Polls are using incorrect methodology.

[Zynx]

They are consistently overweighting Republicans. According to the results of the 2000 exit polls 39% of voters were Democrats, 35% were Republicans, and 27% were independents. Those numbers don't add up to 100% due to rounding. Now, if we apply party breakdowns like we have seen consistently in polling we get something like this:

Democrats 39%- 90% for Kerry, 8% for Bush 2% for Nader
Republicans 35%- 91% for Bush, 7% for Kerry, 2% for others(including Nader)
Independents 27%- 47% for Kerry, 40% for Bush, 13% Undecided or Nader

That comes out to 50.24% for Kerry, 45.77% for Bush, and the rest are undecided. Basically both parties' partisans will vote 90% for their candidate and we have seen such numbers in polling so far, give or take a percent. Sometimes Kerry's lead among independents is larger than what I credited him for, but I think it averages out to about 7%.

[lancdem]

1. Great post

Also, there may even be a bigger margin of Dems in 2004 because the party is so energized.

[Zynx]

2. There are more Dems than there were in 2000 now I think.

So you would be right.

[Feanorcurufinwe]

3. Yep

thanks for reminding folks.

[Awsi Dooger]

4. Party affiliations can be misleading

In the South, especially, where so many elderly registered Democrats have voted nothing but straight Republican for decades. In Oklahoma, for example, there is a considerable registration edge for Democrats but we don't even dream of contesting the state in presidential terms.

Still, I study poll results and agree the methodology is flawed, partially due to new technology like Caller ID. The response percentage is considerably less than it was a decade or more ago. IMO, the sample size is far too low and the supposed 3-4 point margin for error is glorified bullshit.

My pet theory is PAN, or Partisan Adjustment Number. I treat each state individually and analyze how actual voting results differ from final polls, in presidential and senate elections. When there is a distinct trend that indicates Georgia, for example, always sways several points more Republican than polls suggest, I adjust future Georgia polls accordingly. Right now I have Georgia 4 1/2 points more Republican than base. It's not a finished product due to limited number of election results, but incredible how many states differ in one direction EVERY time, yet no one every mentions it in the mainstream media.

[Zynx]

5. In 2000, only 11% of Dems voted Repuke.

http://www.abcnews.go.com/sections/politics/2000vote/general/exitpoll_hub.html

A lot of those "registered" Dems call themselves Republicans on election day.

[Awsi Dooger]

8. I made a faulty assumption

Your original post never mentioned registration, but exit polls. I thought the 39-35 numbers would have been a simple, "What party are you registered with?"

"No matter how you voted today, do you usually think of yourself as a..."

That's an interesting question. I would really like to see it broken down and compared with a followup question, the actual party affiliation.

[DeadHead67]

6. AGAIN, Show me the money ! I'll show you the votes. . .

Kerry/Edwards has raised record amounts of money in very short time, and this is LARGELY NOT 'Big Corporate' money, it's from joe average. The assholes sitting in the board rooms only get to vote once, and the number of people voting with their wallets and checkbooks is, I feel, very telling.:kick:

[Tarheelhombre]

7. And 2000 was a year wherein Democrats were not energized

[Dick_Tuck]

9. There's more under the radar than you think....

"Likely Voter" algorithms use things like did the person vote in the last election and is the person of an age/color/ethnicity that doesn't usually vote.

On the one hand, you have the Stern Schwing voters who don't vote unless Howard tells them to, e.g. NY Pataki v. Cuomo and NJ Whitman v. Florio. Then you have the 18-24 year olds, who only vote when there's a free drink. Both under-the-radar groups aren't going to show up in anything other than "registered voter" polls, where Kerry is dope slapping Bush all over the place.

[Zynx]

11. Likely voter models are horribly flawed.

[Dick_Tuck]

12. I'll give you that...

only in this cycle. Zogby's algorithms seem to look at voter anger. He got it right in '94 (bad for our side), and was the most correct in '96 and '00.

When less than half the people vote, and 1,000 respondents give you 95% confidence that the sample represents people +/- 3 points, they better be adjusting.

So, I don't knock the polls. They are what they are. In this cycle, they tell a great tale. Kerry leads Bush, in the major polls by about 2 points. However, Stern Schwing voters,young voters, and recent immigrant voters are the ones energized in this cycle.

If I were to set book right now, I'd put the over/under at Kerry by 4.

[King Coal]

10. Did you ever take a stats class?

Polls are made to be skewed.