Monthly Archives: October 2008

The State of the Race 10 Days Out

With 10 days left in the race, the political science forecast models from August are looking golden. Any progress that John McCain was making at the national level has slowed in recent days, according to the daily national tracking polls. As of yesterday, the average of the five daily tracking polls I’ve been following puts Obama at 49.6%, and McCain at 44%, for an almost 6% lead for Obama which is not quite 1% higher than it was almost a week ago. There are two important points to take home when looking at these numbers. First, Obama’s average level of support is near 50%, and that support seems solid. Second, the number of undecideds in these polls hovers at just above 6%.  Keep in mind that despite what you may read in blogs or hear from the media, races do NOT necessarily tighten in the last 10 days – they are almost as likely to break wide open as the front-runner pulls away.

 At the state level, things are looking a bit better for McCain, but not by much. In theory, all McCain needs to do is to hold all of Bush’s states from 2004 except for Iowa and New Mexico, both of which seem out of reach for McCain this year. In 2004 Bush won 286 electoral college votes to Kerry’s  252. If you substract Iowa and New Mexico, McCain wins the election with 274 electoral college votes. The problem is that he isi fighting an uphill battle to retain several remaining Bush states, including Colorado, Virginia, Ohio, Florida and North Carolina.   Now conceivably he could lose Colorado but still win by picking up Pennsylvania, which Kerry won narrowly in 2004. However, although McCain continues to make very incremental progress in the key battleground states of Florida and Ohio, he is down by 8-10% in Pennsylvania which is only a slight improvement from where he was a week ago.  Despite erroneous media reports, he has not pulled out of Colorado (although he has scaled back his paid media there) and the state remains in reach. However, his numbers are quite bad in Virginia, and he is struggling to hold onto Missouri and Indiana, all states Bush won in 2004.  At this point, McCain really has two roads to victory: hold onto all the 2004 Bush states except Iowa, New Mexico and Colorado, but steal Pennsylvania, or else retain Colorado.  All this assumes he wins in Virginia which is looking increasingly doubtful. (A third path is even more complicated: if McCain loses Colorado, Iowa and New Mexico, but picks up New Hampshire, and holds the rest of the Bush states, the Electoral College race would be tied at 269!)  

Now, is there anything to suggest that these polls are missing underlying trends toward McCain, or overstating Obama’s lead?  Were I to rely on a single poll with a built-in bias based on distorted partisan weighting (see the Daily Kos tracking poll!) I might hedge my bets.  But, in theory, taking an average of the five tracking polls should ameliorate the house effect of any one poll.  It is true that the standard deviation between polls (that is, how far each poll is from the average of all the polls) is greater this year than it was in 2004 at a similar time.  This means that there is greater variability across polls.  The reason, I think, is that pollsters are not comfortable with their partisan weighting of samples, or their projections of overall turnout, and thus each pollster is “guesstimating” using slightly different projections.  This reflects their uncertainty regarding reports of higher registration numbers, the impact of a potential “Bradley” effect, the potential cell phone bias, and the Palin effect.  All these factors have contributed to an unusually volatile polling year.  Nonetheless, my read on the election at this point (and I admit to relying heavily on the political science forecast models from August as comfort) is that Obama continues to be safely in the lead in the popular vote, and that he has a strong if not invulnerable position in the Electoral College.  To be sure, McCain has found a stump message that is working – one that owes much to Hillary Clinton’s effective campaign strategy beginning in March during the primaries.  But Obama is not coasting to the finish line; he has stepped up his direct attacks on McCain to try to counter that message, and is flooding key battleground states, particularly Florida, with overwhelming campaign advertising.  In short, the opposing campaign frames are largely negating one another, leaving the race to be determined by the fundamentals – the economy and change – that have favored the Democratic candidate from day 1.  In looking at the polls, then, the key message you should latch onto is that every one shows Obama in the lead.


The Powell Endorsement: Why No Impact?

