Tedger.com

Whew, can it come any sooner? November 4th is only 6 2 long days away from us now. Sen. Obama is up in the polls, at his highest point ever, and it’s looking increasingly like he’s going to clinch the White House before polls even close in the central time zone (the networks might not call it till later, they don’t want to repeat 2000 or loose viewers). Hell, it’s a long shot but it’s possible we could see Democrats winning 60 seats in the Senate. In typical Democratic fashion however I hear a lot of people very wary of any positive news for Obama. This includes the Senator himself saying in a recent rally “But for those of you who are getting a little cocky I have two words for you: New Hampshire.”¹ That’s not the analogy I hear most frequently though, it’s an analogy to the crushing defeat of John Kerry in 2004.

The polling situation now isn’t anything like either of those situations however. First of all, New Hampshire wasn’t heavily polled. There were only eight days between the Iowa Caucuses and the New Hampshire primary. Nearly everyone had their eyes on the Hawkeye state leading up to the caucuses and pollsters had little time in the field afterwards to gauge the feeling of New Hampshire voters. With less polling you are far more likely to run into statistical problems with your data. In addition to this , and I think more importantly, New Hampshire is a small state, this makes the state easier to poll but makes it far more likely that it will swing in a manner that is fast enough to be missed by polling. This is exactly what happened in New Hampshire.

What about the Kerry analogy? Well, there are two issues in the Kerry analogy. The first is that people were looking at exit polling. Exit polling has issues that over the phone polling doesn’t have. First of all you are in face to face contact with the pollster. This leads to a comfort level issue, something that results in extreme response bias. Younger, enthusiastic, and individuals with strong party affiliations are far more likely to speak to exit pollsters, also, in the case of 2004, Democrats. The other issue here is comparing apples and oranges. What we have right now are pre-election polls, not exit polls. So lets look at some trends.

Hop on over to this link. Here you see some side by side polling averages for the presidential campaign courtesy of Electoral-vote.com². Specifically lets look at the second and fourth graphs. The second graph is the number of electoral votes each candidate has by averaging polls each day ignoring states that are within 5 points. These states are far more reliable come election day and as far as I know all of these states went for their respective candidates in 2004. The second graph is all for this year. You’ll notice that Obama is above 270 electoral votes ignoring “barely” or what other places call “leaning” states. In other words by these polling averages he could loose all the close states (for EV.com these are Nevada, Florida, North Carolina, Missouri, Indiana, Georgia, Montana, and North Dakota³). Now look at the 2004 graph, the fourth one, notice how Kerry and Bush were no where near that 270 mark? The election in 2004 is significantly closer than it is today.

There is something that is even more significant about these graphs though. Note how stable the 2008 graph is in comparison to 2004. This isn’t because it has fewer data-points, it’s because the race was so close that pollsters had a very hard time gauging the electorate, with such a closely divided electorate the electoral vote will fluctuate wildly because a very small change can result in a swing of the entire election. The Obama numbers are steady, and pretty stable, and they’re at their highest point of the election, Kerry peaked in late July, early August, and never firmly retook the lead. With the exception of the Palin/Republican National Convention “bubble” McCain hasn’t been ahead since before the end of the Democratic primaries.

This argumentation only takes us so far and doesn’t look at what the polls aren’t telling us. One of the most frequently cited polling anomalies this cycle has been “The Bradley Effect.” The Bradley Effect is named for Tom Bradley a highly popular mayor of Las Angeles that lost the 1982 California gubernatorial race. Bradley significantly underperformed polling that had him up by a reasonably safe margin before election day. The explanation is that some voters will tell a pollster that they are willing to vote for an African-American candidate but wont follow through with it on election day. The existence of the effect at all has come into question and many who believe that it existed at one time don’t believe it exists anymore. This is supported by Barack Obama’s performance in the primaries where he actually outperformed polls by about 2 points.

There doesn’t appear to be any sort of Bradley effect any longer, Barack Obama in the primaries, Harold Ford in the Tennessee Senate race in 2006 (which was highly racially charged, if you wanted to find a Bradley effect you would have found it there), Michael Steele in the Maryland Senate race in 2006, all cases where no Bradley Effect were seen. The idea that it exists any longer is outlandish, and the fact that it doesn’t speaks volumes for the way we’ve grown as a country.

