Today's Numbers - October 31

It's not an October Surprise if you wait til November:









So much for tightening. Happy Halloween everyone. I'll see you tomorrow.

Today's Numbers - October 30

Of course McCain likes his internal poll numbers, because his external poll numbers still suck:









Four days to go. Deep breaths.

Another day where, even if the national poll trackers don't look great for Obama, the state poll numbers seem to be holding up. I think what we're seeing is in part a product of Obama's efforts to stretch the electoral map.

We can tend to think of the election as being held in three sets of states: the lean-Obama states (NH, MI, WI, MN, IA, NM, OR), the lean-McCain states (NC, IN, MO, GA, MT, ND), and the true swing states(FL, OH, PA, CO, NV, VA). Obama had two possible strategies going in. First, he could attempt to hold down his own states while picking off a couple real swing states. This would mean the election becomes make-or-break in a few select states, probably the big-3 of Ohio, Florida, and Pennsylvania. This is what Kerry did in 2004, and we know how that strategy worked out.

Alternatively, he could make a play in all states where he had some kind of shot, and force McCain to spread himself thin. This is what Obama did, particularly in the post-convention period, and because of his fundraising advantage he managed to get to where we are now--which is that Obama is ahead by a small margin in a lot of states that were previously red. All he needs is a couple of them to go his way to win, and the odds of that look good. Obama has forced McCain to spend money advertising in red states like Georgia and Missouri, which were thought to be safe even just a few weeks ago. The goal of this strategy is simply to bankrupt McCain and keep him from being competitive in all the states he needs--much as he did against Clinton in the primaries. He doesn't need to win every state he campaigns in to make it worth his while, because McCain has limited resources to spread around.

The interesting trend in the past few days is that while Obama has gained small leads in states where he was previously behind by a couple points (FL, OH, MO, NC), his leads have diminished in a few other key states (PA, VA, CO) that are still part of a solid victory strategy. What appears to be happening here is that Obama's making the choice to diversify--rather than run up large leads in a few key states, he would rather have small leads in a bunch of key states, and so is devoting resources accordingly.

This move is safer in one way and more risky in another than making a play for a few key states in these final days, even after forcing McCain to write off any hope of inroads into Kerry states not called "Pennsylvania". On the safe side, it minimizes the chance that one or two states' screw-ups (bad turnout predictions, voter disenfranchisement, long lines, outright fraud) will jeopardize the entire election. On the risky side, keeping small leads gives McCain a slightly greater chance of winning back a whole bunch of the states at once, if something were to happen to shift the national mood a few points in his favor.

I think that the Obama camp is going with the right strategy, as hard as it to watch. The national polls do seem to be trending in McCain's direction. I said yesterday that much of this change may be base consolidation in red states, given all the red meat he's thrown them lately (accusations of socialism and terrorisms are always redneck crowd pleasers). But even if that's wrong, there are still only four full days left until the election. Maybe if the election were November 15, McCain could come back in a legitimate sense, but four days is simply too short to make up a full 3-5 points in most of these states.

This is especially true in the age of early voting, when large numbers of voters cast their ballots during Obama's peak of the past two weeks. What's more, the trends in early voting by party and demographic suggest that it's Obama's base that has come out early, meaning that the opportunity for McCain to peel away some of his support is even smaller than it would have been in previous years. McCain's whole strategy in the past month has been to run a Bush/Rove-style campaign--throwing every kind of shit at the wall and seeing what sticks--but this isn't what he needs to be doing. This campaign style plays to the base, but as we saw in the 2006 elections, the Republican base is not nearly the size it used to be. What McCain needs right now is independents, and he's unlikely to get them this way. The campaign he's running now makes absolutely no sense--but I'm not complaining.

See? I told you so.

I was trying to wait for this before publishing the Wednesday numbers, but Nate Silver was too busy palling around with Keith Olbermann. Check out today's state-level poll numbers. This should reassure anyone who's freaking out over the national trackers.

And as a bonus: the latest early voting numbers. Looking good...

Today's Numbers - October 29

The "somewhat less than $4 million" edition:









Repeat after me: Five days to go. Don't. Panic.

Another hint of tightening in the national trackers, but it's important to note that we've seen no equivalent shift in the state polls. The most likely explanation for this is that undecideds in red states are finally coming home to McCain, much as Democrats came home to Obama over the past few months. What this means for the ultimate results is... nothing. The election is not decided in the red states, or the blue states, but in the swing states. And in the swing states, Obama's doing just fine.

Today's Numbers - October 28

Because aren't we all, in our own ways, preparing for 2012?:









I'll make this quick, since there's nothing in here that's really concrete. The national trackers took a bit of a dive today, and shook up the confidence of the markets with them (though only to return them to where they were 3 days ago). This could be a bit of a weekend effect (which looks far more ominous than normal this close to election day), or it could be real change. We'll see tomorrow, when Saturday's results drop out of most of the trackers. Til then, not much can be said. Though with 6 days to go, Obama could continue to lose 0.6 points a day (which is highly unlikely) and still win the popular vote by a comfortable margin (3.5 points).

6 days. Wow. Less than 168 hours from now, barring a 2004-Florida-style fiasco, we'll know the next president. Break out the champagne and Xanax.

Just a heads-up: this weekend, I'm hoping to post numbers both days for once (assuming polls are as voluminous as they should be), and to also leave you with a "What to watch for" election night prep guide. Based on the predictions across the country, it will let you know definitively when you can start celebrating (or mourning) your candidate's victory (or defeat). The networks are still paranoid about declaring a winner too early after 2000, but using the wonders of mathematics, I'll show you what constitutes a win and a loss for each candidate. Stay tuned.

Voter Deception

One of campaign season's most notorious and enduring tricks has reared its head, Putin-like, in Virginia: Phony flier says Virginians vote on different days. This is one of the oldest tricks in the book, yet somehow people still fall for it. It's closely related to the outstanding-warrants flier scam that we've already seen in the past few weeks, in that its goal is to trick potential Democratic voters into staying home on election day. It seems like these particular scams always show up somewhere in every election cycle. When I was campaigning in Ohio back in 2004, we noticed that somebody was spreading rumors amongst the student population that voter registration would lead to jury duty (which is true in some states, but not in Ohio at that time), and that it was illegal for students to register at their college address (absolutely untrue).

Let's hope people have smartened up a bit over the last four years. I must give Virginia credit, though; last year, the legislature passed a law making it a class-A misdemeanor to purposefully mislead a voter. Such a law specifically combats campaigns like these. A similar bill was introduced nationally last year but so far it hasn't been put to a vote.

Today's Numbers - October 27

Even Ted Stevens is still a safer bet than McCain:









Mondays are a wonderful time around here. As you can see, damn near everyone updated their numbers today. We've seen the first new maps in a week from Election Projection, Chris Cillizza, Chuck Todd, and Steve Lombardo (of LCG--Lombardo Consulting Group). And with these updates comes a continued uptick in Obama's lead, reflecting the small but steady improvement in the state polls over the past week.

The national polls actually ticked down a little for Obama compared to the past couple days, but it remains to be seen whether this is a sampling issue or actual change. I'm guessing it's sampling--after all, today's tracking poll releases include predominantly weekend respondents, who can differ systematically from respondents who can be reached during the week. Many pollsters believe that these differences can be compensated for by weighting respondents according to demographics, but this doesn't entirely solve the problem; as we see in data on the differences between cell-only respondents and landline respondents, basic lifestyle choices can have effects that are not properly accounted for by demographic weightings. Comparing those who are home to answer the phone on weekends with those who are not, we would imagine the latter group to have less traditional jobs, be located in more socially-active areas--they could even be out volunteering on a campaign!--and all these factors could lead to the weekend samples being less favorable to Obama for reasons that cannot be dealt with simply by weighting.

