My non-statistical intuitions about the election have corresponded almost exactly with the predictions of one of the most respected statistical pundits in the country

Last night I began drafting, but did not finish, an entry in which I said the following two things: (1) that a month ago, before the debates, I had privately thought and said (but didn’t say publicly because I didn’t want to add too much to the conservative gloom) that my own feeling was that Obama had a 90 percent probability of winning the election; and (2) that after the debates, I now feel that that the probability of an Obama victory is 67 percent (2/3), and possibly as high as 75 percent (3/4). (Those two points were a preface to my main point, which was that Romney could have clinched the election if in the third debate he had gone decisively after Obama on Benghazigate and the disaster of the Arab Spring, but that because he didn’t, but instead kept amazingly expressing his agreement with Obama’s foreign policy, the odds remain against him.)

Just now, I saw an article about Nate Silver, the Wunderkind of “scientific” election prognosticators, who became famous by correctly predicting 49 of 50 states in the 2008 election. Joshua Jordan at NRO writes:

On September 30, leading into the debates, Silver gave Obama an 85 percent chance and predicted an Electoral College count of 320–218. Today, the margins have narrowed—but Silver still gives Obama a 67 percent chance and an Electoral College lead of 288–250 …

Jordan goes on to argue that Silver’s model is wrong, and that the election is much closer than Silver thinks.

- end of initial entry -

Stephen L. writes:

I respectfully disagree with your and Nate Silver’s analysis. It will not be close. Mr Obama is soon to be a former president, doing the speaking circuit and raking in millions. Remember, Republicans make their money before getting elected, Democrats after.

Timothy A. writes:

I have long believed that, given the advantages of incumbency, Obama would easily be re-elected. However, as of today, three big national tracking polls of likely voters (Gallup, ABC/Wash Post, Rasmussen) all have converged on Romney 50 percent Obama 47 percent numbers. If we apportion the 3 percent undecided vote equally among Romney, Obama, and third party candidates, we would have a national popular vote of Romney 51 percent Obama 48 percent. A 3 percent win in the national popular vote would certainly result in an Electoral College victory as well.

LA replies:

Go to the RCP electoral map and see how Romney now has 206 solid, likely, or leaning, and Obama has 201. If we assume the solid, likely, or leaning for Romney are all definite, then he needs to get 64 electoral votes from the Toss Up states to reach 270 electoral votes and the presidency. Based on the RCP averages in the Toss Up states, try to put together a likely scenario by which Romney gets 270. I did this last night. I started by assuming that Romney would get Florida (29), Colorado (9), and Virginia (13). That gives Romney 257. But then I couldn’t put together further likely Romney wins providing the additional 13 votes. Indeed, I couldn’t find any other states where I felt confident that he would win.

Whoops, when I said Romney has 206, that was based on the last time I looked at the page last night. Now I see that RCP has moved North Carolina with its 15 electoral votes from Leans Romney to Toss Up, reversing the change about ten days ago when North Carolina was moved from Toss Up to Leans Romney. So now he has 191 electoral votes and needs 79.

It’s entirely possible that Romney could win in the popular vote but lose in the electoral vote and thus lose the election.

Of course Romney could still win. But in my view the path to victory is not readily apparent.

Romney thought he could defeat an incumbent, charismatic president by smiling at him and agreeing with virtually his entire disastrous foreign policy. Does that sound like a winner? Or does it sound like your typical Republican loser—your Bob Dole, your John McCain?

Timothy A. replies:

If Romney gets (at least) 51 percent of the national popular vote, as three of the big national tracking polls seem to indicate he will, then he will win Ohio. That would put him over the top in term of Electoral votes. I say this because for the last nine Presidential elections, the GOP candidate has received a higher percentage of the vote in Ohio then he has nationally. The last time this was not true was in 1972 when Richard Nixon was getting 70 percent or more of the vote in the deep South and ended up with about 61 percent nationally. In Ohio he “only” got about 59 percent of the vote. I believe that the only other time in the history of the GOP that their candidate underperformed the national vote in Ohio was 1964 (Goldwater was getting 70-80 percent of the votes in the deep South and underperformed his national percentage by a few points in Ohio). If one worries that Ohio no longer resembles the historical Ohio, I would point out that in the last electoral cycle (2010) Ohio elected a Republican Governor, Senator, and swept all of the other statewide offices as well.

Now, it may be that the three tracking polls are wrong and Romney won’t win the national popular vote. In that case, he will probably lose Ohio and will certainly lose in the Electoral College.

LA replies:

Interesting argument.

An obvious flaw in it is, Romney is now getting over 50 percent of the national popular vote in the three big national tracking polls, but, contrary to your logic, he is not getting over 50 percent in Ohio.

Timothy A. replies:

It seems to me that the state polls which are aggregated at sites like RCP are older, have smaller sample sizes, and include polls by smaller, less reputable pollsters than do the national polls, and therefore are less informative. The RCP average poll on this date in 2004 (plus or minus a day or so—I looked this up a day or two ago) had Kerry up by 0.7 percent in Ohio. Bush won the state by 2.9 percent.

When I see pictures of Romney at rallies like the one this evening in Defiance, Ohio, they are nothing like McCain or Dole rallies 12 days before their respective elections.

Kristor writes:

If you go to 270toWin, you can play with the electoral map. The University of Colorado Political Science Department model predicts Romney will win 330 electoral votes; the likelihood he will hit that number is 75 percent. The model has had 100 percent success in “predicting” the last eight Presidential races. The model doesn’t operate on polling data at all, but on the delta (or was it the delta of the delta?) to average real household after-tax income in each state over the period preceding the election. It uses these data to predict how each state will vote, and then compiles these outcomes into electoral college votes.

Quoting an October 4 article on the CU Boulder website:

Of the 13 battleground states identified in the model, the only one to change in the update [from August] was New Mexico—now seen as a narrow victory for Romney. The model foresees Romney carrying New Mexico, North Carolina, Virginia, Iowa, New Hampshire, Colorado, Wisconsin, Minnesota, Pennsylvania, Ohio and Florida. Obama is predicted to win Michigan and Nevada.

This was before the debates had had a chance to influence the electorate, and Romney’s poll numbers had improved. I come up with 331 votes for Romney on account of what I have heard about District 2 in Maine, which is going heavily for Romney. Maine is one of two states that split the electoral college vote.

LA replies:

My first reaction is: that’s very hopeful news.

My second reaction is: If this model has such a fantastic record of successful predictions, why isn’t it better known? Why does the whole world focus on polls rather than after-tax household income?

Posted by Lawrence Auster at October 25, 2012 06:48 PM | Send