Silver, Krugman Ignore Data Indicating Romney Victory

Real Clear Politics rephrases the title of Paul Krugman’s latest New York Times piece as “You’re Stupid If You Think It’s Close.” In it, Krugman writes, “Nate Silver (who has lately attracted a remarkable amount of hate – welcome to my world, Nate!) clearly explains, state polling currently points overwhelmingly to an Obama victory.”

The funny thing is that despite being one of Nate’s critics for giving Obama an 87% chance of winning, I certainly have no hate for Nate. In fact I am a huge defender of his against people like Joe Morgan for his statistical work in sports – on which he is amazingly accurate.

The reason he and Krugman do not understand the closeness of the race and potential for a Romney win is that they largely look at the race as a two-linear, two-dimensional affair measured by polling. For those who have worked the ground game on both sides, elections are always three-dimensional. As I’ve summed up in past seminars, “poll results + turnout efforts – actual election day result,” which I’ve covered in great detail in past speeches.

In a linear sense, Krugman and Silver are right. In my mind, there is no question that several million more adults living in America today prefer President Obama to Governor Romney after watching him and Gov. Christie in the wake of Sandy. But that is still offset by the intensity of the Romney voter as indicated by larger rallies, much better early voting numbers than in 2008, and in the end higher participation.

As I did back in 2008 when I began picking an Obama landslide at the beginning of the year, I monitor the following table to get beyond the polls. Each box on this table represents roughly 2 million American adults. The more liberal you are, the further to the left you are on the table (100% Obama the most liberal), and vice versa to the right. That part Silver and Krugman get, and if you look across the top line you can see that the model in 2008 and perhaps today give Obama a slight edge with swing voters (6 of 11 boxes on the top line have “Ob” for Obama while only five have “McC” for McCain and now Romney).

The problem for Obama is the vertical aspects of campaigns, which indicate how likely each adult is to vote.  The vast majority of Obama voters will vote for him again, virtually all of the McCain voters will vote for Romney, and the vast majority of voters who did not vote (NV) or were not registered (NR) will once again not vote. If you count the boxes in the model, Obama had 35 boxes in 2008 and McCain had 30 boxes, so at 2 million a box it was almost a 70 million to 60 million race.

So where are the changes in the model? Obama’s incredible turnout in 2008 had liberals who usually didn’t make it out (100% horizontal, 30% vertical) showing up to vote for him, and clearly he is losing a box there that will not vote this time (smaller rallies, fewer Democrats voting early, etc.). In fairness, they have continued to register additional voters who will vote for him, so the 90%/30% box is voters that went from Not Registered to Obama.

When you go to the Swing Voters column (50% horizontal), Obama dominated in 2008 as signified by four boxes of Obama votes. Even giving Obama the slight edge in polling in battleground states, Romney has clearly taken some of those voters back and some are sitting this one out (1 box moves to Romney, one to Not Voting).

But as you get to the right of the graph, you see how the huge increase in early voting, rally sizes, and a much bigger ground game in places like the coal fields of Virginia and Ohio are changing conservative not voters into Romney voters this time. Those who are slightly conservative (30% horizontal) and typically only 50% likely to vote (50% vertical) sat 2008 out and are voting for Romney in 2012.

When you go through the whole model, you end up with 34 boxes for Romney and 33 boxes for Obama, a potential two million vote win. I am writing in national terms here, but the basic model – albeit with far fewer voters per box – is playing out in every swing state.

A three-dimensional understanding of polling and turnouts points to a slight win for Romney. I do not believe Silver or Krugman are stupid for thinking otherwise, but I do believe they are naïve on half of the equation – political turnout represented by the vertical aspect of these tables.

The table below summarizes the table above to show the 35-30 boxes for Obama in 2008 transforming to the 34 boxes to 33 boxes win for Romney. Of course, on a state-by-state level, each box isn’t really two million voters, but the same table is playing out in every battle ground state with a very similar layout. So whether or not it actually translates to a 68 million to 66 million win is irrelevant for the race – but the fact that narrow edges should emerge in a few of the battle ground states due to Obama’s huge turnout edge from 2008 being gone points to a Romney win.

While I said at the outset I defend Silver against Joe Morgan and his other critics, there are times the stats in a void miss a key result. I am reminded of Joe Morgan hitting a single during a Game of the Week and runners advancing to score and to third base. On the throw home, Morgan could have easily taken second base, but he suddenly turned around and returned to first base.

Even as a child, I saw the wheels turning. Joe knew that if he went to second base they would intentionally walk Johnny Bench with first base open – and Joe had an inkling that he wanted Bench to hit. If Nate and I were sitting in the stands we would have quickly computed that giving up the extra base by retreating to 1st base cost the Reds the likelihood of scoring, and we would have both been correct 90% of the time.

However, Bench came up and drove a three-run homer out of the park. Even with a ton of stats in baseball, Silver is only right 90% of the time and Joe Morgan is right 10% of the time.

With far fewer stats to measure turnout, Nate is going to be wrong a lot more of the time in politics, and in this case, all the things he can’t measure are pointing to a Romney win. True, the polls and other things Silver measures well are pointing to an Obama win, and that makes it close, but all in all I stand by the 58% probability of a Romney win on Tuesday.

Photo of Paul Krugman: Ze Carlos Barretta
Photo of Nate Silver: Randy Stewart