As a quick semi-scientific analysis, look at the Rasmussen numbers over the past few months Looking at the Sunday to Monday change in the last 10 weeks (since the start of Convention season), we see that Obama lost ground to McCain in 5 of those instances, stayed even in 3, and improved in only 2. The sample size is not big enough to draw any definitive conclusions, but it's more likely than not that weekend polling does have an effect.

And on the win prediction front, Obama continues to creep up slightly, but McCain's floor hasn't moved much since I mentioned it two days ago (down about a point on average). There's not much hope for a McCain comeback at this point, especially with early voting under way, heavy, and looking to favor Obama's demographics. The remaining hopes for McCain are, as a I said before, resting on some kind of screwup. But with courts shooting down efforts to illegally remove voters from the rolls, those hopes are looking pretty slim at this point. (Full disclosure: I was involved in the data work for that lawsuit in Colorado.)

So with that, sitting here near two in the morning, I'll leave you with a parting thought:

OMFG! ONE WEEK TO GO! And a good thing, too, because these last two years have probably taken ten off my lifespan.

Today's Numbers - October 25

Watch out, or I might decide to go rogue:









Sorry for the poor image quality on the numbers--on the road, I'm forced to use MS Paint instead of my usual Photoshop, so.

I think we've figured out how far the markets can go. Right now McCain is given a 13-15% chance of pulling this off, which hasn't changed significantly in the past five days--despite a small but real increase for Obama in the poll numbers. I think what we're seeing here is that the markets have decided that McCain realistically cannot pull this off by normal means, but there is a chance that some screwup will give it to him. That screwup could be a massive voter disenfranchisement (for which efforts are underway, but struggling), outright fraud in some key states (hello, Diebold!), a complete failure of the Obama turnout operation, or a massive, systematic failure of polls to accurately gauge public opinion. But what McCain won't do is come back in any legitimate sense. So while this thing's not quite over, it's certainly looking about as over as it's going to get before November 4.

See you monday.

Today's Numbers - October 23

Here John, have another rat:









Not much analysis today (was grading midterms all evening), but suffice it to say that life is looking up for Obama. The McCain camp is imploding. My call? Too late for a strategic comeback at this point. All he can do is hope for either a bin Laden video or that the polls are all wrong--and I'm not sure even those would save him. Surprisingly, I'm finally starting to agree with Nate Silver's numbers.

Traveling tomorrow, so either a late update or else a Saturday update. Stay tuned.

Today's Numbers - October 22

This is getting pretty repetitive:









Things are pretty flat in the polls again today, so the win percentage numbers keep moving up incrementally. (The polls were actually all over the place--for example, we know that Obama's lead is somewhere between one and fourteen points, amusingly--but they averaged out to about the same place they were yesterday, and the day before, and the day before that...)

I'm beginning to repeat myself, but time is running out for McCain, and he's obviously just spinning his wheels right now. The NYT reported today that McCain's basically going all in with Pennsylvania. In my mind, a smart move, but also a sign of how desperate things are. The electoral map is this right now: Iowa and New Mexico are solid pickups for Obama from 2004. That brings Obama to 264 EV. Add Colorado or Virginia to the mix (273 / 277 EV), and it's game over. However, if PA flips to McCain, Obama can have both VA and CO and still lose (266 EV). McCain's calculation is that he's got a better chance of taking PA than he does of holding BOTH VA and CO, and I think he's right. McCain's only strategy at this point is stoking up fears of Obama's supposed -isms: socialism, terrorism, anti-americanism, etc. That won't fly in Colorado, at least not as well as it might in rural Pennsylvania. And I think Pfotenhauer blew McCain's last hopes of keeping Virginia. (The Daily Show link is a great clip to watch regardless.)

McCain's odds of taking PA? Not good. But it's about all he's got right now. And if Obama manages to also take any of Nevada, Missouri, Indiana, North Carolina, Ohio, West Virginia, Florida, or Georgia, or North Dakota & Montana together? Game over.

Cross-Promotion

I was thinking of setting up a "Recommended Reading" list of links on the right of the page (by the links to poll trackers / campaign maps / etc.) to interesting political blogs. So if you write a blog that you'd like me to link to, send me an email (andrew [at] poll3.com) and if I like what I see I'll put a link up. All I ask in return is that you do the same. Thanks!

P2004SD - Post-2004 Stress Disorder?

It's a solid candidate for the DSM-V:

• Democrats' Gloom Deepens
  

This is what I get for not going to therapy like a good New York liberal should. I would quote some Kurt Cobain lyrics about paranoia right now, but I think that would be overplaying my hand a bit, don't you?

Today's Numbers - October 21

A day without movement is a day without sunshine--for McCain:









(the obvious melanoma jokes aside...)

So almost no movement in the EV numbers--almost all the movement in the lower EV number is a product of Pollster switching Florida to toss-up status, sapping 27 EVs from Obama in their calculation because of a string of tight polling. Otherwise, no movement there. Despite the GWU/Battleground figure I had a mini-freakout about earlier, most of the popular vote measures actually went up today, owing to a few very nice numbers for Obama in some non-tracking polls released today. (Our number goes down because the Iowa markets--perhaps the reason I've been popping antacids every day lately--jumped around again today and settled on a far more reasonable number than the +11 they were at yesterday.)

Which leaves us with the win percentage numbers--which, it's easy to see, are at an all-time high. Two weeks until election day, not counting the early voting already going on (and which the Democrats seem to be benefiting from), and time is running out for Team McCain. The bump in the popular vote polls today seem to suggest that the tightening we've seen over the past week has flattened out or may even be reversing. I think the Wright issue is almost inevitable at this point--I suppose McCain may still try to preserve what's left of his legacy by not bring it up directly, but others will on their behalf--but it's old news to pretty much everyone and is likely to turn off just as many people as it turns on.

So this is why the win percentage numbers are high and getting higher--what more can McCain do? Since his campaign is obviously directionless and undisciplined, in start contrast to Obama's (Biden's relapsed foot-in-mouth disease excluded), and with little time, money, or credibility to recover on the merits, he's down to hoping for either a) the polls to be way, way off, or b) the previously-mentioned "external contingency" to occur. The odds of the latter happening are decreasing by the day, which leaves McCain with a surprisingly typical Republican position to rest his hat on: praying that science might be wrong.

-------------

Also: Note that I've finally updated the links on the right-hand-side of the page to include the new poll trackers I've added since I originally set up this page, so if you want to know where I get this data, it's all over there.

Tracking Polls

I have to give a nod to FiveThirtyEight today for putting out a great tracking poll primer which gives a good overview of all the major tracking polls. Nate's obviously got a lot more time to work on this stuff than I do, and in moments like this it pays off.

One Point?

That's what the GWU/Battleground tracker says today. An outlier, of course, but it still worth a couple heart palpitations.