It has been three days since the race-changing, mind-altering shocking unprecedented Powell endorsement of Obama’s candidacy, and it is probably worth looking at the extent to which the electoral campaign was turned upside down in its aftermath.  Recall the immediate reaction to Powell’s endorsement:

Mark Halperin called it “one of the most symbolically important moments of the general election,” and said it contributed to Obama’s campaign momentum.

Chris Matthews described it as “Colin Powell, right in the kisser. Barack Obama gets the endorsement of the  year.”

Andrea Mitchell said, “I think this is a very powerful political statement.”

Mike Murphy of Swampland (a Time blog) opined, “…Colin Powell’s endorsement of Obama today is a real sledgehammer blow to the already staggering McCain campaign.”

Another blogger, reflecting the sentiments found on many blogging sites, called Powell’s endorsement, “the most important, most profound, more powerfully argued 7 minutes of this campaign.” (See Powell).

Oh, and Middlebury College Political Scientist Matt Dickinson said, “Ignore the media hype. It does nothing to alter the fundamentals driving this race.”

Who is right – Dickinson, or the glitterati of the punditocracy? Let’s look at the polling data. You be the judge.

I’ll break down the pre and post-Powell endorsement data into two categories: daily tracking polls and more intermittent surveys that get bigger press coverage.  In all cases we want to see whether Obama got the expected boost in support due to Powell’s endorsement. We begin with the more sporadic but more highly publicized media polls.

Poll Last Pre-Powell Poll Most Recent Poll Net Gain for Obama

NBC/Wall Street          49-43  Obama   52-42  Obama                 +4

ABC News                  53-42 Obama    53-43 Obama                 -1

Ipsos McClatchy       48-39 Obama        50-42 Obama                -1

CNN                           53-45 Obama     51-46  Obama               -2

Ok, so Obama appears to have gained, on average, absolutely nothing from the Powell endorsement in these polls, but note that they sample voters over three-four days, and some were released on Monday, so at least one of the days of their most recent polling took place before the Powell endorsement.  And in some cases the pre-Powell poll is from a couple of weeks ago. It might be better, then, to look at the most recent daily trackers, which include more data from the post-Powell endorsement period. Surely they show significant gains for Obama! In each case, I compare the last tracking poll from the day before Powell’s endorsement to yesterday’s daily tracking results.

Gallup (Expanded) 51-44 Obama         52-42 Obama                          +3

Rasmussen  50-45 Obama                    50-46 Obama                          -1

Hotline             49-42   Obama             47-41 Obama                          -1

Battleground 49-45 Obama                  48-47  Obama                         -3

TPP                 47.2-39.8 Obama        46.9-40.0 Obama                    -1.3

Zogby              50-42 Obama              52-42 Obama                          +2

Once again, we see that, on average, Obama has gained absolutely nothing in the daily tracking polls since Powell’s endorsement.  Now, it may be that the full significance of that endorsement has yet to kick in.  But as of now it doesn’t appear to have had any impact whatsoever.  Why not?   The answer says much about how political scientists view elections versus the media perspective.

In theory, endorsements might matter under one of two circumstances: One, the person providing the endorsement has the allegiance and exercises authority over a significant portion of potential voters. Think of the Pope. If he endorsed Obama, that could potentially swing a few Catholic voters. Two, the endorsement reveals previously hidden aspects of the candidate’s resume or qualifications for office in a way that causes voters to reevaluate their comparison of the two candidates.

Neither of these hold in Powell’s case.   He doesn’t speak for any group of people, having long ago burned his bridges among conservative Republicans and still mistrusted by many Democrats who remember his case that Iraq possessed weapons of mass destruction. In any case, most Republicans and Democrats long ago made up their mind in this election. What about the military – surely Powell carries some clout there? In fact, surveys of those in the military suggest overwhelming support for McCain – Powell is not going change that in appreciable numbers. How about independents? As I noted in my earlier post, this election is about the economy, and Powell’s credentials on this point are not the kind that give him any particular weight on this issue. Most independents are looking to see what the candidates can do about the mortgage and banking crisis, jobs and unemployment – not the “Powell doctrine”. Those that are weighting national security issues more heavily are disproportionately for McCain. Nothing Powell can say will likely convince them that Obama is better prepared to make military decisions than McCain.  Indeed, they realize that on national security issues Powell is closer to McCain than Obama, so his endorsement isn’t likely to win over many independents on this issue either.  In short, when we step back from the media hype, and think this through carefully, we shouldn’t expect Powell’s endorsement to matter.  And it hasn’t.

