One of the most bizarre but oft-repeated myths about polling is the notion that state polls 'lag' behind national polls, and particularly lag behind national trackers. This is largely a relic of a bygone era in which newspapers might commission a poll in their state, and then sit on it for several days until their Sunday editions or some other ebb in the news cycle. Thus, you might be seeing state polls that were in the field a week or so ago, whereas the national trackers were more up-to-date.
This largely is not true today, however. Rasmussen and SurveyUSA, perhaps the two most prolific public pollsters, generally release their data no later than 24 hours after it has left the field; likewise with other pollsters like InsiderAdvantage and Public Policy Polling. A couple of other pollsters like Quinnipiac and Mason-Dixon will occasionally sit on a poll for 24-48 hours, but generally not more than that. Every now and then, you'll have some small college or some fledgling marketing firm release a poll that is a couple of weeks old, but this is unusual, and it's easy to notate the exceptions. Most of the big, business-savvy pollsters recognize the importance of timeliness in this era of 24/7 news cycles.
Conversely, some of the national tracking polls are actually not all that fresh. IBD/TIPP has a 5-day polling window. Battleground also has a 5-day polling window, and they don't poll weekends, meaning that they're usually including some data that is a full week old.
Certainly, there is some mystery as to why the state polls and the national trackers seem to have diverged somewhat of late, with John McCain picking up perhaps 2 points in the tracking polls versus last week's averages, whereas the state polls haven't really budged one iota. The 'lag' effect, however, is not a valid explanation.
BONUS EDIT: Here's a fun little reality check for those of you who are sweating the trackers. Since our last update yesterday evening, I have added 34 polls to my database, including both state and national numbers. Barack Obama is ahead in 32 of those 34 polls. The two exceptions are in Arizona and Alaska, the home states of John McCain and his VP nominee, respectively.
10.29.2008
The Myth of the 'Lag'
by Nate Silver @ 3:45 PM...see also meta, national polls, tracking polls
Subscribe to:
Post Comments (Atom)

415 comments
True.
Dude, anarchists don't belong to a political party -- that would be ironic.
Little known tidbit, Milwaukee was a socialist city for much of the early 20th century. What was the net damage? A bunch of parks, which are quite nice. And some social centers that are now closed. We still have way too many public pools though.
Is McCain suggesting that Obama wants to build pools everywhere? I wish he would just come out and say it.
All kidding aside, I'm not sure rational people think that giving out a middle class tax cut is socialism. Countries with socialist histories must be scratching their heads or laughing.
Damn Nate, you should start in the tv show with David Krumholtz, "NUMBERS." (if it's still on the air, if not maybe you could bring it to Broadway!)
Thanks for letting me ride your very long coattails!.
Nate:
Perhaps it is because much of any 'national tracking tightening' is located in RED STATES that are already in McCain's column?
Well that was relevant to Nate's post.
Speaking of state polls, by the way, seems to be shaping up to be a good day for Obama in Ohio.
Do state polls use different likely voter models?
Is it possible that the majority of McCain's gains are in non-battleground states, where perhaps Obama is not campaigning as heavily?
The 'lag' effect, however, is not a valid explanation.
So what would be a possible explanation? Or are the national polls experiencing random statistical noise?
State polls are more important since we use the EV. I'm not worried about the national trackers as long as the state polls don't tighten too much.
I think the reason the national polls are showing a tightening of the race, while the polls of swing states are not -- indeed some showing the opposite -- is for the simple reason that the Obama campaign is wisely focusing its ad $$, volunteers and personal time on the swing states. He knows where he has to win and isn't wasiting time and money in Alabama, Oklahoma, Utah, Idaho, Texas and elsewhere where McCain may be making hay with his "socialism" campaign.
I've been screaming this into the gale for two months now!
Thanks, Nate!!!
Nate, you're going to make all the heads over at the Corner collectively explode. :)
By addressing this issue (in addition to everything else he has done), Nate Silver now has 98.9% of being officially cool.
Apparently Secretary of State Mike Coffman is getting sued over the 30,000 purged voter rolls.
since i have been coming to this site i have less faith in polling then i did before.
but i do love it. i could look at the numbers all day long
Wesley-
What causes the 1.1% chance that he's not?
Thank you, Nate. I'm so sick of this argument as well. Sure, there are cases where the state polls are older, but you just have to check carefully. I think the point Nate made in a previous post is most relevant: there are so many more interviews in state polling than national polling that the latter is far more likely experience significant variability.
Lag does not have to do with sitting on a poll for 2 days. It has to do with whether the state mood represents the national mood via the polling data.
It is possible to have the national trend in one direction, say a widening lead for obama, but that might not show up in a state poll for several days.
I think you need to do a real lag analysis rather than give it some anecdotal blog time.
This is what I recommend. Compare weekly moving averages of virginia against the weekly moving average of the national poll and then compare peaks and time to peaks.
It looks like red getting redder and blue getting bluer.
"hyporic"
Lots of ways to run with this one.
Apparently Secretary of State Mike Coffman is getting sued over the 30,000 purged voter rolls.
*Sticks out window*
*Looks up*
*Sees overcast skies*
Can someone tell me if the sky is still blue?
Then why use national polls in your model at all?
This is welcome news to me, Nate. I could never remember whether it was the national polls that were supposed to lag behind the state polls or vice-versa. Now I know that it's b.s. anyway.
word verification: cants. definition? GOP candidates in 2008.
thanks jon-good news!
i have grown to loathe that CO creep
my word was "hackw" maybe i should have called coffman a hack
Nate, could I make a suggestion? The Super Tracker is great, but how about a second one showing results from the past month only? I find myself trying to interpret the small dips and spikes at the end of the graph.
I wouldn't be surprised if the tightening in the nationals is to somewhat from already red states. I can imagine the 'socialist' cr*p selling in places like Texas.
Incidemntally I was readin parts of Paul Wellstone's 'Conscience of a Liberal' and the bits on the elections really reminded me in some ways of this years Presidential. All the GOP could through at Wellstone were slander and mud in the end.
IBD TIPP
O: 46.9
M: 43.9
It is possible to have the national trend in one direction, say a widening lead for obama, but that might not show up in a state poll for several days.
This is not true for polls that are in the field at the same time.
National trackers are a movie. State polls are a part of a single frame. They may be lighter or darker than the whole shot, but there are no rules that say that they must follow the national trend.
Also I wonder if some of the tightening is through the political bias of certain pollsters, and maybe thats harder to build into state polls.
yes, it would seem likely that red states would be getting redder, leading to a nat'l tightening... but that doesn't appear to be the case if we look at Nate's own "Poll Detail" Alaska, Alabama, Georgia and Kansas (to name just 4), all seem to show the opposite, movement toward Obama, even though all are decidedly red.
Then why use national polls in your model at all?
Earlier in the campaign we didn't have four polls per day per battleground state. The model was heavier on national trackers and demographics back then.