Today's Numbers - October 20

15 days to go to the election, but I won't make you wait that long:









We see a continuation of the incremental tightening I discussed earlier in these numbers. Nate Silver isn't too worried and says that it's hard to imagine that McCain can make up that kind of ground in two weeks without some "external contingency" shaking up the race. I'm not so certain. I agree that it's hard to make up a half point a day, but that's presuming we really are in a 7-point race. What worries me is that maybe the average poll is getting it wrong, that somehow all these new voters for Obama will act like new voters in other elections and not actually show up in anywhere near the prophecied numbers, or that conversely lots more Republicans will turn out (motivated by rantings about terrorists and socialists--or for that matter, black people--in the white house) than the polls predict. Maybe the race is closer to the 4-point race claimed by Rasmussen and GWU/Battleground in their tracking polls today, or even the 2-point race AP-Yahoo found among adults in a poll last week. In that case, two weeks is a much longer time than it is with a 7-point lead. I don't mean to be all doom-and-gloom about it, but having worked on the 2004 campaign in OH, PA, and NH, I know what it's like to watch everything you worked so hard for fall apart, and I can't stand to see that happen again.

On a totally different note:

The interesting story of the next few days may be this. Obama is leaving the campaign trail to fly to Hawaii and be with his ailing grandmother. The question is, what does John McCain do? He can stay on the trail and risk looking vulgar (taking advantage of Obama's absence while Obama shows his family cred), or he can lay low and give up 2-3 days out of 15 during which he might do something to get back in this. It looks like a lose-lose for McCain, but it will be interesting to see how he decides to handle things.

And finally: Wright alert. If McCain decides to use this while Obama's away, it could be suicidal, especially since Joe Biden would never let it slide--and a Biden tirade in Obama's absence would be major. As TO says, get your popcorn ready.

A Tightening Race

New numbers to come this evening, but right now I want everybody to go look at two things:

• Pollster's estimate of the popular vote trend. (Of all the sites tracking popular vote numbers, I tend to trust Pollster's the most, since they include every scientific poll available, use regressions to estimate the over-time trends [as opposed to simple averages], and their model responds accurately to change in the national mood.)
  
  
• Sean Quinn's warning about the dangers of the complacency that can come from favorable polls.
  

This race has tightened up over the past week--there's no denying that now. And there are still 15 days left until the election, which is plenty of time for things to close up even more. So to all the Obama supporters out there: Don't get cocky. There's plenty of work left to be done. The campaign still needs volunteers. They still need money, even despite yesterday's announcement--much of that money has already been spent just on building and keeping the lead Obama has now, and it will take more to keep it going through election day. And for those of you who live in swing states or have friends there--make sure nobody you know decides to stay home on November 4th. There's no excuse for it, unless they've already voted early. Otherwise, keep pestering, and if necessary, show up on their door and drag them to the polling place. They'll thank you come January.

Two Stories

There were two big stories today that everyone has probably heard by now. Early this morning, the Obama campaign announced a HUGE fundraising haul in September of over $150 million. A few hours later, Colin Powell endorsed Obama on Meet the Press. Let's look at the potential impact of each of these in turn.

To put the $150 million in perspective, the previous one-month fundraising record--set by Obama in August--was $66 million. With his total fundraising exceeding $600 million, Politico notes that he's on track to surpass the combined fundraising of Bush and Kerry in 2004. (This entry will probably refer to Politico more than usual, since I'm watching the Sox game on DVR and don't want to spoil it by accidentally seeing a score report on a mainstream news website.) This is what has allowed Obama to dramatically outspend McCain on advertising, including a half-hour of airtime on CBS and NBC on October 29th.

So the actual haul is nothing but good news for Obama. But my question is, what impact will the announcement of this amount have? There are three perspectives on this. It could have a positive effect, since it could add to Obama's perceived inevitably and serve to further discourage McCain supporters. It could have no effect, since the vast majority of Americans don't care about campaign financing. (And as a certain political type I know pointed out to me tonight, the vast majority of people also have no real idea how the campaign finance system works anyway.) My take, though, is that there's potential for this story to hurt Obama. For one, it allows McCain to bring up the argument again that Obama lied by saying he would take public financing and then backing out of it. And as far as negative attacks go, that may be the most substance-based McCain has going for him, so he may be able to spin it effectively. What's more, the people who are still undecided right now may be the same people who are most likely to react negatively to the announcement. You could imagine that the people who are still uncommitted are the lesser-of-evils types who generally dislike politicians and are unlikely to be fans of large amounts of money in politics. This announcement could turn them away from Obama. While this is admittedly speculative, there is some evidence for this. Looking at the Harris poll linked to above, the people who are most likely to respond negatively to a generic candidate opting-out of public financing are independents--who are also the most likely to still be undecided.

I can't say whether this will make much of a difference, but I think the worry about it was what caused the campaign to announce that number early on a Sunday morning when the main news story was inevitably the potential Powell announcement. A smart move, I think, since the money story gets buried under the Powell story, and because it's too late for Sunday's papers it gets stale by the time Monday rolls around. Any potential damage is largely blunted by the timing, while the people it might help with (the overinformed partisan types--which I suspect includes most of us) will hear about it regardless.

As far as the Powell announcement goes, the prevailing wisdom ranges from "so what?" to modestly good news for Obama. Endorsements typically don't change any minds directly, but (a) Colin Powell is not a normal endorser, with high favorability ratings from both Republicans and Democrats, and (b) in the big picture sense, it wins a good news cycle or two for Obama, takes McCain out of the headlines for that time (taking away his chances to make news), and may help to shore up Obama's foreign policy credentials. My take is that it will probably not have a perceivable impact on the polling numbers, but what it will do is shore up some of the weak Obama supporters (by allaying their doubts about him) while decreasing the likelihood of some weak McCain supporters turning out. So while I don't think it'll affect the polls, I'd bet it's worth half a point or so on election day--and if you don't think that's a lot, remember Florida in 2000.

Today's Numbers - October 18

McCain's slow creeping continues, and I mean in the polls:









A little movement away from Obama's highs of a few days ago, but nothing to panic about yet--just more to stay vigilant. The only number showing a substantial move is the top EV number, and that's just a product of an updated Chris Cillizza number; the Washington Post has three different dates for the same numbers on their site--one on the generic campaign map, one on their estimated campaign map, and one on Cillizza's blog--and I've finally settled on checking Cillizza's blog every day because the numbers on the maps aren't updated regularly. (They also have a map based on RCP data, but I don't include that because it borrows the state-level estimates, so it would essentially be double-weighting RCP, and it is also undated.)

Other than that, not much movement except for the slight inching toward McCain that I mentioned yesterday. It's not being reflected in the EV numbers just yet, but we see the popular vote totals going down a bit (even more if you discount Iowa's number somehow going up to an incredulous 12.9% margin), and the win percentage forecasts are just slightly off their highs. As I said before, no need for panic, but it's just a reminder of the need for Obama supporters to stay vigilant. McCain seems to have stopped the bleeding at this point, by tempering the personal attacks and instead returning to the classic republican tactics of attacking on taxes and emphasizing Obama's "liberal" record. So it's time to stop planning for blowout victory parties and keeping focusing on the race as it is: one wherein Obama has a respectable lead of probably 5 points nationally, but where McCain is far from defeated. Complacency now can make 5 points dry up pretty quickly by November 4.

Since the pattern of little weekend movement has been clear for the past couple weeks, there probably won't be a numbers update tomorrow unless something major happens in the polls. I might post a bit more commentary if anything exciting happens between now and then (Powell endorsement anyone?), but otherwise I'll see you Monday.

Today's Numbers - October 17

Odds of a Red Sox comeback? They're at least better than McCain's:









So baseball was definitely the more exciting sport to watch yesterday. These numbers include a minimum of post-debate data, mostly in the popular vote numbers--which showed almost no change. (The national trackers went down by a fraction of a point on average over the past two days, more yesterday than today--so not debate-related, since today's numbers have the first post-debate interviews--but the Iowa markets are once again showing an Obama blowout. The number looks screwy to me, but I'm trying to avoid cherry-picking, so interpret it as you will.)