Don’t misunderstand. Powell’s endorsement was an important news story from the perspective of pundits and journalists.  It was news! But from a political science perspective, it was not substantively very important because it doesn’t affect how people decide to vote.  The difference says much about how the two look at campaigns and elections.

At this point, with less than 20 days to go, the dwindling numbers of undecided voters are paying almost no attention to who wins the daily news narrative, and instead are thinking of the race in terms of more fundamental issues that I have harped on again and again. From this perspective, Powell’s endorsement was so much media background chatter, to be stored as one more element in the glut of election-related information that now saturates the airwaves.  It made for interesting water cooler chatter, but not much else.  If you are an Obama supporter, it confirmed your support. If you are a McCain backer, it is easily dismissed. And if you are undecided, Powell’s endorsement will not have much relevance to you.

But didn’t McCain “lose” the media cycle?  Perhaps – it depends on what the displaced storyline was likely to be: Obama’s fundraising advantage?  His overwhelming lead in the polls?  Chuck Todd pointing out once more why the Electoral College map favors Obama?  More importantly, however, voters are not looking at the race in terms of who wins or loses the daily news cycle, and McCain’s fortunes aren’t affected by this by to nearly the degree that journalists think they are.

It is perfectly understandably why pundits, who live day-to-day from news cycle to news cycle, and journalists, who need to file a story every 24-hours, would hype the Powell endorsement. It was news!  But the public operates on a different time scale when it comes to the election.  In the end, we should be pleased that as a collective, voters’ decisions are not skewed by the endorsement of a single individual.  I’ve said it before, but it bears repeating: voters are quite rational when it comes to elections.  Perhaps the most telling comment I read in the post-Powell lovefest was by a media commentator who gushed, “I can’t remember an endorsement in the history of political campaigns that had this much impact.”  Think about that for a moment.   Can you remember ANY political endorsement that had any positive impact in a presidential election?  Any at all?

It would be foolhardy, media hype notwithstanding, to think voters could be moved to vote for a candidate simply because one person told them to do so. Indeed, we ought to be thankful that Powell’s endorsement has had so little impact – that voters are not that fickle or easily swayed.

(My apologies for those of you who remember my earlier screed on this topic when Massachusetts Senator Ted Kennedy endorsed Obama, and Clinton proceeded to crush Obama in the Massachusetts primary.   But the media hype after Powell’s endorsement was easy pickings.)

Is the Race Tightening? Some State Level Results

In my last post I said the tightening of the race at the national level would mean more if it was reflected in state-level polling, particularly in the big-ticket battleground states. We are now seeing the same trend: a slight – and I emphasize slight – movement toward McCain in three of the four biggest battleground states for which polls have come out in the last two days.

In Florida, there have been three recent polls indicating a slight movement toward McCain:


SurveyUSA has McCain up 49-47, a gain of 1% for McCain since the last Survey poll three weeks ago.

PPP has Obama up 1%, 47-46, a gain of 1% for McCain since the last PPP survey 3 weeks ago.

Rasmussen has McCain up 48-47, a gain of 6% for McCain since the last Rasmussen poll a week ago.

As a result, Obama’s lead in the polls in Florida has been narrowed to 2%.


In North Carolina, 3 polls have come up in the last two days showing a net average gain for Obama:


Rasmussen has Obama up 51-48, a gain of 3% for Obama in NC since the last Rasmussen poll a week ago.

Civitas has Obama up 48-45, a gain of 2% for McCain since the last Civitas poll 2 weeks ago.

PPP has Obama up 51-44, a gain of 4% for Obama since their last poll 1 weeks ago. .