Why isn't Obama hammering the lies that he's going to put up taxes for ordinary people, and that he's bad for small business?
That, plus the Joe the Plumber meme is clearly doing some damage.
I think the national tracker to watch right now is Rasmussen, which has Obama up by 3 points. For the last month plus a few days, that poll had Obama up at least 4 and at most 8, every single day. If the 3 point lead today becomes 2 points, 1 point and then zero, the electoral college map is bound to change, at least in the 5-10 states which are still close.
Steve said...
Then why use national polls in your model at all?
Not as many state polls, and even with all the polling in battleground states, there is little in other states. The nationals allow gaps in that info to be filled in a bit.
Rasmussem New Mexico:
HEROMCCAIN 49
ZEROHUSSEIN 46
CNN Virginia:
HEROMCCAIN 46
ZEROHUSSEIN 48
The poles are closing in fast, libs! The lag IS REAL! Prepare for McCain victory! Have you purcchased your trane tickets to France? LOL
Word verification: chemat
Simper Fi
It's not necessarily only movement in the red states. The fact is, most states are not really seeing any campaign action at all. This includes states that are solidly on one side or the other. Many of these states are very large (NY, IL, CA, TX) so we could be seeing some tightening in these big, ignored states that drives the national polls.
AP article on early voting favoring Obama:
http://news.yahoo.com/s/ap/early_voting
409,239 voters could changed this election
Lag does not have to do with sitting on a poll for 2 days. It has to do with whether the state mood represents the national mood via the polling data.
It is possible to have the national trend in one direction, say a widening lead for obama, but that might not show up in a state poll for several days.
This only makes sense if you are claiming that some states are backwards, compared to other states. So perhaps some new development shifts attitudes in TX, CA, and NY very quickly, but voters in other states take more time to digest the new information and change their votes.
By some odd coincidence, of course, the laggard states just happen to be the swing states that are polled most heavily, while the fast-changing states are the ones most ignored by the state pollsters.
Sorry, this doesn't make much sense.
And the Word of the Message is: Otionsea!
Actual CNN/Time Va Poll
O-53
M-44
Shameful bit of wishful thinking:
Can it be that nationals are more influenced by party ID weighting and that more independents who planned on voting Obama anyway have been calling themselves Democrats in the last few days, depressing the Obama vote in the Party ID-weighted polls?
That, plus the Joe the Plumber meme is clearly doing some damage.
Not where it matters. (Hint: Joe's neighbors think he's a clown.)
For McCain: No Ohio - No Victory.
CNN polls:
Colorado: Obama 53, McCain 45
Florida: Obama 51, McCain 47
Georgia: McCain 52, Obama 47
Missouri: McCain 50, Obama 48
Virginia: Obama 53, McCain 44
Orange County, FL early voting as of 10/28 at 9:29 p.m.:
Total early votes: 75,090
Registered Dems: 45,005
Registered Repubs: 16,225
Registered other: 13,810
LAS VEGAS REVIEW JOURNAL
EARLY VOTING SHOWS SLIM OBAMA LEAD:
http://www.lvrj.com/news/33433079.html
Exit polling of early voters conducted by a local political consultant suggests a close presidential contest in the Silver State: Democratic nominee Barack Obama had the votes of 50 percent who had voted, while Republican nominee John McCain had 48 percent of the early vote.
The poll, conducted by Las Vegas-based consultant Steve Nathan’s firm Dialing Nationwide Automatically, or DNA, was conducted by automatically dialing everyone who had voted through Friday, based on statewide voter records, and weighting responses to reflect the composition of the electorate so far.-(all emphasis mine)
What an advantage early voting gave Obama in NV..... NOT. Up 2 measly points...
NV FOR MCCAIN!
Rasmussen AK-Sen (10/29):
Begich - 52
Stevens - 44
Then why use national polls in your model at all?
They're useful for guesstimating the trends in states that don't get polled very much. So they'll affect Nate's model far more in states like ID than in states like PA.
And the Word of the Message is: Gantstan!
Time to sweat the trackers boys.
Three national trackers a three points and closing.....
So, Nate, are you telling us that the Ras +5 yesterday vs the +3 today is an improvement or movement for Obama???
I also wonder if national polls may be less accurate than state polls in part because a national poll has a smaller sample size relative to the national population. Also, demographics are easier to pinpoint and extract at a state level than at a national level. Anyone care to critique my assessment? I'm fascinated by this stuff, but I know nothing about statistics.
I think in this election, a few states lagged behind where they should have been based on what we would guess at about their demographics. I think that Ohio did lag behind when the national surge towards Obama started waaaaay back in September.
I've never seen a day where the state and national polls seemed to diverse so quickly.
Those CNN polls are perfectly acceptable from an Obama point of view, be nice to be up in MO as well, but within 5 in Georgia, up in Florida, Colorado and Virginia. All good.
i am starting to think that the Great Blue Road Trip™ from Miami to Seattle might not be doable this year.
it will have to be the 2012 project
TREMTHER!!!!
NATE. You are full of it. You preached on here for months that state polls lag behind national polls.
Therefore it could take 3-5 days before we see movement for McCain in the state polls.
Why?
God only knows, but after the conventions and debates, this was evident.
McCain could get that last minute surge in the state polls just before election day.
I put this on the last thread but it's more relevant here. Please Nate comment on this if you read comments!
Nate,
At this point, what is the legitimate PURPOSE of national tracking polls?!?! Just to create false security or alarm? I think they should stop national trackers 5 days before the election as by then there is no "narrative" to create or watch- it's 100% all about the swing states electorate - and the NUMEROUS state polls for that are more than sufficient.
Yesterday and today, dozens of polls show Obama in strong positions in every swing state. On top of that, Obama leads basically every state in early voting, and in some cases (north carolina), its with half of ballots in and more coming. Yet National trackers continue to "tighten."
If Obama closes the deal in CO and VA and keeps the good sized lead in PA, then things will be just fine. It looks like it's hard to imagine every pundit being wrong at this point -- also with the support of electoral vote projection websites like 538 showing McCain falling deeper into the quicksand.
CNN polls:
Colorado: Obama 53, McCain 45
Florida: Obama 51, McCain 47
Georgia: McCain 52, Obama 47
Missouri: McCain 50, Obama 48
Virginia: Obama 53, McCain 44
Oh I do like those CNN polls. Almost all of them trending nicely toward Obama, with the exception of Missouri. Obama at 53 in both Colorado and Virginia is very nice indeed,
I was wondering about the lag issues after seeing AP's numbers today.
Polling on those states was conducted 10/22 - 26.
However, today's 3 day trackers were polled 10-26 - 28.
So IFthere had been some mvt picked up in the last 2 days, wouldn't that theoretically make the state polls suspect?
Michael --
It's called learning and adapting to new data, that's what educated people do. Nate started this site this year, he's a stats expert, but there's still a learning curve associated with polling data as oppose to raw baseball stats like he's used to working with.
Chill out.