From the looks of the latest individual polls, Obama's wave may have crested a bit last week and might be coming down slightly, so look for the electoral vote numbers to come down a bit over the next few days as these results filter through. That said, the markets don't seem distressed by it--they're maintaining their post debate bounce--and I agree. Obama's got some breathing room, and he can take a point's loss here and there. Looking at the latest polls on Pollster, Obama supporters should be comfortable. McCain can have Ohio and Indiana if Obama gets to keep Colorado, Florida, Missouri, Nevada, New Hampshire, New Mexico, North Carolina, North Dakota (!?!), Pennsylvania, and Virginia. That seems like a pretty good deal to me.

Today's Numbers - October 15

Joe the Plumber, meet Andrew the Political Scientist:









Not much changes in the electoral vote or popular vote numbers today. The uptick for Obama in the top EV number is mostly an artifact of a weighting adjustment; digging into the Washington Post site, I realized Chris Cillizza's prediction was older than I thought, so increasing the age of that data point (the worst for Obama in that set) upped Obama's overall numbers. The rest of them moved around pretty much at random. The popular vote increase is due pretty much solely to post-debate increases in the Iowa markets.

Looking at the win percentages (these were taken at about 11pm EST), we see that the markets think the debate was a definite plus for Obama; not necessarily that he won, but that he certainly didn't lose in any way that would help McCain, so that missed opportunity for McCain made the probability of an Obama win jump significantly.

My take on the debate? I think on the merits, Obama won by a small margin, but in the grander scheme, that means a huge victory for him. I was concerned at the beginning, and agree with the pundits (at least those on CNN that I saw) that the first 30 minutes showed McCain much better than he's been at any other point, and better then Obama was. But as the debate went on, McCain grew steadily worse, starting with Schieffer's question baiting him to make his Ayers and ACORN charges. He did not disappoint, but came off as petty and small, and I think Obama's response, though a bit muted, was straightforward and sincere. (I'm surprised, though, that McCain didn't push on the paying ACORN point; Obama flatly denied it, which isn't quite true, but McCain moved on.)

But more than that, McCain presented himself physically and emotionally in a very poor manner. We saw him jumpy and excitable, grinning maniacally at totally inappropriate times, and looking absolutely shellshocked for about 9 seconds after Obama explained the small business exemption to his health plan's employer mandate. He kept repeating things even after Obama had discredited them, as if he was reading from a prepared script he couldn't stray from. He looked condescending and unfriendly, ill-tempered, and ultimately un-presidential.

Obama, on the other hand, really showed himself well. Aside from a few overly-broad smiles when McCain made false statements and a few unconscious head nods while McCain accused him of wanting to raise taxes, he carried himself with an overall composure that I think most people will find reassuring, especially in a crisis as we're in now. He came across as very even-keeled and mild-tempered, and showed the appropriate seriousness and gravitas when discussing important issues that McCain sorely lacked. Throughout all three debates, he presented himself as cool in the way JFK was cool, and despite being far younger than McCain, Obama was consistently and definitely the adult in the room, to McCain's disgruntled adolescent.

When this is all said and done, I think this will be the story of these debates. Not the talk of issues, which didn't give us much that was new or exciting, but the perceptions of the candidates. Obama's task was to make himself familiar, comfortable, dignified, and reassuring to the American people, and he succeeded wildly at that. McCain, on the other hand, needed to show that he could offer a way forward that was neither a continuation of the Bush years nor a watered-down version of the change Obama offered. I think that on the policy front, he began to make progress on that tonight, but any progress was overwhelmed by the repulsiveness of his personal performance. McCain came into these debates with a reputation for candor, dignity, and competence earned during his years as a media darling. He blew that all away by showing a level of irritability, aggression, negativity, sarcasm, and desperation that would make Dick Cheney proud. In the final verdict, Obama's debate performances made Americans feel more comfortable with the thought of an Obama presidency, while McCain's performances made Americans have second thoughts about McCain.

In his liveblog of the debate, Nate Silver posted at 10:31 to say, "Congratulations, President Obama." I'm not quite there yet, but despite my "don't jinx this dammit!" Sox-fan inclinations, I'm having a harder and harder time seeing how McCain could pull this out. A week ago I said that McCain's only real shot was to put together a comprehensive economic plan and unleash it this week, culminating in the debate. He did put together an economic plan, but it wasn't comprehensive and has very little in the way of ideas, and he didn't promote it much at all tonight, certainly not enough to give it any real attention or credibility.

Going forward, McCain doesn't have any real opportunities left to get a game-changing level of attention unless he does something dramatic, and there's not much that he can do at this point. Meaningless stunts won't work--he already cried wolf and spent that credibility on his campaign suspension act. Now he'd need to up the ante by a factor of ten to get the same level of attention. Right now, Intrade is trading Palin withdrawal contracts at 40 cents to pay off $10. To my mind, that sounds like a bargain, because it would take a shake-up of that order to give him any shot of winning this thing. So while I won't say it's over, it got a hell of a lot closer tonight. Don't pop the champagne just yet, but it wouldn't hurt to get an ice bucket ready.

Tonight's Debate

So this would seem to be McCain's last chance to make some news. There's been talk of the Ayers charge coming out, but as the Obama team seems pretty sedate about it (by their standards), expect Obama to have an effective retort. McCain's goal tonight is to earn some headlines and choice quotes, and generally to hurt Obama. Obama's is to put us to sleep. I was just talking with a colleague who was heading home to watch the debate, and she was complaining about how boring the last one was. The thing to note, though, is that boring debates are good for the frontrunner. Anytime something happens, it could go either way, but if nothing happens then the frontrunner remains the frontrunner.

Obama would love to have the election today (and in early voting states where that's kind of happening, he's doing quite well), but the next best thing for him is (to continue the bad sports analogies) to take a knee. Lots of people on the left are frustrated that he's not attacking McCain right now, but he's playing this very smart. Attacks can backfire (as we seem to be seeing right now with McCain), and ultimately it's not worth the risk. If his lead were one or two points on average, it might be worth it, but he has a comfortable enough lead to sit on it for a while. Despite my prior cautions about getting too excited just yet, I do believe that, if the election were held today, Obama would win--my skepticism is directed only at the assumption that the conditions will be the same in three weeks.

So Obama's goal for tonight, and for the next three weeks in general, is to keep things boring and run out the clock, so that the electoral landscape is the same November 4th as it is today. That's his best path to victory, and it's clear the campaign knows this, so don't expect any fireworks from him tonight. He may well take McCain on if attacked, but probably won't go after him unprovoked. After all, at this point, anything but a total collapse is a victory for Obama, so look for a quiet, competent, solid, and yes, boring performance from Obama tonight.

Today's Numbers - October 14

I'm starting to feel like a killjoy:









Things are looking better and better for Obama. Check out the CBS/NYT poll, which isn't even factored into most of these popular vote numbers.

Analysis tomorrow.

Today's Numbers - October 13

Because I need to talk about something other than the Sox:









Regular readers will note the inclusion of a couple new trackers today: CQ Politics' electoral map and the Washington Post electoral map by Chris Cillizza. The inclusion of these two maps makes the total numbers not exactly comparable to the previous numbers, but only by a small margin. Without these, the tossups-allocated number would have gone up a bit more, the tossups-excluded number, a bit less.

