Nevada Early Voting--Dems hold edge

[WI_DEM]

According to the secretary of state, Democrats accounted for 162,801 in-person early voters - or 43 percent of the total - just over the party's share of registration. The GOP turned out 156,264 early voters, or 41 percent of the total. That's 4 percentage points higher than the party's registration in Nevada.

Democrats held a 21,502-to-19,087 edge in mail-in ballots for the state's two most populous counties, Clark and Washoe. Statewide figures for mail-in ballots were not immediately available.

Together, the final tallies for two weeks of in-person voting and a count of the mail-in ballots for the two counties gave Democrats about a 9,000-voter edge. The slim margin stands out because Democrats hold a 60,000-voter edge in statewide registration, and Republicans typically turn out in higher percentages on Election Day.

Campaign officials on both sides predict a close finish but disagree on the interpretation of the early vote.

In a year when polls show Republicans are highly motivated, Reid spokesman Kelly Steele said Democrats held the GOP below its average turnout rate for midterm elections, a key measure going into Election Day. Also, because of higher early voting by the GOP, Steele said a higher Republican turnout on Election Day is less likely. "You can't vote twice," he said.

Republicans credit Reid for a strong get-out-the-vote operation but say the early vote shows Angle has momentum.

http://www.washingtonpost.com/wp-dyn/content/article/2010/11/01/AR2010110106332.html

[still_one]

1. Democrats better vote today /nt

[SuperTrouper]

3. I know. It is neck and neck in Nevada nt

[DarthDem]

2. Great News!

For John McCain.

Oh, and probably for Harry Reid.

[FBaggins]

4. Could be closer than expected.

The final poll from PPP had a sample that was 41%D and 40%R. This early turnout is 2% higher for democrats and 1% higher for republicans. They had Angle ahead by one, so this would imply a dead heat if you assumed that the early voting was in any way indicative of the overall turnout (a big assumption that I don't happen to agree with).

[Igel]

5. I don't believe the polls.

Well, at least not without reading and understanding the assumptions. Screw the error intervals: More important are the assumptions.

Many of the assumptions the numbers spouted include conflicting, incompatible definitions: If you don't know how people define their terms you really, really have no clue what they're saying. You may have erroneous beliefs about it, you may believe that you know them better than they know themselves, that they really aren't aware of their true meaning. Most of that kind of belief is below commenting on further.

No, you have to read the fine print.

The early voting party-affiliation numbers are by voter *registration*, whether your voter ID card says you're a (D) or a (R). You register as (D), you're (D)--it really doesn't matter what you call yourself this time around, it doesn't matter who you vote for. You're a (D). That's the label the Board of Elections has for you.

The polls' party-affiliation numbers are voter *self-identification." It really doesn't matter what your voter ID card says. Today you can self-identify as (R), tomorrow as (D), the next two days as (I) and the fifth day you can simply say that you're a evangelistic nihilistic gnostic Incan pandit with an omnipotent, all powerful hamster who you'll vote for via write-in ballot because he crawled into your brain during the night and reprogrammed your wiring. (We'll put that down as (L) for "loony party".) Which label you wear depends on which you put on as you pick up the phone.

The Inner Outer County precinct number -342 in Upper Southern North Dakota lists voters as 100% (D), with voter turnout as 79% while the Gallop-away poll gives it 94% for the (L) party (based on a likely massive infestation of hamsters)? I don't see the necessary contradiction. Any caviling *must* be based on whether or not that many (D) are likely to self-identify as (L) and not just on a mismatch when you count the different labels assigned by different entities at different times.

If the poll does not give registration cross-over numbers (after all, the BOE can't know until after the polls close, one assumes) then you don't know how off, how reliable, either number is in determining who's going to win or lose. Even then, the polling model has to try to account for people who say they've shifted parties but haven't, or haven't shifted parties but have, as well as people who really are (D) in registration and self-identification but are voting for (R) candidates (often considered the primary meaning of "Cross-over voters"). For example, in some states you reregister when you cast a primary ballot; in others, it doesn't matter what you're registered as, so people don't update their voter registration often at all. Since the 2008 election had high turnout and a lot of reregistering, while the last primaries had fairly low turnout, you can't know how many people reregistered as (D) (or (L) :-) ) and saw no reason to reregister this time around.

This is not true of just the most recent polls. This is true of all pre-election polls. As my math (chemistry, physics, and engineering) teachers/professors all said, you have to understand the number you get from your calculations--otherwise you have no clue what it means or if it's even possibly right. (Most recently pointed out to me by a physics TA who railed against a midterm test that had answers for how long it would take a metal ball to drop 7 meters than ran the gamut from 1x10^8 years to -3000 years, meaning that the ball started falling before the Big Bang *or* the person dropped the ball 2980 years before he was born.)