As a result, the RCP average has Obama ahead in NC by 2.3%


In Ohio, three polls have come out in the last two days:


Rasmussen has McCain up 49-47, a gain of 4% for McCain since their last poll a week ago.

Suffolk, however, has Obama up 51-42.  They have no other polls in October.

NBC/Mason Dixon has McCain up 46-45. They have no other polls in October.

As of today, the RCP average has Obama ahead in Ohio by 2.8%.


In Pennsylvania., two polls came out in the last two days:


MorningCall (a daily tracking poll) has Obama up 52-42, a gain of 2% for McCain in the last three days.

Susquehenna has Obama up 48-40. They did not poll previously in October.

RCP still gives Obama a commanding lead here of 11%

Interestingly, McCain has devoted a huge amount of campaign time to Pennsylvania, a state that I think frankly he has little chance of winning. Although the tracking poll suggests he might be cutting into Obama’s lead, he still trails by a considerable margin. So why campaign there at all?  All I can think of is that McCain’s internal polling shows that he has an outside shot in this state, and given its size, and the fact that he has few other options, McCain believes it’s worth making a play here. Of course, John Murtha’s recent comments (in his defense, he no longer is calling western Pennsylvanians racist – – instead they are “rednecks”) may keep this state in play for McCain.

It is important, however, to keep these battleground state-level movements in perspective.  All are in states that, with the exception of Pennsylvania, Bush won in 2004, but which RCP currently gives to Obama.  So McCain is fighting to hold onto states that went Red in 2004. Obama can conceivably lose three of these (maybe all four) and still win the election.  So we ought not to make too much of these microtrends – they are interesting so far only in that they confirm the national tracking polls that show this race is tightening.

Not surprisingly, there is no evidence at either the national or state level of a bump in support for Obama due to the Powell endorsement. This is to be expected, of course, and it’s worth a separate post explaining why the media consistently misreads the importance of endorsements generally, and Powell’s endorsement in particular.  I’ll try to get to that in my next post.

No Need to Panic

Because of their typical partisan slant and intolerance for opposing viewpoints, political blogs are wonderful places for like-minded people to engage in a form of internet devotional services.  Obama and McCain supporters can get together in their own blogs, without worrying that their world view and political assumptions might be challenged, except by the occasional troll. Even here they can take collective pleasure by ganging up on the infidel and inflicting an online beating and virtual banishment. On the flip-side however, when political trends appear to be turning against one’s candidate, the online discussion among like-minded people can produce a type of virtual mass hysteria, as they talk themselves into believing that doom is just around the corner.  We are beginning to see the first indications of this on the comments sections in many pro-Obama sites, such as Five Thirty Eight, Huffington Post or the Daily Kos, in response to signs that the presidential race is tightening.  I have waited before commenting on the recent polling data so that the full impact of the last debate on the three-day tracking polls registers, and to be sure that the trend is real (at least as real as can be determined from polls in which the internals are often not released.)

Here’s what the five national tracking polls indicate has happened in the last week

Rasmussen had Obama’s largest lead at 8 points on Oct. 6 – it dropped to a low of 4 on Friday, but is up to 6% today.  One week ago, Obama led by 6%.  So, a drop of 2% (which is within the margin of error) in two weeks but no movement either way in the last week.

Hotline has Obama up by 7% as of yesterday, down from a 10% lead a week ago, so McCain gains 3%.

In Gallup’s expanded poll (the one most favorable to Obama), Obama’s lead has shrunk to 4% (it’s down to 2% in the “traditional” polling of likely voters) from a 10% (53-43) lead on Oct. 13. McCain gains 6%.

Battleground went from 53-40, Obama leading on Oct 13, to 49-45 yesterday a net gain for McCain of 9%.

The one counter-trend is IBD/TIPP.  They had it for Obama by 2% at 44.8-42.7 a week ago. Yesterday his lead had increase to more than 7%, 47.2-39.8.  So McCain loses 5%. However, they also show 13% of voters still undecided, almost twice what the other tracking polls indicate, so it’s hard to know what to make of their results.

