How can the media project

[Mass_Liberal]

with only 20% of THE FUCKING OFFICIAL NUMBERS!!!!!!!!!!!!!

[displacedtexan]

1. exit polls

eom

[Mass_Liberal]

2. exit polls aren't sure

not everyone is in the exit polls. Plus, they said the margin was very small in exit polls!

[Ripley]

4. Exit polls are notoriously WRONG.

Remember Florida?

[Edwards4President]

7. The exit polls in Florida were VERY accurate

that's why the networks initially called Florida for Gore. When they polled people on their way out, they told them who they THOUGHT they had voted for. It was only after their ballots got screwed up, lost, whatever, that the totals changed.

The exit polls were spot on. It was the official count that was inaccurate.

[Mass_Liberal]

9. but CNN

isn't showing the exit polls. It's just showing 30% of the reported #'s. WTF?

[Edwards4President]

10. You're right, they're just showing the actual vote counts

but their projection is based upon the exit polling.

[lovedems]

3. They didn't learn their lesson in 2000 I take it.

I wish their jobs could be outsourced.

[Indiana Democrat]

5. Because it's a proven fact...

...that exit polls are very reliable. These people know what they're doing.

Just take note of them now and compare them to the final numbers. You'll never ask this question again.

[leyton]

6. The percentages have stayed consistent,

from the beginning to the current status in New Hampshire. It was the same in Iowa - the results early on were pretty close to the results later in the evening (although Kerry pulled away).

[Adenoid_Hynkel]

8. because it worked so well in 2000

with no problems whatsoever

none at all

not a one

really

it went swell

[Edwards4President]

11. The exit polls weren't the problem in Florida

they were very accurate - that's why they projected Gore the winner. The actual count was skewed by Republican shenanigans.

[unblock]

12. 20% is a HUEUEUEUEGE sample!

let's say that the total tally is that out of 1,000,000 people voting for either dean or kerry, dean gets 500,001 and kerry gets 499,999.

well, then, you would expect that randomly picking 1,000 voters, the split would be somewhere around 50-50%. you wouldn't expect to see 100 kerry to 900 dean unless there was something unusual about the sample, e.g., they're all from a town kerry spent way more time in.

well, now consider a sample of 200,000 dean-or-kerry voters. if dean really is going to beat kerry statewide, what are the odds that, out of the first 200,000 counted, kerry has a 15% lead over dean?

is it possible? sure, especially if the first precincts reporting were kerry's favorite haunts, while dean owns the rest of the state.

but from a statistical perspective, it's not bloody likely.

if kerry has a huge lead with 20% counted, odds are REEEEAAALLLY high that kerry won statewide.