Battleground Poll-Obama (D) 49% -Romney ($) 48% -ONE POINT CHANGE*

[DemocratSinceBirth]

http://images.politico.com/global/2012/10/politico_gwbgp_trackingoct8_questionnaire.html


When you factor in the negative fallout from the debate a one point change is encouraging. A bit less encouraging is the fact that major parts of the base such as Latinos and young people need to be energized.

All things considered it could be a lot worse.

[ananda]

1. It should get better this week after Biden debates.

nt

[DemocratSinceBirth]

2. The Battleground Poll Despite Being Bipartisan Leans A Bit Right

I don't remember every year but in 00 they had Bush* by 5 and in 08 they had Obama by only 3.

[FBaggins]

3. Not encouraging at all.

More than half of the interviews were made prior to the debate and republicans continued to show the enthusiasm gap (though smaller) that slaughtered us two years ago.

What sucks is that this is my favorite national poll.

I haven't seen yet whether they break out pre/post debate numbers to take a swing at any kind of "bounce".

[DemocratSinceBirth]

5. Why Is It Your Favorite?

They had Bush* beating Gore by 5% in their final 2000 pre-election poll:

http://www.ncpp.org/?q=node/20


Gore won the pop vote by .05%


And in 2008 Battleground Tarrance had Obama winning by 2%:

http://www.realclearpolitics.com/epolls/2008/president/national.html

The president won by more than 7%

[FBaggins]

6. Because they're the best.

And in 2008 Battleground Tarrance had Obama winning by 2%:

Well... no. Ed Goes' spin on their final number said 2%... Celinda Lake said that the same raw numbers meant a 5% gap.

They had Bush* beating Gore by 5% in their final 2000 pre-election poll:

True... but those aren't the only two polls that these guys have ever run. Does Rasmussen become credible because of the one or two years when their numbers were quite close? If you run a poll twenty times, a few of them will be off by that much and one will be way off... that's statistical variation, not polling error.

In general, they have an exceptional track record. More importantly, I know that they (in the past - I haven't reviewed this year) do a great job in identifying a sample group and craft their questions better than just about anyone (since they balance each other out of asking questions that would benefit the other side).

Just as importantly, their analysis has been invaluable. Each of them gives a different take on the same numbers.

Lastly... I think they shift from a more detailed strategy to a rolling tracking poll late in the cycle. I don't think that tracking poll has as solid a record as their more comprehensive polling.

The best hope for us in this poll is (IIRC) it's one of the ones that intentionally weights by party affiliation - based on a weighted multi-election average. As many republicans have been complaining, poll after poll this cycle "over-samples democrats". If we're right in believing that there actually are more people this cycle who consider themselves democrats, then (hopefully), it's possible that Battleground is muting that effect.

[B2G]

7. Here's their sample

The data from the Battleground poll shows the sample of 1000 likely voters was made up of 30 percent Republicans and 10 percent independents-leaning Republican while it included 38 percent Democrats and six percent independents-leaning Democrat. That is a ratio of 44 percent Democrats/independent leaning Democrat to 40 percent Republicans/independent leaning Republican. The sample included 16 percent independent that do not lean in the direction of either party.

http://www.politico.com/polls/politico-george-washington-university-battleground-poll.html

[DemocratSinceBirth]

9. The Race Is Close Because Battleground Shows The President Doing Poorly Among Independents

That mitigates the advantage in that poll that the president has in Party ID.

It's a respected poll... I just don't elevate any poll or over any other poll though there are polls I consider of dubious value.

[DemocratSinceBirth]

8. How Are They The Best?

That's your subjective evaluation. You have no established metrics to base your opinion on. Like most pollsters they have had good years and bad years.

As a social scientist I cringe when I see a pundit on tv say this or that poll is a solid poll or a good poll. There are no good polls or bad polls. There is only good social science and sloppy social science.

And most pollsters eschew weighting by party as party isn't a fixed characteristic. I can decide to become a Republican tomorrow though that is highly unlikely. But it is not nearly as unlikely as my becoming a Latino or along those lines a Latina tomorrow.

[FBaggins]

12. And I would disagree for precisely the same reasons.

There's good science and bad science. Good science can still end up with statistical variation that looks like "error" when cherry-picking specific results, so a couple counter-examples don't mean much to me (nor does Pew's apparent accuracy with a final call when their numbers leading up to the end were all over the place) - particularly when the rest of their work has been among the most accurate. They simply put much more science into this than the bulk of pollsters out there (so many of which have gone the robo-calling route). That doesn't mean that this poll is "accurate"... it's just why they're my favorite firm.

And most pollsters eschew weighting by party as party isn't a fixed characteristic.

And that's just what republicans were saying two years ago in response to democrats who didn't like what the polling was saying. They were right in that case, so hopefully we're right this time. But historically, party affiliation in exis polls has not varied as much as opinion polls leading up to the race. It all depends on the ground game.

On edit - And now I'm particularly happy that I mentioned how much variation Pew has had... because they just came out with one showing Romney up by four. - And now that I think of it... Nate Silver ranked their house effect as +3 D.

[DemocratSinceBirth]

13. We Know Party ID Isn't Fixed

In 00 the Democrats had a four percentage point advantage. In 00 there was no advantage for either party. In 08 the Democrats enjoyed a seven point advantage.

"Party identification is another thing entirely. Most fundamentally, it is an attitude, not a demographic. To put it simply, party identification is one of the aspects of public opinion that our surveys are trying to measure, not something that we know ahead of time."


http://www.people-press.org/2012/08/03/party-affiliation-and-election-polls/

But I agree, Pew's current four point lead for Romney doesn't strike me as any more credible than the eight point lead for Obama from just last week.

[JonLP24]

4. I only caught the last 20-30 minutes of the debate

but don't understand how it is widely accepted Obama lost the debate.

One thing, when they were discussing "Obamacare" and if it should be repeal, I'm not even a fan of the plan, but he pointed out Romney had no specifics (Romney didn't use the opportunity to detail them and harped on states' rights or such), didn't protect against pre-existing conditions the current health plan does, allows you to stay on current health plan (to counter "government takeover" lines)

[OldDem2012]

10. Old poll - covers dates October 1-4, 2012. nt,

[DemocratSinceBirth]

11. That's A Good Point

There should be at most a one day gap between completion of the polling and tabulation and release in this fast moving news cycle.