EDIT: I kind of buried the lede, I think, so I'm presenting it here in a revised intro/headline:
In fact, every single newly registered voter that Team Obama has signed up, anywhere in the country, over the past couple of years is automatically disqualified as a Likely Voter under Gallup's methodology.There's a surge happening, President Obama's support is solidifying in swing states, stabilizing nationally, with a probable margin of about 4% to 5% over Romney (52-47) on the horizon. Ohio is solid, Wisconsin, Nevada, and Iowa as well, and odds are starting to favor Virginia, Colorado, and New Hampshire, with an outside shot at North Carolina. Why? Well, because a majority of Americans, and of voters, approve of the President, distrust Mitt Romney and the Republicans, and want to keep Barack Obama in the White House.
What might stop that from happening? If Democratic voters don't show up at the polls. If a big chunk of those out there who want and intend to vote for President Obama decide to be lazy, stay home, and blow off the election, while Republican Romney supporters faithfully carry out their electoral duties.
That's what the Gallup organization thinks will happen. More precisely, Gallup employs a methodology that inherently assumes that a large portion of the electorate will stay away from the polls, even including many who tell them that they fully intend to vote. Much has been said about Gallup's heavily conservative (in more ways than one) Likely Voter vs. Registered Voter methods and results, but I haven't seen any detailed discussion of exactly how Gallup applies its LV screen, and how it systematically excludes a great many people who are, in fact, highly "likely" to vote for the President.
I'll explain on Page 2.
Gallup's website helpfully provides a complete explanation of the method they use to screen LVs from RVs in their telephone interviews. In summary, they ask respondents 7 questions. Only those who give the "right" answer on 6 or 7 of those questions (with a few exceptions) are considered "likely" to vote. Even a score of 5 out of 7 gets you kicked out of their LV results.
What are the LV screen questions? To paraphrase:
1. How much have you thought about the election?
2. Do you know where to vote? (exception for mail-in voters)
3. Have you voted in this precinct before?
4. How often do you vote?
5. Do you plan to vote this year?
6. How likely are you to vote this year?
7. Did you vote in 2008? (exception for age 21 and under)
Notice, then, how this can work. You can be the most enthusiastic, dedicated voter in town, but you could still easily be kicked out of Gallup's LV pool. For example:
Q2: "Well, I just moved to this precinct, so I haven't found out where the polling place is yet.
Q3: "Like I said, no I just moved here, so I haven't voted here before.
Q4: "Sure, I always voted in my previous address"
Q5: "Yes, I'm definitely going to vote!"
Q6: "Totally sure, 10 on a scale of 10, I'll be voting!"
GONG! Nope, sorry, you're not a Likely Voter, you failed questions 2 and 3. How many thousands of Obama voters (especially young people, poor people, inner city folks) fit this description?
Q1: "I think about the election every day, I read Daily Kos religiously."
Q2: "Sure, I know exactly where to vote, I've got it all staked out."
Q3: "Well, no, actually I just registered to vote this year, because I got fired up about the state of our country."
Q4: "Well, I haven't voted before, because I just registered, but now I'm totally excited about voting for the first time."
Q5-6: "I'm definitely, absolutely, unequivocally going to vote, for sure!"
Q7: "Um, didn't you hear what I said before?"
BEEP! Another response for the RV scrap heap! This one only scored a 4 ... total non-voter. In fact, every single newly registered voter that Team Obama has signed up, anywhere in the country, over the past couple of years is automatically disqualified as a Likely Voter under Gallup's methodology.
It turns out, you see, that Gallup's "Likely Voter" screen isn't really trying to measure people's "likelihood" of voting; it's measuring consistency of voting. Those who have been living in the same place, voting in race after race, are not only the "most likely" to vote, under this method, they're treated as the only ones who will vote.
Other pollsters are less obnoxiously convoluted in their methods: they generally just ask "do you plan to vote?" and one or two other questions, and accept voters' stated intentions as a reasonable proxy for their likely behavior. Gallup is trying to second-guess people's own assertions, by adding other hurdles: "Sure you say you're going to vote, but you didn't vote last time, did you? So why should we believe you this time?"
The net effect is clearly a bias in favor of demographics that lean Republican, particularly older, higher income voters, and those who stay in one place, and a huge bias against newly registered voters. Even if their screen correctly excludes a significant portion of potential voters who might indeed fail to show up on election day, this method doesn't leave any room whatsoever for those on their blacklist who actually will get out and vote. They apply no gradient of probability, no proportions: you're either in or out.
We've all seen the results. Gallup's LV tracking poll figures don't measure the true "likely" vote intentions or outcomes. They measure, at best, a worst-case, lowest turnout scenario, that inherently favors Republican fortunes.
What is "likely" to happen? The actual vote tallies will further tarnish Gallup's sinking reputation. And President Obama will be in office for another four years.