@micheal
Chill out! A McCain SURGE of this magnitude is how likely?
*achtn*
Blogger Rich Rifkin said...
I think the national tracker to watch right now is Rasmussen, which has Obama up by 3 points. For the last month plus a few days, that poll had Obama up at least 4 and at most 8, every single day. If the 3 point lead today becomes 2 points, 1 point and then zero, the electoral college map is bound to change, at least in the 5-10 states which are still close.
Right now, I'm not overly concerned about the RAS poll. It's tightening comes on an increased support for McCain—Obama's support remains stable. Of some concern is this comment:
It will take another day or so to determine whether today’s numbers reflect a lasting change or statistical noise. Two of the last three nights of polling show a closer race than was found in the previous month.
I've been turning this comment over in my head all day. This says to me that McCain had a very good night in the polling last night. From the overall numbers and the way the comment is phrased, I think McCain had a good night on Sunday night as well, but not on Monday. The numbers are bouncing around (or at least McCain's number is), and RAS isn't sure what to make of it. Kos has noted that their polling numbers have been bouncing erratically as well, especially in the 30-44 male demographic. So something seems to be going on. No one seems quite sure what, though McCain's lines on "taxes" gaining some traction seems the most likely rationale. (I'm not really buying it, though I don't have a better explanation.)
On the other hand, my sense remains that what we are seeing in this "tightening" is primarily pollsters make adjustments in the softness of McCain's support. That is, the tightening is coming from pollsters pushing undecideds into McCain's column. Whether that is warranted—well, we'll see on Nov. 4th. I remain doubtful. Personally, I don't see McCain breaking 46% on Election Night, and I think he is more likely to poll below 44% than above 48%.
Word verification: ovent.
A slight majority (52%) of Rasmussenreports.com viewers think McCain is going to win Virginia...interesting, but then I realized the site is overwhelmingly GOP (comments are uprated if you support McCain for the most part, haha.)
I like the site 270towin because i am a pessimist and i always give MO, IN, NC, and GA to Mccain even though they are toss ups.
338
200
@Michael
You preached on here for months that state polls lag behind national polls.
Provide one example where Nate has "preached" that.
Lack of lag is WONDERFUL NEWS!!! For John McCain!!!
Big news will break soon.
The LATimes Tape is just 1 of 2 bad tapes going to be released that will send undecideds screaming to McCain!!!
@ant
Did you see my message on other board? I thought Ras had a NM poll coming out with the three you posted.
*trousec*
Nice. Ed Rollins is on CNN basically saying that "internal memo" is basically BS and he said if it is true, they should invite in reporters and show them the evidence.
What you can see is a lag in the electoral map if some states aren't polled frequently, and their results remain on the map after opionion has begun to change.
However, that's different than what seems to be happening now, when the results of new state polls seem to trend in a different direction than the trackers.
michael said...
NV FOR MCCAIN!
McCain campaign gave up NV
i wonder why
LOL
Totally off topic, but I just wanted to announce another early vote for Obama in Colorado - mine. No lines, but people were coming and going. I signed up, got a booth, and voted in less than 10 minutes.
Anyone have any stats on early voting in CO?
"The LATimes Tape is just 1 of 2 bad tapes going to be released that will send undecideds screaming to McCain!!!"
Yes, brining his support up from 39-42 percent to maybe 45-47 percent. Which is still a losing hand.
Michael said...
NATE. You are full of it. You preached on here for months that state polls lag behind national polls.
Bullshit.
Therefore it could take 3-5 days before we see movement for McCain in the state polls.
Bullshit from bullshit.
Why?
Why not?
God only knows, but after the conventions and debates, this was evident.
No it wasn't! They didn't DO state polls during the conventions, so any fucking "lag", was because they had to go out into the field, come back, compile and publish. Meanwhile we're looking at trackers bouncing around in the interim.
McCain could get that last minute surge in the state polls just before election day.
Yeah... He'll get a 4 point surge in PA and a four point surge in VA and still lose both. BFD.
Take your weak sauce out of here.
Unless McCain wins one of PA, VA, or CO, he's done, and he continues to tread water in those states,
While I know zippo about statistics, it seems to me that the closer you get to the actual election, the less important national polls become, and the more important the state polls do.
So, it's not as if things remain static on what's important and what's not.
It's like the economy, stupid! We need to swing back towards 'socialism' now, but 40 years from now (thankfully, I'll be dead), we'll do the trickle-down thing again, crash, hit rock bottom and so on. The pendulum needs to swing both ways.
"i always give MO, IN, NC, and GA to Mccain even though they are toss ups."
Missouri Bellwether holds for another election...I'll say 50-49 for Obama.
I think NC will be McCain by 1, GA will be McCain 50-47 (Barr and McKinney will have their highest draw there,) and IN will be McCain by 4.
The poll, conducted by Las Vegas-based consultant Steve Nathan’s firm Dialing Nationwide Automatically, or DNA, was conducted by automatically dialing everyone who had voted through Friday, based on statewide voter records, and weighting responses to reflect the composition of the electorate so far.-
Where do you even start with this one...
Word: Wheamet. Where?
Michael -- only GOP wingnuts will give a second thought to those tapes. Besides, when the McCain campaign has staffers like Timmons whose linked to lobbying for Hussein to the U.S. government, it's not really smart to start playing the associations game.
You're probably just upset because Ayers, Rezko, Wright etc. have been countered by more damaging Liddy, Keating, AIP, and Palin's witch doctors. Go to free republic if you want to spout such garbage.
The national trackers are moving because the Republican base continues to solidify. McCain is going to rack up some great numbers in deep-red states like Oklahoma. Those states don't get a lot of polls. Hence the discrepancy.
Michael,
I have some swamp land to sell you if you believe those NV numbers.
How much of this national tightening is due to people already voting and (my assumption) not being polled?
"insnes"
MCain an Plin iz teh insnes.
If McCain gains 3 points or so in CO, VA, and PA, he's still cooked. Those three states have outpaced Obama's national margin for months.
Nate, how can you not speak of lag, when the three-day tracking polls were conducted October 26-28 and for example the polls released today by the Associated Press/GfK and Quinnipiac were conducted October 22-26?
I keep my eye on the Gallup tracking poll they seem to be the quickest to respond. Obama had a slight up tick today in that poll. Rasmussen seems to lag a bit and is probably responding to the little dip this weekend. Polls like Battleground, IBD and Zogby are just garbage they use old models that don't mean a thing. These three polls in particular us almost equal party affiliation models. For example Obama has had a consistent 5 point lead with independents in IDP and a 9 point lead with seniors in their poll. If that was true Obama would have a 6-10 point lead.
McCain is gaining some undecided voters that were probably were leaning McCain all along but waiting to commit. Like some folks already said I think much of McCain's increase has come from stupid red state voters that want to believe anything bad about Obama. Also Obama's number has not changed much in weeks and until that top number consistently goes below 50% Obama will win easily.
In states like OH, VA and CO in particular Obama is making some impressive gains.