Of course, each of these polls has a built-in house effect depending on partisan weighting, treatment of “leaners”,  number of undecideds, etc.  If we average them with other polls, using the RealClearPolitics trend line, we see that Obama has lost about 3% nationally in the last week.  On Oct. 14, Obama led in the RCP poll of polls, 50.2 to 42, a lead of 8%.  Today the lead has shrunk to 5%, 48.9 to 43.9.

So, what should we make of this?  In response to this apparent drop, partisan bloggers who are supporting Obama have advanced a number of largely inaccurate explanations in order to rally the troops. Nate Silver at Five Thirty Eight suggests it is simply the case that McCain is solidifying his partisan base – that “red states” are becoming redder.  This may be part of the story, but it is not the whole story. The Reuters/Zogby poll, for example, shows that Obama’s lead among independents dropped from 16% to 8%.

More generally, Obama’s support according to the RCP average has dropped marginally by about 1.5%, while McCain has gained about 2%, so the change is combination of Obama losing some support and McCain gaining some.

Other Obama supporters argue that races always tighten in the last two weeks.  Again, this is not true.  I looked back at Gallup poll trial heat data, and it shows the presidential race tightening across the last two polling points in 2004, 2000, 1996, 1976, 1968, 1964, 1960 and 1952.  But in 1956, 1972, 1980, ’84, ’88 and 1992 the frontrunner maintained or increased his margin at the end of the campaign.

So, if the tightening numbers are not simply McCain solidifying his base, or the “natural” tightening of the race that occurs every four years, what is the explanation?  First (and not surprisingly to readers) there is no indication of a post-debate “bounce” for McCain – the race began tightening before that debate, although the debate did not halt the trend (at least not as of today). For that matter, neither is there evidence of the bounce predicted for Obama because he “won” the third debate according to the instapolling. I trust I need not elaborate at this point why we shouldn’t expect a debate bounce for either candidate.  Some partisan bloggers, like Silver, argued that Obama received a polling bounce after the second debate, but in fact his upward polling trend preceded that debate and was largely driven, in my view, by the continuing bad economic news.  In any case, there’s no evidence of debate bounce here.

Since I do not have access to the internals of all the polls, I can only speculate here, but I think that McCain’s gain is largely due to three factors:

  1. Most importantly, the spate of bad economic news has, temporarily, not dominated the news cycle to quite the degree that it had in the previous weeks.  The economic crisis is by no means resolved, but we are in something of a holding pattern.  The less frequently we see George Bush on television urging patience, the better for McCain.
  1. McCain has finally found a message, in the terms of an economic policy directed both toward middle-class homeowners and traditional Republican-oriented small business owners, which puts him on the “right” side of the economic crisis.  It is not that he has obtained the high ground here, but he at least has a message, rooted in a policy, to compete with Obama as an agent of change, rather than as part of the problem.
  1. McCain – and Palin – have dropped the “Obama hangs out with terrorists” sub theme on the stump and focused instead on the economy.   If you have watched their stump speeches the last week (and I have) you can see that they have both honed their message and that the audiences’ responses are much more enthusiastic.  There is a renewed vitality to their campaign stops as a result.

So, McCain has clearly made a slight gain – and I emphasize slight – in the last week, (although you wouldn’t know it if you listened to the weekend talk shows!)  Does this mean there is reason for Obama supporters to panic?  No.  Obama is still up in most polls by 4-5%, with only about 6-7% undecided voters left.  Even with his slight drop in support, Obama is still polling close to 50% in most national polls, and the Electoral College map is still heavily tilted toward him. Although we see McCain gaining ground in key battleground states (the latest polls in Florida, North Carolina and West Virginia all show movement toward McCain), these are states that Bush won in 2004 and that McCain still trails in (except for West Virginia). And because voters are polled much less frequently in these states, we can’t be sure how much the national trend reflects movements in the key battleground states where Obama is focusing his air and ground campaign, bolstered by a huge fundraising advantage.

