Why does CNN always use the 'likely voter' numbers...

[truthpusher]

...instead of the 'registered voter' numbers. CNN just announced 48 Bush 46 Kerry for likely voters when in fact it is tied at 46 with registered voters. These people need to start explaining there numbers.

[mikehiggins]

1. Why does CNN show states as clearly for * when no one else

does?

Why oh why?

I just can't figure it out.

[high density]

2. They want Bush* in the lead.

The race is over if they can't keep him in the "win" column. This has been made clear by their use of Mason-Dixon polls last week. Probably this week they'll have to rely on Strategic Vision polls.

[FloridaPat]

3. They are actually two different groups. Likely voters are ones who

have said they will probably vote in the election and are registered. They also make the poll more in Bush's favor.

Theoretically it would give closer results to actual reality of what should happen. But they there are so many other variables which no one is talking about - percentage of repubs, Dems, African Americans etc involved in the polls.

[H2O Man]

8. That's right, althought it goes a little farther ....

They count people who have been registered for at least 4+ years, and who have voted in elections on a regular basis. This eliminates potentially large voting blocks, such as those under the age of 23, as well as people who only vote when they feel it is very important (such as to get rid of Bush), and minority groups such as Native Americans who do not traditionally vote (this could make a huge difference in several southwestern states, and in the Dakotas).

[Mountainman]

4. At some point they will have to admit Bush is losing, I just don't know

if it will be after the election or before. CNN has a greater number of repukes as likely voters then Dems and that is the only way they can show fuckhead in the lead.

[baldguy]

5. It gives them an excuse not to include newly registered Dem voters.

The polls are biased 5-8% toward the GOP anyway.

[central scrutinizer]

6. "Dewey Wins"

The famous 1948 photo of Harry Truman holding up the newspaper headline "Dewey Wins" was a vivid reminder that any poll is only as reliable as the population it samples. Public opinion doesn't mean squat - only those "likely to vote" should be polled and exit polls are the most reliable of all (at least they were, pre-BBV) since you are polling those who actually did vote.

[tomfodw]

7. Each poll tries to estimate who is most likely to vote

Each asks a series of questions designed to ferret out the true likelies. It's not just one question, "Are you likely to vote on Nov. 2?" it's several questions.

They also look at the percentage of each category who actually voted in 2000.

The problem is when they estimate smaller turnout this year in any category than actually voted last time.

Anyway, keep in mind that the polls are designed to sell newspapers or attract viewers. They're not intended to affect the race, although of course they do. The pollsters are not scientists, and they're not partisans - they're a bastardized hybrid. All of the, including the ones who report what we like to hear.

[magnolia]

9. I was under the impression...

...that "likely voter" was a person who voted in the last election.

[ThoughtCriminal]

11. There's something I rarely see

I think they usually ask "Who did you vote for in 2000?", but so far I have not seen the results of that question published.

Back when Bush Jr.'s approval ratings were up in the nose-bleed range 80%+), I noticed that the number who claimed to vote for him was much higher than it should be. Something like 70%, when we know that it should be under 50%. Some attribute this to dishonesty by the respondent (wanting to be associated with the winner) but it could also be very very poor sampling.

By now, a valid sampling of likely voters should show close to a 50-50 split on the Gore-Bush voters in 2000. Since most of the polls I've seen have a disportionate number of Republicans (38% vs 31% Democrats seems common), I would not be surprised if the sampling of likely voters is biased towards Bush. The 2000 Bush-Gore ratio would pretty much prove it one way or the other.

[undercover_brother]

10. CNN will be showing Gallup polls of Bush ahead until December

[atre]

14. A couple weeks ago

Gallup had Bush down 1 to Kerry and Zogby had Bush ahead 4. Notwithstanding the fact that they PAY for their name to be affiliated with the Gallup numbers, they led with the Zogby numbers on all of their programs.

CNN is unabashed in their whoring for Bush.

[louis c]

12. Are those Gallup's numbers today?

[atre]

13. They DON'T

They use whatever numbers are more favorable to Bush.

Follow mediamatters.org or watch it yourself and you'll see unabashed *-whoring.