The popular vote numbers have gone down by over a point; this is mostly due to a correction in the Iowa markets that brings them more in line with the rest of our numbers (from Obama +11.2 to +7.6). I'm not a huge fan of the Iowa markets (though others are), simply because the low cap on investment amounts makes investors less risk-averse than they might otherwise be if more money were involved. There was also a marginal decline in the national vote trackers, but nothing to really note.

The win percentage figure continues to go up, but this is mostly due to a big increase in Obama's win percentage from the FiveThirtyEight figure. This may be the last straw for me--there's definitely something wrong with Nate's model.

My specific problem with the FiveThirtyEight model gets a little technical, but bear with me. FiveThirtyEight assumes that as the election gets closer, the polls will get tighter. What this means, in effect, is that whoever is in the lead is predicted to be still in the lead but by a smaller margin by election day. The simulations then run in the context of this framework, allowing for some margin of error in the polls.

What this doesn't account for, however, is the fact that such a simplistic prediction gives an undue advantage to the person who's ahead, because in the model the leader will retain the lead at a smaller degree and any other variation is random. Since Nate is a baseball statistician by trade, I'll give a proper analogy. If his model were predicting the results of a baseball game, he would be assuming that whichever team was ahead at a given moment had run into a string of good luck, and more likely than not the game would tighten by the end. Whether that's true or not, it doesn't help much in predicting how each at-bat will play out. When it comes to that, his model assumes that the variation will be random between at-bats.

But we know that's not true. When a pitcher is throwing well, he throws well to batter after batter, but if he's off, he's off to everyone. The same works for polls. If a candidate improves one day, there's likely a reason for it, and that reason is likely to carry over to the next few days as well. So a proper adjustment for tracking the national mood between now and election day would be to model each day's electoral change individually, with every day's prediction factoring in the previous few days, in each course of the simulation. The regression-to-the mean assumption could still be factored in, but in such a way that provides a balance between allowing for trends, compensating for lucky polls, and accounting for potential diminishing returns (since each new supporter gets the candidate further from his base, and is thus harder to convert).

This sounds complicated, but it's really perfect for a straightforward time series analysis. Find the autocorrelation between days, test for the general trend (which Nate assumes to be asymptotic to 50/50), and see to what degree the absolute margin affects the day-to-day variation. We have plenty of data to work with--we could use this year's polling data, or previous years which run straight through to the election. After establishing the parameters of the model, then we set it to work simulating electoral outcomes.

With that sort of correction, I'd expect the FiveThirtyEight to forecast the probability of victory at a much more reasonable level. As it stands now, the prediction of a 94.9% chance of an Obama victory is ridiculous, and it points it the biggest flaw in the FiveThirtyEight model: things happen, and it has no way other than completely-random variation to account for them.

</rant>

One last bit of housekeeping: Gallup introduced an alternate likely voter model today. Pollster and (sigh) FiveThirtyEight have commentary. I tried to find the internals (the breakdown of the samples by demographics) but failed, but from Mark Blumenthal's report on Pollster, it seems the standard LV model undercounts the youth demographic when compared to past elections--an assumption that is absurd in this election, given the increased youth turnout rates in the primary. Frankly, I don't like either model. Excluding past voting history is a foolish throwing-away of information. Instead, Gallup should factor into its standard model when the voter registered, using data from past elections as a guide, as well as whether they've been contacted by a campaign. The closer to the election the voter registered (all else being equal), the more likely she is to vote, since enthusiasm may wane over time, and voter contact is a very strong predictor or turnout. This allows them to factor in past voting behavior, while still accounting for the dramatic increase in mobilization this election cycle.

As I teach my American Politics undergrads, the best predictor of whether someone will vote is whether they voted last time, and the second best predictor is whether they voted the time before that. But being mobilized by a highly-efficient and effective campaign apparatus is also a clear predictor of turnout behavior, and Gallup's likely voter model does not account for this in any way beyond the voter's stated intention.

More Reading

On the "is it over?" topic:

• Daring to utter the 'L' word: Obama on track to a landslide
  
  
• Is Obama heading for a blowout victory?
  
  
• What will an Obama blowout mean?
  
  
• Grim picture for McCain
  
  
• McCain down, but can Obama hang on?
  
  
• National Journal's Political Insiders Poll
  
  
• Going Obama's way
  

UPDATE:

• How John McCain could still win. Shapiro's columns are always worth reading.
  

Irrational Exuberance vs. the Continuing Panic

The debate continues on both sides as to whether John McCain is just a little screwed or completely screwed.

Leading off, John Harwood in the New York Times today makes the case that, in Gallup's 72-year history of polling the presidential election, only one candidate (Reagan, in 1980), has overcome a 7-point deficit to win the presidential election. He then cites Larry Bartels, a notable professor of political science at Princeton, who places Obama's odds of victory at over 90%, since "front-running candidates have typically preserved three-fourths of their October leads" since 1948 (Harwood's wording).

Nate Silver over at FiveThirtyEight today responds to Harwood's article in a generally favorable fashion. He does clarify a few of the mathematical flaws in Harwood's article (such as his failure to mention that not every one of those 18 elections polled by Gallup gave one candidate a 7-point+ lead, and that many of the leads in non-close elections were even larger), but generally comes to same conclusion--which, nicely enough, places the odds at about the same level as his model's predictions.

As I said yesterday, though, this is no time to get cocky. With reference to the above arguments, I simply don't think they're valid comparisons. Here's why:

• There's no comparing the level of information voters have access to in this election to the amount they had 70, 50, or even 20 years ago. When Gallup started polling, TV news didn't even exist, and it took until the Reagan era for national news to consist of more than just three networks airing an hour each of evening news. Today, by comparison, we have three networks (CNN, MSNBC, and Fox News) airing round-the-clock political coverage, along with a host of other networks (even including HBO and Comedy Central devoting at least part of their broadcast day to election coverage. Add to this the incessant attention paid to politics on the internet, and what this means for us is that the election can change far more quickly. New information which would have taken days weeks to disseminate through newspapers and magazines can now be passed to voters in minutes. Because of this, the odds are greater that something major will happen to change the course of the election. Imagine a news cycle as a spin of a roulette wheel. Each time something happens, the effects have some random probability of affecting each side to a greater or lesser extent, but the odds of something truly game-changing are very small. But with the dramatic shortening of the news cycle, that means in effect we are spinning the roulette wheel more and more often, and so as a matter of probability the odds of a game-changing event get higher and higher.
  
  
• Also a product of the increased information availability is the transition from a policy-centered campaign to one where personal impressions seem as important as policies. Since politicians have an incentive to keep their policies consistent, a policy-oriented campaign has a low likelihood of major changes in the last few weeks of a campaign. After the conventions, with the party's platform set, the policies were out in the open and voters made their decisions. But now with such a focus not only on the candidates' personal histories, but also that of their spouses and even people they kind of, sort of, not really know, there are more opportunities for some individual thing to blow up. News organizations are always on the lookout for the exciting story to give their audience, and no doubt every candidate has a skeleton or two (as every person surely does), so there's some chance that any of these could catch on and change the outcome. To extend the roulette analogy, when there are more bets on the table, the odds of a big win get higher.
  
  
• Lastly, Harwood's assertion that the polarization of politics has grown larger is based not on established fact but on debatable conventional wisdom. Morris Fiorina, one of the leading scholars of American politics, finds that the idea of a red and blue America is a myth and that Americans today are no more polarized than their ancestors. There are equally laudable opponents of that idea, and I have a number of problems with Fiorina's book myself, but it's certainly not settled fact that the conventional wisdom of polarization leads directly to more stable presidential preferences. (Full disclosure: I'm currently teaching the former book, and I've studied under the author of the latter.) One trend that may auger for the opposite effect--that voters are in fact less stable than they were in years past--is the possibility of a decline in the strength of partisanship over time. Simply put, voters have become more independent since the 1940s and 50s, and as such their votes are less predictable than previously. Other studies have contested this result (and the above-mentioned Bartels is a prime proponent of the resurgent-parties theory; Fiorina notes, however, that the evidence for this is weak at best), but my point is not that a big change will occur, just that it still might, and evidence of uncertainty in the purported effects is thus evidence against certainty about the outcome.
  

