbered by a huge snowstorm while the South has a real nice day...
I bet that would skew the popular vote at least a few %...