Obama must have heard me.
http://www.huffingtonpost.com/2008/10/29/obama-sharpens-tone-again_n_138915.html
I have some swamp land to sell you if you believe those NV numbers.
Nevada swamp land at that...
@boulder-liberal
early vote stats here
*bronst*
Lol, let's go with those tapes, because:
"McCain also has ties to Khalidi through a group that Khalidi helped found 15 years ago. The Center for Palestine Research and Studies has received more than $800,000 from an organization that McCain chairs."
-MSNBC.com
i think that the lag will turn into a signficant tightening of the state polls in 7-8 days
aunt karen said...
Lol, let's go with those tapes, because:
"McCain also has ties to Khalidi through a group that Khalidi helped found 15 years ago. The Center for Palestine Research and Studies has received more than $800,000 from an organization that McCain chairs."
-MSNBC.com
OMG !!
LOL at don't panic!
I can't help but think that while up-to-the-minute results are cute, they are borne more of a sports-fan mentality than any real political importance. A weeks-old or months-old presidential election poll will still reflect any meaningful trend just fine.
I need to qualify that. "Meaningful" does not mean "who will win." Our system allows someone to become president when half the population hates and loathes him. In the long term, these vagaries of fate ought to even themselves out (except of course for the fact that politicians try to perpetuate the systems that gave them power; look at the trends in education or state-sponsored religion). If half the population hates the president, that is not exactly a mandate. That is a sign that it's time to split the country into two, or to decrease the power of the federal government and give back to the states the powers that the Constitution left to them.
So by "meaningful" I mean a real, honest trend in the way the public thinks about real big-picture issues, not how rednecks react to the claim that Obama had a childhood acquaintance who later became a terrorist. If Obama wins by 80%, that's exciting. If he wins by a similar margin to Bush's, that is a sign of a crumbling society.
Of course, in an era of politics-as-sports and headline news for a generation without an attention span, up-to-the-minute "trends" are exciting to watch! And there's no denying that there's a lot on the line this time. I check 538 every few hours too. But I still believe that this country will not see meaningful trends for a long, long time.
don't panic said...
i think that the lag will turn into a signficant tightening of the state polls in 7-8 days
what ?
election will be over
Nate, how can you not speak of lag, when the three-day tracking polls were conducted October 26-28 and for example the polls released today by the Associated Press/GfK and Quinnipiac were conducted October 22-26?
FFS!! THAT'S NOT LAG!!!! IT'S ONE POLL TAKEN AFTER ANOTHER!!!
@Michael
The only times I recall him doing that are when we are on here hyperventilating twelve hours after a debate or an endorsement or a news cycle that just hasn't had time to be fully reflected in a pollster's full window of time in the field; thus, it would take a few days for polls to show the full impact of that news cycle.
That's not lag, which is more the sitting on your hands notion Nate mentions in his post, that's just giving people time to do their jobs.
I don't know about what Nate is saying. For many of the state polls being released today, their last day in the field was the 26th. For whatever reason over the last 3 days McCain has closed the gap in the national tracking polls. Rasmussen, which Nate says he trusts the most, shows a 3 point lead for Obama, with McCain at 47 percent, the closest the race has been and the highest McCain has been in over a month. Obama had a really good end of the week last week which the state polls seem to be picking up but the national trackers are typically a good indicator of how the state ones will go, so we won't know more for a couple of days. I don't think it does any good to argue that Obama is way ahead. The only thing worse than complacency with a big lead is complacency with a small one.
Anyone have any stats on early voting in CO?
As of yesterday, 958,508 votes had been cast, representing 44.6% of all votes cast in 2004. In comparison, 47.9% of ballots were cast during the entire early voting window in 2004. Expect to see this number surpassed within the next day or two.
As of 10/23, 38.6% of CO early votes were by registered Democrats, and 37.8% were by registered Republicans.
Source: http://elections.gmu.edu/early_vote_2008.html
Word of the Message:
Luretat
don't panic said...
i think that the lag will turn into a signficant tightening of the state polls in 7-8 days
I see what you did there.
What is the "internal memo" people are talking about? Sorry if I've been out of this loop...
[dimudis??]
I gotta disagree with Nate. It seems reasonable that trends would first appear in states that are most partisan, since undecided’s in those states should naturally be more susceptible to social pressures than people in battleground states. Think about it this way – if you are on a jury that is deadlocked 6 to 6 it’s harder to move anyone’s opinion than if you were one of three holdouts. But once the trend starts to go, it gets easier to move people in more balanced states.
just been playing on 270 to win, gave McCain PA(just for a laugh!) and I have NV, MO, IN, OH, NC and FL left to call. Any one of those would give Obama 270+. Not totally impossible that McCain sweeps those states but without Pennsylvania McCain really cannot win.
@antmatic
"Unless McCain wins one of PA, VA, or CO, he's done, and he continues to tread water in those states."
Frankly, I would even take CO off that list. If Obama wins PA and VA (and picks up Iowa), that gives him 272. That means he could lose CO, NM, NV, OH, FL, NC, IN, MO, etc and still win.
I continue to care mostly about PA and VA polls. When we know results from those two states on Tuesday, we'll know whether Obama is president--just a question of the final EV tally. (Although if he loses one of those states, we wont necessarily know McCain will be president, then all those other states matter.)
real joe said
what ?
election will be over
i see nothing escapes your keen eyes, real joe ;o)
It would be great if Nate could weigh in on this:
On the national tracking polls -
In getting their answers, do they make an attempt to distribute throughout all 50 states, or as close as possible? Or is that not one of the criteria?
I ask this, because with all the intensive polling of the battleground states (which means higher degree of non-response from fed-up answerers) - could some or all of the national trackers be AVOIDING those states, and instead sampling more heavily in the eastern and central clear lead states?
If that is the case, might it not skew the results? (since people are seeing far fewer commercials)
I'm curious - anyone know the answer?
One of the possible reasons why the national numbers may look worse for Obama right now than do the state numbers is the phenomenon, which I've seen discussed elsewhere, by which McCain seems to do worse in media polls (those by the networks and national newspapers) than in the daily tracking polls. So at a time (like right now) where there aren't a lot of recent media polls, Obama's numbers look worse. RCP right now, for example, is pretty much all trackers except for Pew (which is +15 Obama, as compared to their average of +5.9).
Recently excluded media polls are:
Newsweek +12
CBS/NYT +13
FOX +9
NBC +10
Most of these organizations are due for another poll in the next few days -- it'll be interesting to see what numbers they come up with.
My argument, if you choose to accept it, is that state polls are structurally more like the national media polls than like the daily trackers. I'm not saying which approach is more accurate, but they do split into fairly distinct camps at the national level.
Daniel said:
"For many of the state polls being released today, their last day in the field was the 26th"
CNN was in the field until yesterday.