So I still maintain that this race is far from over. Maybe this is just the paranoid ranting of a formerly-long-suffering Red Sox fan combined with a lifelong Democrat (I was the only kid on my schoolbus who admitted liking Dukakis), but this race is too close still to start measuring the drapes in the White House.

That said, it certainly looks like McCain is trying hard not to win. I mentioned yesterday that he seemed to be taking up a half-baked version of my advice. Now it seems that he's abandoned that plan and is working on his third economic proposal in the past week. With each new scheme his credibility seems to be going down, and as such Republican strategists' long-delayed realization that it's about the economy after all may well be too little, too late.

So my prediction? I'd have to agree with the traders on Intrade, and put the odds at about 3-1 in Obama's favor, but no better than that.

Sunday update

No numbers today--there were almost no new polls released except for the tracking polls, which moved around a little but in no particular direction, so there's not really anything to see. Tomorrow should be better...

One quick point I want to make, though. I've been hearing a lot of "It's over!" talk lately, from Joe Scarborough on Colbert Wednesday to a friend of a friend in a bar last night, and I have to disagree. McCain's in a deep hole right now, but he's not out of it. We've got more than three weeks left, and though Obama has something in the 6-7 point range for a lead in the popular vote polls, it's by no means decisive. We still don't know how McCain's ground game will be, or Obama's. We don't know if there's a Bradley-Wilder effect going on. We don't know how many of the new registrants will actually turn out. We don't know a lot of things going on right now, and so even though we aggregate lots of polls here on this site and on the poll trackers themselves, I still would give the total numbers a +/- 4-5% margin of error when compared to what the final vote returns will be. That means that in reality, Obama may well only be up by a point or two, and that's a number that can easily be overcome in three weeks, especially if Obama's supporters declare victory, stop working, and go home before this thing is actually over.

Some weekend reading:

• Sometimes I hate being right: McCain is coming out with a new economic plan. That said, from the looks of things there don't seem to be any real ideas beyond tax cuts (which is what the Republicans always propose, whether conditions are good or bad) and the already-panned mortgage-buyout plan. So I don't expect this to have much effect, if any. Are they trying to lose?
  
  
• It seems the whole McCain-Palin angry-mob meme is really taking hold. This weekend's New York Times sees Frank Rich taking on the topic with his typical vigor--which is to say, disturbingly-credible hyperbole. Worth a read, even if it's a bit spine-chilling. (I also think Maureen Dowd wrote something on the topic, but I can't tell, since in her oft-annoying wittiness decided to publish it in Latin. And sadly, I'm not joking.)
  
  
• For those who are still thinking worst-case scenario: Canada wants to help us out.
  
  
• The New Yorker's verbose presidential endorsement is an exhaustive and scathing indictment of the Bush years and a compelling manifesto of why Obama is the choice to move beyond them. If you have any undecided friends with extraordinarily long attention spans, send them that way.
  

Plan on a late numbers update tomorrow, but I may try and also get in a bit of analysis earlier in the day.

Today's Numbers - October 11

Almost no change from yesterday, so you didn't miss much from the delayed analysis:









Is this Ayers crap all McCain's got left? Really? Because right now Obama is running out the clock on him. For the past four or five days, we've seen very little actually movement in the campaigns, and as a consequence Obama's position has grown stronger and stronger. It's clear now that the Ayers accusations aren't having any effect, and may actually be backfiring--the coverage of the "angry mobs" at McCain events can't help but hurt McCain with independent voters. And the McCain camp seems to have no strategy for handling what they've started. Reports yesterday of McCain being booed at his own events for simply calling Obama "decent" show a campaign that is completely out of control at this point.

If this keeps up for the next 24 days, we've got the potential for a blowout. If that happens, plan on a plethora of books and articles debating how and when exactly the Republican Party imploded. Was it Bush? Was it Palin? Was it the primaries? All I know is, it's a refreshing change of pace to see the Republicans pulling the circular firing squad act instead of the Democrats for once.

Today's Numbers - October 10

ALCS game one = analysis tomorrow.

Today's Numbers - October 9

No big changes today:









The EV trackers are still catching up with the national numbers, but the real news today isn't from the political world. It's the stock market. With the Dow having fallen 39% since this time last year (with almost half of that loss in the past month), we're rapidly approaching the line between recession and depression. Nobody's quite sure how the losses in the stock market are going to translate into the numbers that matter more acutely for the middle- and working-classes--unemployment and inflation, in particular--but it's not going to be pretty.

The conventional wisdom would have this help Obama, but two things might change that. First, if McCain manages to diminish Obama's advantage on the economy (see yesterday's post), things could tighten up real quick. And second, if the crisis gets so bad that it starts to affect basic day-to-day life in the country (food or gas shortages, massive layoffs, catastrophic bank failures, utilities companies collapsing), the experience issue may come back into play. This is where McCain really hurt himself with the Palin pick; with her lack of qualifications firmly established, the Obama/Biden ticket is suddenly looking like the "safe" choice. But if McCain can regain that mantle, voters might flock to him if the situation gets dire enough.

As I've said before, I think the odds of McCain pulling this off are pretty slim, but this race isn't over yet. The stock market's hemorrhaging is the kind of once-in-a-generation crisis that has the potential to be a game-changer, and nobody can predict with certainty what effects that will have on the election--there isn't really any precedent for it.

For reaction to the numbers today, see any of the past few days' posts. There's nothing really notable here, and (to repeat myself ad nauseum) a day when nothing happens is a good day for Obama.

Today's Numbers - October 8

First signs of a post-debate reaction:









While we obviously can't see any effect of the debate in the polls (since there are no state or national polls yet which were conducted post-debate), the markets tell a big story. Our win percentage margin increased by more than 10%, and that's including the (non-market) FiveThirtyEight prediction; the three markets in our index had an even greater leap.

The story here is simple: while we could argue who won the debate (most say Obama, as would I, but it wasn't a blowout), there's no arguing that McCain didn't do anything to change the nature of the race. Since Obama's ahead, that means McCain lost by default. And with only one debate left, and no reason to think McCain will excel or Obama will fall flat, the markets don't see much chance for a McCain comeback.

As James Carville said on CNN last night, "You can call the dogs in, wet the fire, and leave the house. The hunt's over."

I'm not quite in agreement with the Cajun yet (he also predicted race riots in the event of an Obama loss--typical Carville), but it doesn't look good for McCain. If I was his campaign manager, I would not be sending him off to these everyday campaign stops in swing states, sometimes two or three a day. None of these events ever moves more than a handful of votes at a time, and he needs to move a truck's worth in less than four weeks. Given the set of issues at play now, McCain needs one of two things to happen: either a terrorist attack occurs, forcing attention back to his strongest point, or he somehow gains an advantage over Obama on the economy. He has no control over the first (hopefully), but he can do something about the second. McCain and his advisers need to pull off the trail immediately, gather some heavyweight advisors in a conference room with McCain for 3-4 days of highly secretive meetings (which will generate great rumor-press and attract attention from Obama), and come up with a comprehensive economic plan. It needs to be a grand plan, on the scale of the New Deal in ambition if not in ideology, and then they need to write it up as a law and submit it in the Senate on October 15, the day of the third debate. After getting a day's media coverage pre-debate with his surprise bill, he then introduces it straight to the public in the debate (which is on the economy), and responds to each question with an element of his plan.

