John Harwood gives some insight into NBC/WSJ LV screen

[smorkingapple]

From his Twitter...

NBC/WSJ shows 47-47 Obama-Romney tie w/likely voters (O +8 women, -10 men). ABC/WashPost shows 49 O, 48 R (O+14 women, -12 men). Why?

margin-of-error sampling is one answer.

Another: different definitions of "likely voter". NBC/WSJ asks 3 things: vote in in '08? '10?

is election interest 9/10 on 1-10 scale? yes to 2 of 3 = likely voter. (7/8 OK if too young to vote earlier). ABC/WP method different

Makes no sense to me. I have to have both voted in 2010 and say I'm at least a 9/10 interested in this year's election to be considered a likely voter?

What if I don't give a shit about any race other than Presidential? Off cycle election years are infamous for lower turnout for this very reason.
What if I'm only 5/10 interested in this years race but still plan to vote? How do people even know how to answer this question? Do they expect every likely voter to be as informed and engaged as we are? What about those that treat it like a every 4 year tradition but really don't care to watch the news every night?
They admit they're excluding young people who maybe wanted to vote for Obama in 2008 but couldn't.

Now I'm starting to see why the polls are so wildly off this year. It seems they're only capturing voters who've consistently done it in the past or register as extremely interested in the race. Helps explain Romney's bounce from 1st debate as Republicans will now start to pass LV screens while Dems might have dropped off screens. Also helps explain the RV-LV disparity we see.

Thoughts?

[hoosierlib]

1. Yep!

We are up about 3% nationally and after this debate we will be up 5% - 6% come election day. African Americans, Latinos, Women and moderates will come out in droves, we will keep the Whitehouse, the Senate and come damn close to taking back the House!

[WCGreen]

2. When I ran a few polls back in the 90's, I aways used the universe of likely voters to be...

the last even year election and one other vote say in a primary or munipal election.

Granted, it was easier back then to get that information from the BOE and overlay the phone book.

But to me, that is what the definition of a likely voter should be.

It would be my guess that most people who would vote in 2010 would also vote in 2012.

[regnaD kciN]

3. I seem to recall a lot of progressive Democrats talking about not voting in 2010...

...in order to "teach a lesson" to Obama for abandoning the public option/not immediately withdrawing from Afghanistan/not filing war-crimes charges against Dubya/etc. By NBC's standards, it seems like most such people would automatically be considered "unlikely" -- I doubt there'd be any such situation on the Republican side.

[geek tragedy]

7. No, it was mostly demographics--young voters just don't vote in off-year elections.

Frustrating but true. If Democrats coalition showed up every 2 years regularly we'd never lose.

[Thrill]

4. These LV models just seem shady to me

.

[TexasCPA]

5. I don't think an incumbent can do good in that model

The supporters of the incumbent are naturally less enthused because they are not as hungry for a win. They will still vote though.

Who cares what Gallup thinks? The only poll that counts is the actual vote.

[VirginiaTarheel]

9. But New ABC poll says Obama supporters are more enthusiastic than Romney voters

and the huge early voting numbers reflect that is true.

[geek tragedy]

6. LV screen is never perfect. A lot of unlikely voters still vote, the

questions are (1) do the likely voters who don't actually vote favor either candidate and (2) do the unlikely voters who do vote favor either candidate.

[woolldog]

8. I'm not a likely voter according to their screen.

I didn't vote in 2010 and my interest in the election is only 7/10