The big change in Ras today is that McCain is at 47, instead of 44-46. That is it. That is the only difference between this Ras poll and every other Ras tracker for the last five weeks.
don't panic said...
i see nothing escapes your keen eyes, real joe ;o)
;-)
Is it possible that when you randomly call 3000 (give or take some) people a night from a country of 270+ million you're going to have more noise or, sometimes, hit more red than blue (or more blue than red) than if you're calling 600 (give or take) people in a state of ~11.5 million (my state of Ohio)?
Evan:
305,529,978 (est)
Evan, it's actually only 1,000 people per night for Rassmussen, and less for other trackers.
@Mason
Nate's 1.1% chance of being uncool due to rumors of a surprise endorsement of McCain by the reputable poll analyzer.
I just saw on MSNBC that the US Embassy in Syria will be closed tomorrow due to "security concerns".
Something to make you go "Hmmm..."
@blue in pa
The memo is basically a rah rah pep talk. Ed Rollins, who has been pretty level headed and conceded already that McLame will lose, was just on CNN saying that he doesn't believe the memo because of all the public polls out there and he also said that if the internal numbers were accurate, then they should invite in reporters and actually show them the evidence.
Response to:
Pat Andriola said...
Great numbers in Colorado and Virginia, GA is whatever, it won't matter in the electoral map, MO is a toss-up which I'd like to see more polling of, and Florida number is great too.
So the Obama path is:
Kerry + IA + NM = 264 then all he needs is one of the following:
Here are my odds for each for Obama (Eric):
1) Colorado (97% if voters will is shown) Because of Coffman's attempt to "steal" it for the GOP, 90% for Obama instead of 97)
2) Nevada (tie goes to Obama) 99%
3) Virginia 98%
4) Ohio 90%
5) Florida 80%
6) North Carolina 65%
7) Missouri 55%
8) Indiana 55%
9) Georgia 50%
10) Arizona 30%
11) Montana 55% + North Dakota 55%
both together (40%)
13) West Virginia 25%
Check mate.
Can someone give a quick synopsis on the tapes? The only one I had heard about was the one the LA Times is apparently holding.
Or are we back to the "whitey" tape again?
The only real move against Obama recently in the trackers, at least that I really care about, is the Ras poll, and even that only really comes out in the national polls. Ras state polls seem to strongly favor Obama when it matters.
And remember its PA or CO and VA, if he doesn't get PA then McCain needs both CO and VA. All three seem very tough for McCain right now.
When you first introduced the tracker that applies national data to individual states I was skeptical, but have since come to accept it based on empirical results.
However, the current divergence might be a manifestation of my original doubt, which is very simple (and simple minded):
The lopsided states can get more lopsided without having any effect on the electoral college.
Perhaps the undecided or weak Obama voters in UT, WY, AL, MS, TN, KY etc. are finally coming home to GOP, or the admonitions of their Sunday preacher, or their fears of terrorism or blacks or gays or whatever. And at the same time, perhaps the groundswell of new Obama voters is keeping pace with this surge in the battleground states.
Peixegato -
Maybe there is a Syrian gov't organized demonstration tomorrow about the border incursion over the weekend?
Doesn't make me go hmmmm - probably not anything important
State polls "lagged" national polls earlier in the year because states were polled much less frequently than the nation. Therefore, the national trackers could predict movement in the state polls
This is not to say that state polls were less accurate than national polls released at the same time-- this is the "lag" Nate is addressing here, and it is very different then the "lag" that was present earlier in the cycle.
Now that state polls are released frequently in battlegrounds, any motion should be detected concurrently with national movement.
Like I said before Nate: just give us a signal when we should all run outside and pass the ball around between ourselves to run out the clock. We're ready.
Polling on those states was conducted 10/22 - 26.
However, today's 3 day trackers were polled 10-26 - 28.
So IFthere had been some mvt picked up in the last 2 days, wouldn't that theoretically make the state polls suspect?
Maybe. So let's chop 3 points off of Obama's leads in the AP/GfK polls:
OH: O+7 now O+4
FL: O+2 now M+1
PA: O+12 now O+9
NC: O+2 now M+1
VA: O+7 now O+4
CO: O+9 now O+6
NH: O+18 now O+15
NV: O+12 now O+9
Result: Electoral College Landslide
Has anyone been paying attention to the fact that the Iraqis are likely not going to agree to the status of forces agreement, and if that happens, the US has to pull out ALL of its military operations from Iraq on Jan 1, 2009? Scary stuff. But I say, if the Iraqis don't want us there, lets spend that 10b a month in fixing the problems in this country.
@mason
I should have gotten a US population. For whatever reason, I can't get the population figure I read somewhere as a kid out of my head. Of course it's 300+ million now.
@rwd
You're right. That's my bad. I was trying to cite the totals for the whole poll and wrote the wrong thing.
I really fouled that up. So much for my first attempt at making a point on my favorite website (although I wonder if the argument is valid).
Anyone?
ITS OVER FOR OBAMA !!!!
Michelle Obama "Whitey" Video
Link
Final lap:
Obama will be in Missouri on Thursday
Indiana on Friday
Colorado and Nevada on Saturday
Incidentally just to add to my previous post, I just had a combination on 270towin [my electoral map predictor of choice) that gave McCain VA and CO and its still a 269-269 tie, and other than PA I can't see another state I would be prepared to give to Mccain save maybe NV or PA. The electoral map really is tight for McCain
"You're right. That's my bad. I was trying to cite the totals for the whole poll and wrote the wrong thing.I really fouled that up"
ehh, don't be so hard on yourself. the corrections just make your point stronger.
can't believe that IBD surveys less than 200 people per day nationwide....
real joe,
Ouch!
dammit! someone recorded over it! all I get is a stupid video of rick astley singing a stupid song.
Damn you Real Joe...I prefer a Barack Roll myself though LOL
Jeff NYC Dem,
Dunno if this has been answered already, but:
> I also wonder if national polls may be less accurate than state polls in part because a national poll has a smaller sample size relative to the national population.
This doesn't jibe with what I learned in Stats 101. I was told that a sample of, say, 1000 people is just as accurate for a population of a million as it is for 10,000. Yeah, it didn't make sense to me either.
I'm just waiting for the McCain c*nt tape myself....
@real joe
You made me fall out of my chair in fear. Then I pressed play and I fell out laughing.
We got rove-rolled.
Is it bad I have that mp3?
Nice to see Rasmussen's McCain bias become OFFICIAL. What was this part of, the Joe the Plumber push poll?
http://www.rasmussenreports.com/public_content/politics/election_20082/2008_presidential_election/mccain_trusted_more_on_taxes_and_economy
Nate, it's only fair, you gotta call this wanker out. He's push polling to make the race look tighter than it presumably is.
Stay classy, Joe! :)
Heh... okay Real Joe, you got me. That's the first time I've laughed at that in a while.