The Bradley Effect would skew results against Obama but there are factors that actually favor him that won’t appear in polls. One getting major coverage is cell phone polling. Why does it matter? Cell phones are under-polled, it’s hard to get the numbers and cell phone users are far more likely to screen their calls than landline users. Cell phone users lacking landlines and/or mostly using cell phones are also overwhelmingly young. When you effectively have a systematic removal of one population group from the polls you only have an issue if that population group has a particular lean that is different than the population as a whole, which in this case you do. Young voters favor Obama by 2:1 in some polling and up to 3:1 in other polling. This has remained consistent throughout the campaign and has the possibility of having a huge impact. The question is how much of an impact. This hinges on two things, how much the under polling of cell phones has skewed polling, and how many young voters turn out on election day.

The skewing of polls as a result of lacking cell phone numbers is easy to see and according to Five Thirty Eight could be almost a 3 point bump for Obama. If you’re a reader of Five Thirty Eight you’ll be familiar with the mentioning of “house effects.” House effects are the leans that polling firms have shown over the course of the election. This isn’t due to attempting to cook the books (Fox News actually has a slight Democratic lean) it’s simply due to how the polling is conducted and how the questions are asked. Five Thirty Eight accounts for these effects so the “cell phone effect” is being controlled for as if it’s a systematic problem with the pollster when in fact it could be that the pollsters that call cell phones could be ~2.8% closer to the truth.

The “cell phone effect” is only an issue if young voters turn out. Young voter turn out is something like Godot, people keep waiting for it to happen, this year is always the year, but never seems to materialize. This year however if it’s going to be any year, appears to be the year. We’ll have to wait for Tuesday to know for sure but the GOTV⁴ efforts this year have been phenomenal at ISU and I have no reason to think it’s unique. Added to real empirical evidence of a youth vote turnout in the primaries I think that this might be the year.

The less covered advantage that wont be captured in the polls is Obama’s massive ground operations. Senator Obama has amassed the largest campaign apparatus in American history. He has a paid campaign staff edge of 3:1 in some battleground states which is matched by a similar edge in field offices⁵. This huge campaign apparatus is being predicted to turn out Democratic voters in record numbers in a year that Republican vote is probably going to be suppressed by the seeming unstoppable juggernaut of a campaign that bought up 30 minutes on every network to air an informercial, a move costing $4 Million, 4.5% of McCain’s total general election budget. Campaign organization has been credited by as much as a possible 3 point shift in some places, but it’s something we wont know till we see it on election night.

Predictions

I made these predictions on the air Wednesday and they’re based on what I’ve outlined here. I expect to see an Obama win, and a big one, something that might look a little like this:

Now that map is not what I think is most likely, just what I think might happen. More likely is that Montana and North Dakota will remain in the McCain camp, but barely. Montana, North Dakota, and Georgia are the only three states that the Obama campaign set out to compete in that they aren’t currently winning (except Alaska which was removed from the list when Gov. Palin was added to the GOP ticket). Georgia and North Dakota both have large numbers of young voters and Georgia has a large African-American vote. Montana is probably the least likely of the three but with a highly popular Democratic Governor and an independent streak I think there is potential (not to mention close poll numbers). These factors I think create a potential upset. I also think that the recent shift of Indiana and Missouri in the polls is going to be outweighed by the factors I’ve outlined above. I think it’s near guaranteed that Obama is going to take the states I’ve marked above except Montana, North Dakota, and Georgia, and highly likely that he’ll take Georgia; I’ll admit I’m going out on a limb with North Dakota, and even more so with Montana.

Additionally I think that Obama might carry some heavy coattails leading to a run-off election in Georgia’s Senate race, possibly taking the Democrats to 60 seats in the Senate.

We’ll see if I’m right come Tuesday.

Editorial Note: I wrote this over the last few days and I really just want to get it up so I haven’t heavily proofread it, apologies to spelling and grammar nazis.