Is Romney Really Ahead

Right now, the RCP average has Romney up 1.3%.  I believe for the first time this election, every poll that makes up the average is either in Romney's favor or tied (two of them are).  That tells me that Romney is ahead nationally, if by the narrowest of margins.  And as Ken has mentioned numerous times, it is a bad sign for Obama that he consistently scores below 50%.  Even if Obama wins this thing, it will be by a margin far narrower than four years ago.

But is Romney in the truest sense winning?  When you look at the RCP Electoral College map, Obama still has a lead of 201-181, with a large number, 156, undecided.  Nate Silver's model is, as it has been throughout, more favorable to Obama.  While his popular vote model has Obama winning by the slimest of margins, 49.9-49.0, his Electoral College prediction magnifies that lead (as it usually does) by giving Obama 285.4 to Romney's 252.6 (why doesn't he round those numbers?).   

When we look at this state-by-state we can see the reasons behind the RCP vs. Nate Silver discrepency.  RCP, for example, is now listing Michigan and Pennsylvania as swing states.  Silver says Obama has a 93% shot at winning Michigan and a 90% shot of winning Pennsylvania. Those are not "swing state" percentages.  Even giving Romney Florida, Virginia and Colorado (all of which Silver gives a narrow lead to Romney) leaves Romney at 257, not enough.  This makes Ohio crucial.  Silver is still giving Obama a 65% shot of winning Ohio.  Romney, let's make clear, with near certainty cannot win the presidency without Ohio.  Well, I should point out that if he lost Ohio but won Wisconsin and Iowa he'd also win, but I find it very hard to believe Romney wins those two states while losing Ohio.  Indeed, I find it hard to believe Romney wins Wisconsin (all recent polls show Obama with a slim lead there). 

Right now Silver says that there is a 4.6% chance of Romney winning the popular vote but losing the Electoral College.  I could easily see that happening.  Romney's support is not as geographically diverse as Obama's.  I can see Obama winning many key states, like Wisconsin, Iowa, Michigan, New Hampshire, Pennsylvania, etc, by slim margins (again, they will certainly be slimmer than four years ago) while Romney cleans up with huge majorities all over the South, and perhaps even surprisingly large majorities in Florida and North Carolina (I am not talking huge, but perhaps beyond what people are expecting). 

In short, while Obama trails in the popular vote by a very slim margin, I still consider him the favorite.  Ken is not as fond of Silver's model as I am, but I find Silver's prediction of 62% chance of an Obama win just about right.  Romney is at 37% (I have rounded) chance of winning.  That means there are a handful of realistic scenarios in which he wins (note how closely his chance of winning the whole enchilada tracks with his chance of winning Ohio), but as I have been saying for about six months now, Obama STILL has an easier track to 270 than Romney.

Note that all data in this post is time sensitive.  These models change every day.  If you follow the links you may find the numbers different than I have posted here.   Further note that somehow I missed Ken's post below that makes similar but not really as eloquent arguments.   

Comments

[Stefan]

RCP electoral map lags behind the RCP popular vote. This is because the tracking polls only track the popular vote. State polls are much more scarce. For instance, since your post (sometime today) Missouri has gone from swing state to leans Romney - this being a domino falling as a result of the bounce Romney got from the 1st Presidential debate.

It took RCP about 4 days of ~1pt national Romney lead for Missouri to topple over. In a few more days, we'll have a good idea where the electoral college is as a result of the debate bounce.

[Ken Blanchard]

The only problem I have with the above is that it seems to treat Silver's forecasts as data when they are not data. They are interpretations. Do I think Romney is going to win? I wish I did. I will start thinking that when Romney is leading by five points in the RCP average for Ohio. If that happens, then all Nate Silver's magic will evaporate.

I repeat, if in early November Romney is ahead slightly in the national polls but lagging in key states, then Obama will very probably win. If Romney is leading by four or five points nationally then he will probably be leading by a point or two in Ohio. If he is, then he will win. You don't need Silver to tell you that.

[larry kurtz]

whistling in the dark attacks predators fellers: earth hatred in a nutshell.

[Jon S]

How about In Trade that has numbers similar to Silver's? Today it is 61-38 Obama. Is that data or interpretation? Is Silver's "interpretation" based on ether or on data?

[Bill Fleming]

Silver isn't going for "magic" KB. From what I can tell, he's going for science. And he's actually more critical of his model than anyone else is. (Sign of a good science guy.)

http://fivethirtyeight.blogs.nytimes.com/2012/10/13/oct-13-arizona-and-the-spanish-speaking-vote/

p.s. looks like there may be a fly in the ointment in everyone's model when it comes to the Latino vote. Like cell-phone-only voters, sort of a systemic flaw in the data gathering technique until recently.

Whoops. There goes Florida, Colorado and Nevada. Maybe even Virginia? Who knows?

[Ken Blanchard]

Bill's comment makes my point, which was simply that you can't support Silver's bottom line projections by citing his other projections. If the four states mentioned by Mr. Fleming really are shifting decisively toward Romney and if Ohio and Wisconsin follow suit, what will Silver do? He'll adjust his projections. Silver is doing more, but not a lot more, than what anyone does who follows the RCP average on a regular basis.

Like you, I take the In Trade numbers seriously because they have been very reliable in the past. However, I remember once taking the Weekly Reader poll very seriously because it had never once been wrong. Until it was wrong (Bush 41).