Nevada early voting:
http://washingtonindependent.com/15662/democratic-surge-in-nevada
"Early Voting Reveals Democratic Tilt"
http://blogs.wsj.com/washwire/2008/10/29/early-voting-update-15-million-ballots-and-counting/
"Nevada breaks down its vote by county and reports that Democrats have filed 54% of the early votes in the Los Vegas area, compared to 30% for Republicans."
Nice one, Joe!
:)
Australians favour Democrat four to one: poll:
http://www.theage.com.au/world/us-election-2008/australians-favour-democrat-four-to-one-poll-20081029-5bdy.html
Modern black people dont call white people "whitey" as a derogatory term. Thats some 1960's early 70's black panther era shit.
Michelle is too young to even speak like that.
It always just seemed completely unnatural for any african american of her age to even have that slang in her natural vocabulary, even if she were to say something derogatory about white people.
Chris Bowers explains the National Tracker vs. State Poll "discrepancy":
http://www.openleft.com/showDiary.do?diaryId=9503
"...a partial answer is that there is no discrepancy between state polls and national polls, at least in the largest swing states. Earlier state and national polling diverged from one another in these states, but current state and national polling does not."
ahead... OR TIED...
I'm not getting the explanation. I see four national tracking polls every day here (Gallup, Hotline, Zogby & Rasmussen), and I can track them individually and as an average and see trends and patterns. How do I do that with a specific state? I don't see any four sources with results posted every day. I can't see trends the way I can with the nationals. So to me, there is a lag. It looks like McCain is closing fast in the nationals, but I don't have enough data from a single source to see if that's true with the states.
Other than the fact that the national trackers are measuring a lot of states which simply do not matter at this time, consider that there are very important differences within the battleground states.
They are getting hammered by advertising and canvassing at a level vastly higher than the rest of the U.S.
To extrapolate what is going on in the battleground states (which is all we care about) from a national sample is not only difficult, it verges on pure folly. I agree with what an earlier commenter said - stop the national trackers two weeks out. They are not helpful.
Further consider that Obama basically shut down his national advertising last week. Those national trackers *should* be tightening - he basically ensured that they would. He may, in fact, *want* them to tighten up. If I were in his shoes, I would. He needs his base mobilized.
A black people calling white people "whitey" in the 21st century, is like white people refering to black people as "coloreds"
You must have just hopped out of a time capsule to still use that language.
"This doesn't jibe with what I learned in Stats 101. I was told that a sample of, say, 1000 people is just as accurate for a population of a million as it is for 10,000. Yeah, it didn't make sense to me either."
Yeah, I remember something like that maybe a stat guy can explain. My logic follows this line: Assume 1% of a population of 1,000,000 people believe the world is flat (10,000), and the same is true of a population of 10,000 (100). If you poll 1000 people in each there is at least SOME statistical chance you will get a result that says 100% of the group of 1,000,000 believe the world is flat. There is no chance of such a result when polling 1000/10000.
PorridgeGun said...
Nate, it's only fair, you gotta call this wanker out. He's push polling to make the race look tighter than it presumably is.
From your own link:
1* Who do you trust more on the War in Iraq, Barack Obama or John McCain?
2* Who do you trust more on immigration?
3* Who do you trust more on handle environmental issues, Barack Obama or John McCain?
4* Who do you trust more to balance the federal budget?
5* Who do you trust more to negotiate trade agreements?
6* Who do you trust more on taxes Barack Obama or John McCain?
7* Who do you trust more on social security?
8* Who do you trust more on healthcare, Barack Obama or John McCain?
9* Who do you trust more on education?
10* Who do you trust more on abortion?
How is that push-polling?
Real Joe:
Can you please enlighten me? Everyone always posts links to that Rick Astley sensation. But why? I feel like I have to laugh along with everyone else so I don't have to admit I don't know what's going on. So, please, what's going on?
Charlie speaks:
http://thehill.com/leading-the-news/crist-says-internal-polls-show-mccain-ahead-in-fla.-2008-10-29.html
However, Crist, citing an internal poll, said on MSNBC that McCain is turning things around in a state that is critical in his quest for an upset win next Tuesday.
“I’m hearing some very encouraging numbers, you know, some polls reporting [McCain] three or four points up already in the Sunshine State,” the governor said.
From Ras's write up today
'Obama has a five-point advantage among those who plan to vote but say that something might come up. Hispanic voters are more likely than others to say that something might come up to prevent them from voting.'
Something about that line makes me laugh very loudly, BUT also, just shows how important the GOTV is for
Obama in the end.
sloth said...
Further consider that Obama basically shut down his national advertising last week. Those national trackers *should* be tightening - he basically ensured that they would. He may, in fact, *want* them to tighten up. If I were in his shoes, I would. He needs his base mobilized.
I noticed that, too. I'm in CA so I only ever see the national ads. I saw a few Obama ads earlier this months but those have completely stopped in the last couple weeks. However, I see a TON of anti-Obama ads from indy groups and the RNC on a regular basis.
I also figured Obama has shifted resources to the GOTV effort in the battlegrounds and purposely let his national effort die and create the perception of a tightening...
anyone have details on Obama's final rallies? If he's back in CO I'd like to attend...
ABC/WaPo tracking poll...
O: 52
M: 44
O+1
@James- of course it's possible for "a" state to lag behind the national changes. Maybe the Pony Express gets there late. It is not mathematically possible for all states to lag behind the weighted average of all the states! The general lag phenomenon was purely a result of less frequent state polls. So we might correct slightly aged state polls by national tracker changes, as Nate already does, but right now it doesn't look like that would affect the net result. /mbw
I was also entertaining the possibility that the closing national polls might reflect increases in support for McCain in deep red states. But like Zak, I've also noticed that there is no apparent trend toward McCain even in states like Oklahoma and Texas. In fact, the trend has been flat or toward Obama even in these very red states. Aside from the lying smears from the McCain campaign, there is no other factor other than racism that does not work against McCain and for Obama.
Despite the paucity of state polls in some states, it seems like it should be possible to predict the national vote from the individual state polls, using estimates of the total number of votes to be cast taken from state voter registration statistics. I imagine Nate has probably played around with this notion already, so I hope he will post a comment.
Just looking at the state polling results, I wouldn't be surprised (but would be delighted!) if the Pew poll turns out to be right on target.
I see the FReepers are back.
I've got something for you swell guys!
MC CAIN NOW TIED OR LEADING!
Oh, I was talking about Arizona!!
Hark,
At RCP, you can get a graph of the state polls over time. I looked at PA, FL, NV, CO, and OH, no sign of a drop-off in the past several days...
Incidentally with all the panic about Ras today, I don't feel Obama back up to a double digit lead in Gallup is getting enough copy on this site today. Thats big as well. Suggests something up with those Ras numbers today. And if its the male 34-49 group thats causing the problems, well thats a group that McCain would want to win anyway, so why aren't THEY certain right now?
matt said...
ABC/WaPo tracking poll...
O: 52
M: 44
O+1/
McCain Surge !!!
Maybe it's not so much that the polls themselves lag as it is that there are so few of them (except in the big swing states) and, therefore, the averages tend to lag.
