Morning Status Update for 10/8
Mark Blumenthal | October 8, 2008
Topics: Status Update
Another day of new polls, polls, polls confirming the recent trends to Obama and Biden at the statewide level but showing either a leveling off or perhaps a slight narrowing nationally.
All but one of the 21 new statewide surveys released yesterday represented net gains of at least one percentage point for Obama, although most (17 of 21) were updating results previously collected before the first debate. The new polls tipped the balance in both Ohio and New Hampshire, shifting both states to the light blue lean Obama category on our map, and increasing Obama's electoral vote total to 320.
The more important issue than our admittedly arbitrary line between "lean" and "toss-up" is the way the new polls affected the trend estimates in the battleground states. In 10 or 11 states with new polls, the estimates moved in Obama's direction.
The biggest changes were in Pennsylvania and Ohio. In Pennsylvania, three different polls by SurveyUSA, Rasmussen and Muhlenberg College now show Obama leading by 15, 13 and 10 percentage points respectively.
In Ohio, new surveys by PPP and CNN/Time show Obama leading by 6 and 3 point respectively -- narrow margins, but enough to reinforce and solidify the move in the trend estimate in Obama's direction. However, commenter Mike yesterday flagged something to keep an eye on: The margins in Ohio and Florida have been closer as measured by automated pollsters SurveyUSA and Rasmussen.
Our national trend, which is heavily influenced by the daily tracking surveys, moved slightly in the opposite direction, now showing Obama leading McCain by a 6.2 point margin (49.8% to 43.3%). That change is the result of two new surveys from ARG and Zogby/Reuters showing a closer race than our trend estimate, and a shift in McCain's direction on the Diageo-Hotline tracking poll.
By Mark Blumenthal | October 8, 2008 7:31 AM | Permalink | Comments (13)
Comments
Hotline down to a one-point lead today.
R2K/Kos narrowing from its stratospheric numbers.
Are the recent 1-3 point leads nationally a correction, buyers' remorse, or a response to negative ads?
Three points:
Ohio early voting much smaller than expected/hoped. Implications?
McCain camp openly putting Medicare on the table. Doesn't this look likely to hurt them in Florida?
McCain going harder at nuclear power. Have they written off Nevada, knowing that Yucca Mtn. is a key issue there?
Conclusion -- McCain looking to trade PA, MN, and NH for FL. Hold OH.
I would like Mr. Blumenthal to comment about what could explain the disparity between the polling results of Gallup and Rasmussen compared to those of Diageo Hotline from yesterday. Gallup has Obama up by 9 pts. (51%-42%), which mirrors Rasmussen at +8% (52%-44%), yet Diageo Hotline shows a clear tightening of the race with Obama only up by 2 pts. A closer look at Hotline's party ID shows a 2% dispartity (40% democrats, 38% republicans, 18% independents)from the poll yesterday even though on Monday their own poll showed a 5% spread. It is my understanding that the weighting of party ID can heavily influence polling results and be misleading as a representative sample of likely voters. Rasmussen is consistently polling with a 5-7 % differential in party ID between Democrats and Republicans and given the predicted higher turnout for Democrats this election cycle as evidenced by the higher turnout in the 2006 congressional elections and the 2008 presidential primary, the 5-7 % difference may be an underestimate. Otherwise, the party ID differential completely skews the results of this recent poll thus invalidating the results as being based on comparable weighting values. So my question is why did this poll weight the party ID to near parity between the two parties if not to mislead the public with the false perception that John McCain is starting to narrow the gap? If the Hotline poll does not weight by party ID, then the pollster should be more honest and state up front if in fact the poll was not weighted by party ID and my not be a representative sample of likely voters. Let's face it Fox News took a laughable poll of its own viewers to determine who won last night's debate and surprise, surprise, people who watch Fox News thought McCain won in a landslide, but that biased sample did not reflect the polls taken by CBS and CNN of undecided/independent voters.
Diageo Hotline has a very small sample size compared to other tracking polls, and therefore will show erratic results due to larger uncertainty.
Personally I ignore this poll because I don't think it is sampling enough voters to give any meaningful information on a daily basis.
In both Pennsylvania and Ohio the numbers seem to be converging to the Quinnipiac's numbers 9 (when the seemed like an outlier) a week ago - mid single digit in Ohio and low to mid doble digit in PA.
Mark,
What about the WI (Rasmussen) poll?
Gallup daily through yesterday shows Obama up 1/McCain down 1 nationally, so an 11 point spread..
Does anyone know of any Virginia polls in the field right now? that's the widest margin toss-up state leaning toward Obama - and I'm just curious when we'll see a new poll.
Nevermind - just found out PPP will be out tomorrow for VA polling ...
Tom Jensen on their post about the release - "We haven't finished it yet, but at first glance Survey USA and Suffolk look pretty reasonable there."
SurveyUSA +10
Suffolk +12
And PPP is saying those numbers look pretty reasonable!
The slight tightening at the national level might be due to McCain's increasing his lead in red states. A quick look at the larger red states seems to confirm this.
Mark, I love the state-by-state trend graphs. Would it be a huge undertaking to permit trend lines for more than one state on a single graph? Or even the average of several states on a single trend line? This capability would allow us to graphically test the above hypothesis.
Re: "Bradley effect" in OH and FL:
In FL, the automated (IVR) polls have McCain up 0.5%, while the live calls have Obama up 0.8% - I'd say this is within MOE. But put together, the net result is... Obama +3.2%!
...As for VA, the IVR polls show roughly the same gap as all-polls; in fact, while Obama remains at 49.2, McCain actually drops to 46.4 from 47. As a side-note, if Obama wins VA, he doesn't need either OH or FL.
Anyone care to handicap the over/under date at which McCain pulls out of PA? It looks gone already; certainly worse numbers-wise than MI did last week.
If I were advising him, I'd say pull out now, and concentrate everything on a firewall of OH, FL, MO, IN, NC, VA, and CO.
ProfMark
If McCain pulls out of PA he wont announce it. He alrdy got crap from GOP in michigan for pulling out there. I do suspect him to quietly start shifting funds to defende home territory though like MO NC IN and CO or make an offensive push in the now leaning Obama VA and FL.
I havn't seen any traction from McCain in PA since the crash so im not sure if he thinks he can get it or not. But if he spends too much time on PA he may just be swaping for other states and cause no effect or a net loss on the EV count.
I am trying to understand why polls are tightening somewhat. R2K tommorow will show Obama's lead under 10% unless he has a huge day today. Although R2K is now looking more and more like Ras and Gallup. Should I make a boomshak like prediction and say that Hotline will be tied today? Sure why not.
As far as last night's debate Obama won, McCain had a few moments but as David Gergen and Pat Buchanan said. Obama comes across as calm, cool and collected, McCain as erratic. People seem to like that calm, cool and collected persona versus Bush's erratic,stubborn deny reality mode of the last 8 years. Hopefully as the debate days are rolled into the trackers Obama's lead will stabalize at +5-+6
Posted on October 8, 2008 8:19 AM