(The questions on the economy are quite predictable at this point: blah blah jobs, blah stock market blah, blah mortgages blah blah blah, deficits blah... There's no excuse for either candidate not to have a solid response given any one of these keywords.)

He then needs to spend the last two weeks berating (a) the Senate Democrats for not taking up the bill and (b) Obama specifically for not having a better alternative. Obama knows that solving the economic crisis isn't that easy, and thus wouldn't try and compete to have something done by election day, at which point McCain can accuse Obama of being both clueless and unable to act. There's no way this plan needs to actually be good--there's an exactly 0% chance it would ever pass before the election--it just needs to sound impressive to voters and have enough believable elements for the talking heads to push it.

I'm not saying that I think this would work necessarily, but I think that it's the best shot. Voters care about the economy right now, and they believe Obama will be the best to fix it. McCain needs to do something to convince voters both that he's competent on the issue and that he has ideas which are specific, comprehensive, and (crucially) different from Bush's. By overloading the media and voters with specifics, McCain can make that case, and steal some thunder from Obama and make him look small in comparison. I think there are still a lot of undecideds and "soft" Obama supporters who like McCain personally and perhaps Obama not so much (for issues of race, perceived elitism, and the like), and McCain needs to give them a way to justify voting FOR him and not just AGAINST Obama. A comprehensive economic plan would give them that cover, and combined with a strong debate performance, there's a chance it might be enough.

That's about the only chance I can foresee. I think even a bin Laden video is only worth a point, maybe a point and a half now, given the Republicans' failure to catch him and Obama's hawkishness on Pakistan. An actual attack somewhere (small- to moderately-sized; a large one would postpone the election I'm sure) would get him somewhere around 3-5, which Obama can still afford (though it would be tight). Wiping out the Obama advantage on the economy is pretty much the only chance McCain has.

(And if he does just this and wins, please don't come after me with torches. Really, I didn't think anyone actually read this! And angry mobs kind of freak me out.)

Today's Numbers - October 7

Obama's electoral votes keep rising, the popular vote stabilizes, and the markets creep up a hair:









Not much more to say than that. Obama's rapidly approaching his best-cast scenario--with a credible opponent and a fairly thin resume, a true blowout has always been out of the question--but he's got to hope that he's not peaking too early. I'm hesitant to make any more predictions with the 2nd of 3 debates in 60 minutes, so I'll save the analysis for tomorrow.

Today's Numbers - October 6

Sox win in 4, so the next few days should be on time, but no guarantee after that.









To more clearly state yesterday's discussion of which numbers are "forward" and "backward" looking, we can think of it as a past-present-future comparison. Electoral vote calculators look at the past few weeks' state polls for their numbers, so they lag behind the other numbers. Popular vote trackers look at the past few days' national polls, so they're better for tracking changes in the election as they happen. And markets look forward--traders bet on future outcomes, so they use their collective knowledge to forecast what will happen over the next few days or weeks, much as stock traders do.

You might ask, then, why we don't just look at markets for our snapshots. The simple answer is that while they may be best at predicting future movement, they're worst when it comes to stability. After that come popular vote trackers, which are highly subject to the timing of specific polls' release. (There are five national tracking polls--Rasmussen, Gallup, DailyKos/Research2000, Diageo/Hotline, and GWU/Battleground--which provide updated results each day, and then another 8 or 10 firms which release national polls on a somewhat regular basis, generally every week or two. The house effects of each poll vary widely, so the national numbers can change arbitrarily if a particular day sees a release of a few strongly Republican- or Democratic-leaning polls.) Finally, since they make use of the widest set of data collected over a longer period of time, electoral vote counters are often the most stable of our indexes.

We can see all these dynamics in today's numbers. Both electoral vote counts continued to trend in Obama's favor, reflecting the increase in his popular vote numbers over the past two weeks. Meanwhile, the popular vote margin went down more than half a point, owing mainly to three new polls (from CBS, Democracy Corps and Zogby--see Pollster for details) which have tended to be McCain-friendly throughout the election cycle. Still, the overall trend is fairly stable. And in the markets, Obama's win percentage has picked back up to where it was two days ago, indicating most likely that the McCain camp's new strategy is not expected to be able to counteract the continuing turmoil in the economy.

(And yes, I know that FiveThirtyEight and IEM throw some wrenches in the above generalizations, since the former is none of the above--they use simulations based on national and state polls--and is included in three of four indexes, and the latter is a market being counted in the national tracker. The fundamental assumption of this model, however, is that none of the methods will have a large effect on the total index on its own, and none of the changes today is driven primarily by either one of them.)

So where is the race, in plain English? Right now, Obama is in a very good position to win, ahead by a significant margin in the popular vote and with a comparable advantage in the Electoral College. As such, if the landscape doesn't change before election day, Obama will win. But that's a big "if". Still four weeks to go, and with the McCain camp getting more aggressive by the day, the economy still plunging with no end in sight, and who knows what new event on the horizon, McCain is still very much in this thing, even if the numbers don't like good. I'd take 2-1 odds on him at this point, which is about what the markets predict as well.

Today's Numbers - October 5

Some new additions:









A quick update today--Sox in the playoffs, etc. First off, two changes in the inner workings of the site. The first is that I've added two new electoral maps to the numbers--Chuck Todd's and Karl Rove's. The second is that I've decided to add a weighting factor to each tracker, based only on the age of the numbers. Specifically, each day the tracker's figures are discounted by 10% (meaning that they influence the overall averages 10% less), and stop factoring into the averages at all by day 10. The purpose of this is to make our averages respond better to day-to-day changes in the race by lessening the impact of numbers that are a week old or more (Election Projection, I'm looking at you...).

Still don't know what to do about CNN, though; I can't tell from their site when the numbers were last updated except when they change. If anyone knows where to find this info, I'd love to hear it.

Regarding today's numbers, we see an interesting dynamic at play. Poll trackers look backward--they mainly reflect voters' opinions over the last week or two--while markets look forward. As such, the EV and PV projections all continue to rise, reflecting Obama's continued strength in the national polls and its trickle-down effects on state polls. The markets, meanwhile, have retreated a bit today. I would suspect this is an effect of McCain's decision to go negative; with Obama's numbers about the highest that could reasonably be expected at this point in the race given the poll numbers, any new uncertainty would be likely to tend toward the middle. Even if the odds of the strategy working are low, it seems the markets expect those odds to be higher than the odds of it backfiring. I'm not sure I agree with this, but maybe I'm not at my cynical best today.

We should see within the next week what the impact of this strategy will be. Which begs the question--why didn't McCain hold off another week or two? Right now, there's plenty of time for these accusations to be refuted and forgotten. Later in the race, you might get the peak of a "dead-cat bounce" (a short-lived and small bounce) to hit exactly on election day. There's no way that will happen now. My guess is that either, as Chuck Todd said on Meet the Press today, the McCain camp is all tactics and no strategy right now, or else they've got something big planned for late October. I don't have any clue what that would be, nor do I really give them that kind of credit, but it is a bit worrying.

Today's Numbers - October 4

poll3.com: Now with charts!