On any given day, you have perhaps eight national polls out. You are very unlikely to have that many state polls out at any given time.
In the best cases (PA for example) that won't have much of an effect. But when you get a state that isn't polled much (RI for example), just taking the raw average can give you some screwy results that don't seem to fit at all with the national trend.
Luckily, Nate's model is prepared for that.
Real Joe said...
matt said...
ABC/WaPo tracking poll...
O: 52
M: 44
O+1/
McCain Surge !!!
You get bonus points for Degree of Difficulty by attempting your surge backwards!!
"Nate, how can you not speak of lag, when the three-day tracking polls were conducted October 26-28 and for example the polls released today by the Associated Press/GfK and Quinnipiac were conducted October 22-26?"
Guys, Nate's model incorporates ALL state polls into the current prediction so that they effect ALL the margins to some degree or other, based on demographic similarities. Now that there is so much state polling coming out daily, it overwhelms the national tracker polling in AGGREGATE.
Verification: "boodat"
[shamelessly moved here from last thread]
Everyone who is accusing Rasmussen and any other independent polling orgs of bias - please get real.
They all share a common goal - to be declared the most accurate pollster in 2008 - why would they compromise that goal with push polling?
And if, say, Rasmussen was as biased as some of you are accusing them of being, how come Nate has them as #2 in his list of list accurate pollsters?
tosec - as in "I'll sec your tose if you sec mine"
There is not a lot for McCain in the polls.
PA just isn't tightening. Worse it is looking increasingly likely that PA just won't be enough, even if their is a significant general tightening.
Nevada would then be the firewall & it's moving towards safe Dem.
Then again Ohio is finaly on the move. Ohio is a monster of a state, and about as easy to change course as a supertanker. Even if the National Polls dive Ohio is probably fixed blue.
So with all of the trackers out today, here's where we stand:
Obama gained in Gallup RV and traditional LV, Zogby, and ABC/WaPo. Gains limited to, at most, 1 point.
McCain gained in Ras, IBD, and Hotline. His two point gain in Ras was the highest, however.
No change in Gallup extended LV and Battleground.
This clearly means one thing: MCCAIN SURGE!!! MCCAIN LANDSLIDE!!!
Blogger hark said...
I'm not getting the explanation. I see four national tracking polls every day here (Gallup, Hotline, Zogby & Rasmussen), and I can track them individually and as an average and see trends and patterns. How do I do that with a specific state? I don't see any four sources with results posted every day. I can't see trends the way I can with the nationals. So to me, there is a lag. It looks like McCain is closing fast in the nationals, but I don't have enough data from a single source to see if that's true with the states.
The nationals aren't closing fast, for one thing. Or rather, they are only closing from one side: McCain's marks are going up. That suggests to me that undecideds and Other are being moved into McCain's column. The question is whether the pollster has good reasons to be making that adjustment (are they pushing undecideds harder, and those undecideds are leaning McCain?). On the state side, on the other hand, I haven't seen the same movement in McCain's number: the whole configuration has been more stable.
The question to ask is whether there is a difference with respect to methodology toward undecideds in the trackers vs. the states?
once more on RAS and todays #s on economy, taxes and SS
how many polls do you see now with Bush at 34% approval ratings?
none!!! but this poll does!! m ras has really gone off the deep end...
Do the state and national polls we are seeing right now include people who have already voted? I. e., when we see a poll taken in the last 3 days show a 2% lead for John McCain in NC, does that reflect all voters, including ones who have voted early? Or is it rather a measure of how people who have not voted yet will vote?
What about this story that early voting among new voters, young people and Hispanics is lagging in Nevada? I'm worried:
http://www.lvrj.com/news/33494194.html
Can't find the site anymore, but I'm pretty sure in a methodology segment on the ABC tracker, it mentioned that they poll a sample of cell phones. Possibly may explain Obama's consistent better numbers there.
becky sharo,
Rasmussen is using 2004 turnout models. He will be woefully inaccurate this cycle.
Tom Delay is fucking batshit lmao...
Thanks for the final note, Nate. I'm finding it very difficult to avoid looking at the polls every 10 minutes and find myself getting very upset whenever I sense the tracking polls tightening. Can these next 5 days go faster...please!!!
becky sharp,
Rasmussen is using 2004 turnout models. He will be woefully inaccurate this cycle.
You can't edit a Rick-Roll.
That's shenanigans.
IMHO, the national polls are tightening now because the sample sets are getting skewed toward the battleground states. Pollsters are conducting so many polls now on a daily basis, they are probably short of people and time. I suspect they are keeping costs down by using their state surveys to populate their national sample, using their model to keep the demographics from their sample consistent with national demographics.
But most of the state surveys being conducted now are of battleground states. In the battleground states, the races are tighter (by definition of "battleground") regardless of the demographics, so the results of the national polls are skewed tighter because the samples are skewed geographically.
[torded] - not sure what to make of that.
otf:
I thought it was Zogby who was using 2004 turnout models.
Rasmussen's targeted turnout model right now gives Dems about a 7 point advantage, IIRC.
"What about this story that early voting among new voters, young people and Hispanics is lagging in Nevada? I'm worried"
why? Obama is still winning the state, even with these "lags".
Nate, your article is welcome but I don't think it definitively addresses the possible explanations for the difference that you do confirm exists.
Just for the sake of (future) research, here's my summary of the working hypotheses to account for the differences. Something that maybe Nate, or Erikson, or Gelman, or the three together should check out in future.
Possible explanations for national vs. state poll differences that have been suggested on this blog:
(1) Time lag. Probably not true, according to Nate, at least looking at average time lag from field to publication of numbers -- though hypothesis remains that maybe this differs for battleground vs. nonbattleground states, based on poll frequency. Could be tested.
(2) Party ID weighting. National trackers may be more consistently subjected to a weighting by party ID than are state polls which for that reason alone have more sampling variance. Could be checked.
(3) Dynamic relation in real time between national and local (state) trends. Not an artifact of the field dates or reporting lags in state polls but a result perhaps of "diffusion" of information via national media, and hence difference in timing at which news events are reflected out on the hustings. Could be tested, as someone suggested, by looking at smoothed trends on time track, comparing states (e.g., VA, FL) with country as a whole (using trackers).
(4). Likely voter models are more consistently applied (and differently applied) in national trackers than in state polls. Hence the LV adjustments affect the apparent trends at each level in different but systematic ways that lead the state poll estimates to diverge from national estimates. Could in principle be tested if we knew a lot about the LV methods of each poll. I suspect the state polls are often not truly LV models, whereas national ones are; but the wide gap between the two Gallup LV models tells us that this factor can make a whole lot of difference.