My recommendation is that if you are an Obama supporter, rather than obsess over slight movements in the polls with your fellow Obamanauts, take some time away from the blogosphere today and instead go outside and enjoy the beautiful weather. The fundamentals driving this race have not changed.  It’s still all about the economy and change. And that favors your guy.

By the way, the long awaited Powell endorsement of Obama that came through today will dominate the news cycle for 24 hours but likely will have no impact on the race. I’ve talked about the marginal value of endorsements before, but if you’d like, I can run the data for you again.  They just don’t matter all that much.  Had this been a national security year, it might have carried some weight, but Powell’s economic credentials are about as meaningful as mine.   Ignore the media hype.  It does nothing to alter the fundamentals driving this race.

Take a Deep Breath…

And exhale slowly. Do it again. Ok, everyone calm?  Several of you have emailed me re: the Drudge Report, or other sites, reporting the latest Gallup Poll indicating that the race at the national level is a statistical tie (Gallup has Obama up 49-47, with a 2% margin of error).  “Gallup Shock,” trumpets the Drudge headline. What’s going on?  Has the race tightened that much?  Not according to Nate Silver, who accuses Drudge of “cherry picking” which poll to report in order to reframe the campaign narrative (a tactic Silver knows well, as evidenced by his selective reporting on the debate results.)

For regular readers of this post (and for my students who heard me lecture on this today), this development isn’t necessarily unexpected. Recall my earlier post on Gallup’s continued reliance on samples of registered voters, as opposed to likely voters. Historically, most pollsters move toward samples of likely voters as Election Day draws near, for the simple reason that we want to sample people who are actually going to vote, rather than those who are simply registered to vote.  Gallup was the last of the pollsters doing daily tracking polls to make this switch, but they did so this last week.

Generally speaking, Obama does better among polls based on registered voters, but McCain does better among polls of likely voters. The reason is that typically Republicans turn out in a greater proportion to their registered numbers than do Democrats, so when pollsters switch to likely voters samples, they traditionally increase the weighting of Republicans.  This year, however, the heightened enthusiasm among voters, especially in the Democratic primary, coupled with a spike in voting registration totals, has led some pollsters to wonder whether the Democrats might in fact turn out in much higher numbers during this cycle.  Pollsters are thus left with a dilemma – do they use the traditional procedures that weight Republicans more heavily, or do they adjust their samples to account for what might be a different, more heavily Democratic turnout this year?  Rather than choose between the two options, Gallup decided to present both versions of likely voter samples. Under the traditional turnout model, McCain gets a bump, and the race is essentially tied at the national level.  But under the new likely voter model developed by Gallup, Obama is comfortably ahead by 6%, 51-45.

So, which is more accurate?  One way of finding out is to look at the other national tracking polls. Do they show a movement toward McCain?  Yes and no. Three of them show a slight movement toward McCain, but one (Zogby) shows Obama gaining a point and the other three show no movement.  Based on this, Silver concludes that there is no real trend toward McCain.

In truth, we simply can’t be sure for at least two reasons. First, and most important, no one really knows what the actual partisan distribution of likely voters is; given the unprecedented registration figures it’s not clear whether the traditional sampling strategies still hold in this election cycle. Second, if there is a movement toward McCain, we need more than two days to be sure that it is happening.  But neither can the possibility be dismissed.

Most importantly, to be significant, national trends must translate into gains at the state level, particularly in the battleground states. I don’t see any evidence that this is happening, although because state polls are conducted less frequently, national level trends typically take a couple days to show up at the state level.  So my advice is neither to overreact to these polls nor completely dismiss them. Anyone who tells you they know what is happening likely has a partisan axe to grind.  My instinct says this race hasn’t tightened, but I have absolutely no evidence to support this claim, and thus my assertion is no better than Silver’s or Drudge’s or any other pundit’s.  It is a guess, plain and simple, and should be treated as such.

For now, your best bet is to turn on the Red Sox game – they’ve come down from 7 runs to tie the game in the 8th inning.

Now this is something that should take your breath away.