Exactly one month to go til election day. Sorry again about the lack of updates yesterday; if it's any consolation, there wasn't much new to report. As we see today, the race is pretty stable at this point. There was a bit of minor variation in the electoral vote counts, but this is just random noise, not any meaningful change as far as I can tell. The popular vote percentage dropped by half a point, but this change is entirely due to a correction in the Iowa markets. They'd always been a bit of an outlier (coming in at Obama + 8.7 for the last two updates), more than would be explained by using proportion of two-party vote instead of total vote--as I explained in an earlier post--and now it looks like they've fallen in line with the other markets.

In terms of predicting the outcome of the election, this is good news for Obama. As I've said in other posts, every day when nothing happens is a good day for Obama, and the win percentage predictions have ticked upwards to reflect that. While 4.4% may seem like a big jump over two days if nothing happened in the polls, that figure reflects two missed opportunities for McCain: the VP debate and the bailout plan passing. As far as we can tell, neither event did anything to help McCain and might have hurt him, so the win percentage improvement reflects the dwindling number of chances McCain has to avert defeat.

Speaking of McCain's strategy, as I and many others predicted, he's getting desperate and starting to go highly negative. The topic du jour is William Ayers, a former member of the Weathermen--a violent radical group in the 1960s--turned professor of education. Obama lives in the same neighborhood as Ayers in Chicago, and the two have crossed paths in both their personal and professional lives, but as the NY Times reported yesterday, the two have never appeared to be close. Despite this, the McCain camp is trying to paint Obama as guilty by association.

I really don't think there's anything to see here, but it's seems this is all McCain's got left at this point. He's played the policy card and the personality card, and seems to have lost on both, and he played the VP card but that's either faded or backfired. Driving Obama's negatives up may help McCain with his base, but I don't think Obama's supporters will pay any attention, and neither will most independents. Maybe there was an upside to the long primary season after all for Obama--after getting all the dirty laundry (Ayers, Rezko, Wright) out in the open early, it now feels like old news. Still, Politico now reports that McCain will be trotting out Rezko and maybe even Wright as well as Ayers. I don't think it'll work--the public doesn't seem to expect absolute perfection from Obama anymore--but short of dumping campaign manager Rick Davis, I don't see much else he can do. Palin held her own in the debate enough so that there would be nothing to gain by replacing her, and it's highly unlikely that McCain will have a game-changing debate performance this late in the process. And there's no single stand-out thing that Davis has done to screw things up, and not much a replacement could do differently at this point. The fundamentals are against them, and it's late in the game, so the best they can do is to go deeply negative and hope for a miracle.

On completely different topics:

Today Slate released a Poll-tracking app for the iPhone. It relies on data from Pollster, which is probably the most highly-respected poll tracker out there, so it may be worth giving a try.

And finally, keep an eye out for guest columns in the next few days or weeks. I'm hoping to bring on a couple others from the political science world to contribute their perspectives on the presidential election and on elections in general. Stay tuned.

Correction

Working on today's post, I realized that some of the margins were off in the polls for 10/2--forgot to update them in the rush to get to the debate. The revised numbers are online now, and today's numbers are imminent.

Calling in Sox...

Sorry for no numbers tonight--Red Sox game ran late, and I'm a man with priorities... Full update tomorrow.

Today's Numbers - October 2

No time for analysis--I've got a debate to catch.

Fight! Fight! Fight!

There's an interesting dustup going on between FiveThirtyEight and Real Clear Politics today. This morning, Nate at FiveThirtyEight published a post blasting RCP for the arbitrary inclusion and exclusion of polls from their averages. Specifically, he accuses RCP of excluding a number of credible polls and including a number of non-credible polls in order to coax the average in the Republicans' favor.

Background: RCP does not disclose much about its methodology (one of Nate's complaints), but from appearances their figures are the unweighted average of a handful of recent polls. They do not spell out what the criteria are for inclusion or exclusion from that average. FiveThirtyEight, meanwhile, has an exhaustive FAQ detailing the ins and outs of their methodology. (If anything, it may be too exhaustive--I could hardly make it from beginning to end in one sitting, and reading political science stuff is what I do for a living.) With regard to inclusion or exclusion, all scientific polls (ie, with random sampling--not the instant polls you might find on some websites that only include people who visit the site) are included in the model, subject to a weighting procedure that adjusts for the size of the sample, the recentness of the poll, and the pollster's previous accuracy in predicting election outcomes.

RCP hasn't responded on their site, but evidently contacted Nate directly. In a follow-up post, Nate mentions that he got a call from John McIntyre, founder of RCP, and that it was "contentious". He walks back a bit from allegations of bias, but reaffirms his complaint that RCPs methods are too subjective. In other words (mine), they may not be screwing up to fit an agenda, but they're still screwing up.

My take on this? It's inexcusable for a poll aggregator not to publish its methods. Pollster does it. Electoral-vote.com does it. The New York Times does it. And that's just the first three I checked. We expect pollsters to give us information on how they reach their numbers (sample sizes, margins of error, whether the sample is likely voters, registered voters, all adults, etc.), so the people who compile the polls should hold themselves to the same standards. And in the absence of this information, then Nate is perfectly right--RCP has opened itself up to charges of subjectivity and bias, because in the absence of published standards it's not unreasonable to assume that none exist.

This is especially perplexing for a site like RCP, which does not use complicated methodology at all. Whereas FiveThirtyEight's FAQ runs many pages in length, RCP would only need to put out two or three sentences. Something like: "Our average number is the unweighted mean of all the polls included in our database. To be included in the database, a poll must have a minimum sample size of XXXX, be less than XX days old, be gathered by a random sampling method, and be conducted by a nonpartisan organization." In two lines, RCP would be able to defend itself against complaints such as those raised today. That's not to say we couldn't critique their methodology (and I intend to do that at a later date), but having and publishing such simple standards would insulate them from charges that they are trying to rig the numbers behind the scenes to produce a specific outcome. If they were trying to rig the numbers, we'd be able to see it clearly in the standards they applied, and then we'd be able to evaluate their methods appropriately. As it stands, they're asking us to simply trust that they're unbiased without giving us much to back that claim up.

In a media environment where being accused of bias is often enough to severely damage an organization, I cannot fathom why RCP hasn't taken such a simple step to protect itself. If they're truly aiming for objectivity, publishing their standards is a simple and obvious step, so the fact that they don't makes it very hard for me to give the benefit of the doubt.

Today's Numbers - October 1

Up, up, and away...



The good news continues for Obama. There is one new set of polls out this morning that's got everyone talking on both the right and left. Quinnipiac released pre- and post-debate polls in FL, OH, and PA. The numbers: Obama +8 in Ohio, Obama +8 in Florida, and Obama +15 in Pennsylvania. If these numbers are accurate, McCain would get a better return putting his campaign funds in some Lehman Brothers stock. But I doubt it... Quinnipiac is a reputable pollster and there are no obvious flaws in the methodology of these polls, but most likely these are outliers. Still, at the risk of seeming to worship at the temple of Nate Silver again, there was as he put it an "oh sh*t set of state polls" [his emphasis & censoring] that was bound to come out of the change in the national numbers. This looks to be it. But I would highly recommend waiting for some confirmation from other polls before beginning the celebrations.

Other than that, everything to report is a repeat from the previous days. National tracking numbers are still strong (though Gallup seems to be tightening up), the state polls are beginning to finally show the movement that we've been seeing in national numbers, and the predictors keep moving toward Obama in the absence of major developments as it gets closer to November. And the NYT tracker got updated today, so they're on my good side, but CNN is still on notice.

That's all for today. Be sure to buy some popcorn and/or antacids for tomorrow night's VP debate--it's going to be interesting, whatever happens.