|
Edited on Thu Aug-07-08 02:22 PM by Kurt_and_Hunter
We elect presidents by state. True.
State polling is more indicative of the state of things. False.
It seems like that ought to be true, but it isn't. A gambler picking sides would rather have the national data than the state data if he could only have one.
Historically, national polling is a better predictor. It has to do with the nature of polling, not with how we chose presidents.
Compare electoralvote.com for November 1, 2004 (showing a Kerry landslide) to national polling on November 1, 2004 (showing Bush up by 2-5%).
State polls COULD be more predictive, but because it takes almost as many people to poll Rhode Island as to poll the whole country quality national polling by state is too expensive. Nobody does 50 state polls on the same day using the same methods and assumptions. So the accumulated state polls are a mix of apples and oranges.
The first thing you'd need to say national polls are not predictive (let alone 'meaningless') would be an election where the electoral college and popular vote didn't match. There aren't any since the pony express shut down. (Unless you want to argue that Bush got more votes in Florida in 2000.)
Since the correlation between popular vote victory and EV victory is 100% in the era of modern polling the discrepancy is theoretical. And since national polling is smoother and more accurate for what it measures than state polling is, the practical swamps the theoretical.
Another fallacy is thinking that current EV projections show Obama up by "more." When Obama leads by 5% in a national poll, that's an electoral college run-away. %5 is a lot. It sounds better to say Obama is up by 150 EV than 5% popular vote, but it's just using a different way of keeping score. The Electoral College exagerates margins. EV are easy come, easy go. A 100 EV lead sounds big, but can be as fragile as a 2-3% popular vote lead.
|