Statisticians--does the early voting percentages help predict Tuesday?

[Lex]

In other words, statistically speaking, can what happened during the early voting extrapolate to the percentages we will see on November 2nd, in any way?

Looks like there was a BIG Democratic turnout in these early voting states and if that translates into what will happen on Tuesday, then we're in good shape!

[JuniorPlankton]

1. NO

You cannot make any extrapolations to Tuesday.
Unfortunately :(

:kick:

[Lex]

3. Well DANG.

Is there a reason why not? I'm just curious, not challenging the point.

[alcuno]

7. No, but it does make it easier for Democrats on election day.

They can tell who has already voted and contact those who have not. In Iowa, for example, Democrats are historically better at the absentee balloting thing. That gives them an advantage in the work ahead.

[endnote]

11. Basically, the people who go out and vote early may be VERY

different from those who go out and vote on Tuesday. It's not like those early voters were a RANDOm subset of the entire population. It's like the CNN online polls where there is self-selection. All the statistical tools become useless in this situation since they assume a random sample from the population of interest...

[TwilightZone]

2. In Iowa, early voting was on the plus side for Kerry, but polls of those

who intend to vote on election day were tied.

So, if that's any indication, then no, we can't extrapolate, but a tie on election day would favor Kerry since the pre-election day vote leaned heavily toward him.

[Lex]

4. I'm wondering if the people who 'intend' to vote are the same as

the pissed off Democrats who WILL, by God, VOTE--the motivation to vote seems strong in Dems, in other words.

[writes2000]

5. Just saw on Fox News, That Kerry Leads...

among those in Iowa who intend to vote on Tuesday.

I saw the article which said they were tied but it gave no numbers. Based on the Fox News guy, I bet Kerry is ahead even in the "intend to vote" but only by a slim margin. JMO.

[Awsi Dooger]

6. Normally, the GOP has a 10-15% edge in early voting numbers

This year that has been nullified, if not reversed. We also seem to be doing better among absentee ballots, traditionally a Republican advantage.

Still, I saw David Boies speaking from Miami the other night on CSPAN. He is a Democrat and said the party is not doing enough to encourage or facilitate early voting, especially in Florida.

[DemocratSinceBirth]

8. Awsie....

I want your opinion...


How confident are you that the composition of the electorate will be similiar to 92, 96 , and 00 specifically party ID...

In 92 it was even but in 96 and 00 the Dems had a three and four point advantage...


I think if the electorate is even slightly more Democratic we will be fine...

[Awsi Dooger]

9. That is one of two wild cards, along with percentage of undecideds

I've run Excel models with dozens of permutations. The partisan split is more significant than undecideds. In fact more than three times as decisive, given the tiny number of undecideds. Even if we get a huge 2/3+ share of undecideds a 50/50 Dem-GOP split will nullify that and beyond.

You're right about '92 and that is troubling, since the dynamics of that election are more similar to 2004 than '96 or '00 were. I wasn't following political math closely until '96, so it's difficult to backcheck and backfit and understand what changed. It's weird because that was a high turnout election with hostility toward an unpopular and out of touch GOP incumbent, so you would think the Dem partisan edge would be equal or higher, not zero. I may research that more tonight.

My base model has us with a 2-2.5% partisan edge nationwide, adjusted in various states due to historical trends and my best guess among new registers. When I look at my projected statewide raw numbers, the high populus states with a huge Gore edge in 2000 are the ones that drop significantly -- California, New York and New Jersey for example. We have lost registration percentage in states like that compared to 2000, and especially compared to battleground state numbers from 2000 to 2004. So a national partisan change may not be completely representative. In California, for example, our registration and party ID numbers jumped after '92 with the Pete Wilson decisions and Hispanic upswing.

You ask a very significant question, truly vital to this result but seldom queried on DU. Has there been a party identification shift in the direction Gallup detects? Certainly their projections are wildly exaggerated. But the GOP taking control of so many state houses and local offices makes you wonder. Again, I would think a fever pitch presidential race would draw out our numbers in full force. A GOP shift is more likely to be felt in midterms and obscure elections. If forced to estimate, I tend to believe our edge will be less than '96 and '00, making independents even more critical. My friend who is very astute in politics has looked at state registration numbers and theorizes a 10% block of newly registered independents is really more Dem leaning than standard independents, but for whatever reason has chosen not to officially identify with us.

[StupidFOX]

10. YES

Polls that show that Kerry is winning mean a morale-booster for Democrats, which translates to a better chance for a WIN on Tuesday