(5) Compositional effects. State polls are occurring most heavily and hence most recently updated in battleground states, while national trackers track central tendencies of entire US (likely voter) population. I.e., there is some sort of selection of compositional effect in the state polling data. Note: Nate's SuperTracker combines information from states + national tracker + turnout assumption; it would be interesting to see whether if Nate calculated a SuperTracker II -- which did not use any national polls, just the state polls + nearest neighbor adjustments -- it would diverge from (1) Pollster national poll LOESS trend, (2) Current 538 SuperTracker. Would SuperTracker II (again requiring a turnout assumption) "lag" behind SuperTracker I (current one) or the Pollster trend?
Of course there is a lag. We'll see movement to McCain in about a week. Right around November 5 or 6.
Blue in CO -
If history is a guide, Obama likely mostly sticks to flying from airport to airport the last day to get in as many media markets as possible - and that of course would not be the best scenario for a big crowd in Denver.
Also, the pattern usually seems to be going from east to west to follow the sun, so the last stop (after Colorado somewhere) likely is Las Vegas.
My bet for the final Obama campaign rally: somewhere in NW Indiana (Merrillville?) just after dawn, maybe combined with meeting commuters.
I think the national tracker to watch out for is the one that most favors my preferred candidate. If that one starts moving 1 point, 2 points, 3 points in the direction of my preferred candidate, the electoral map will look different. In the business context, jeff angus calls that management by wishful thinking. If Nate Mclouth keeps improving his performance at his present rate, he'll hit 60 doubles and 40 homers next year, too.
marky
is it a dbl digit number? as TPM has it as O_51 vs M_44 (which I shall take thanks!)
word verification: peterkentisabuggeryarsehole
I think the most positive news out of all the polls today, INCLUDING the ones that McCain gained ground is that the Undecided number is disappearing. The ABS/WAPO has only a 4% Undecided.
scalze, which is a more direct proxy for Obama vs. McCain, the % of Democrats vs. Repubs. who have voted early in Nevada, or % of young people and Hispanics?
When will RCP drop the Pew +15 poll from their average? When that happens, RCP goes from O+5.9 (6.0 when they update ABC) to )+4.8 or 4.9. The McCain people are going to scream that the race is tightening fast.
OMG! OMG! If McCain keeps gaining like he is he'll be ahead in 2 weeks!!! OMG!
McCain campaigning in FL when his only path to victory requires winning OH, VA, NV, and NM where he trails distantly is GREAT NEWS!!! For JOHN McCAIN!!!!!
I voted for That One.
Yes We Can
ABC/WaPo tracking poll...
O: 52
M: 44
O+1/
Hopefully a sign of stability or trending back toward Obama.
Btw, whatever thread is newest... for the people who are watching Obama's 30 minute TV spot, post what they think of it.
I suspect that a source of difference between national and state polling is that state LV screens are evolving as early voting progresses (and less-likely voters are actually showing up and measurably so), whereas there is no similar evolution on the national level.
My report of early voting (absentee) in VA: In Alexandria last Sat, voted at 10:30am poll workers said that at least 500 people had already come through there. And this is in a state where you have to have an excuse to vote early.
From overhearing chatter in line, the vast majority were making sure they were getting their votes in to "finally turn the state blue" and almost all were first time voters or voters that haven't bothered voting in the past few elections because as a solid red state they felt they were just throwing votes away so it was a waste of time- so the odds of them being polled? Minimal. Granted that's just in "communist fake" virginia but still interesting to me.
I think the reason why people often speak of a lag in state polling is because all states are not polled every day. So if you look at the state polling overall, you'll witness a lag in many states. For example, California hasn't been polled in 5 days, and Illinois hasn't been polled in 6 days. This phenomenon is more pronounced in the primary campaigns, when there is less frequent polling. It was also more pronounced in the general election campaigns in the past than in this year, as the amount of state polling has expanded in 2008.
So, when considering battleground states only, that are polled so frequently, there may not be any state lag anymore in 2008, but the reasons that people have talked about a lag were real in the past.
Thanks sfergus483 - someone higher on the thread mentioned Obama having a stop left in CO so I was looking for details on that. Doesn't have to be the final rally... I just want to see him. Keep missing him by a hair...
It seems like Occam's razor should apply here.
If the national polls are tightening, but the state polls are not, the tightening is most likely coming from states that are not being polled: the ones that were never close to begin with.
It's not a surprise given the McCain campaign's message over the last week, which appealed mightily to the base, but rang hollow among independents.
Texas, Oklahoma, Nebraska, etc. are probably getting redder, which would lead to a few points tightening nationwide -- but the McCain message isn't playing (or cutting through the noise) in the states that Obama continues to focus on.
I reckon Obama will have something planned for the last day, something unusual. Maybe a big rally somewhere in the morning and an evening rally somewhere late, maybe in Indiana or Missouri, somewhere nearish to Illinois, so he can be in Chicago election day to vote.
Nevada swamp land at that...
Nevada actually has a lot of swamps. A topography consisting of a series of mountains and valleys will do that.
The past couple of days I've been hearing/reading quite a bit about the effect of "undecided" voters (granted, from Repub operatives and Mccain's pollster). Specifically, they have been saying that most will choose McCain and that "what you see with Obama is what you'll get". If this is true, then the poll margins for Obama must be large enough to beat back this sway, IF all undecideds do break for McCain. See the "McCain Pollster Predicts Massive Turnout" under Politics '08 on Politico. Any thoughts/data on this? Keepin' me up nights...argh.
Mouse-
City or FFAX Co.?
ver: commi
What really cracks me up is how people freak out when Obama's lead falls below 5 %. If he even declines by one percentage point liberals think he's sliding into the tank, and conservatives fantasize that its the beginning of a McCain avalanche of victory. It's hilarious how much everyone over-reacts.
Let's face it. The Democratic nominee has lead in the majority of polls for almost the entire election cycle since June with one exception immediately after the GOP convention when McCain got a temporary convention bounce and Palin was introuduced before people discovered her flaws. That's it. The GOP has lead in the polls for all of one week out of 5 months of polling.
How does a person who has been so far behind suddenly become a victorious dynamo because his opponent's lead narrows by a point? Use your heads people. McCain is in a worse position than almost any major party Presidential nominee in modern election history. Everyone loves to try to compare Obama's numbers to some fantasy campaigner leader who lost their lead, but no one really looks at McCain's position. This guy is in the absolute basement of poll numbers compared to history. No one, and I mean no one has ever reversed such a losing position in modern election history. Yet conservatives think a last minute tv ad or snappy speech will somehow reverse what has conistently been a losing position. Heck, folks, McCain only has 6 days left to change all this. Where is his ground game or his GOTV effort? He has nothing except poorly attended rallies and faux plumbing spokespersons to boost his effort. I'm not saying it's totally impossible for McCain to win, but let's just say I think it might take an extinction level asteroid strike into the earth's crust to shake up this race.
And don't forget Obama has his primetime speech tonight. That I'm sure will have a nice last minute sweetening effect on those poll numbers. He's a savvy campaigner.
So have a little hope people, and remember all those people standing in line in early voting are not waiting to vote on a loser or a fear peddler.
Post a Comment