One change today. Just barely Romney Florida becomes just barely Obama Florida:

Now, some of you may remember my commentary on August 22nd when Florida moved to Romney. I mentioned that what pushed the average over to Romney was a poll showing a 14% lead for Romney… which was way out of line with any polling in the previous year… many sites that do poll analysis were just dropping it entirely as an outlier. I explained that I just leave everything in, and figure outliers will wash themselves out on their own before too long. That is exactly what happened here. As soon as the outlier poll aged out of the “Last 5 Polls” that I use for my average, Florida popped right back to being (just barely) leaning Obama. If I had simply excluded this poll, Obama’s lead in the five poll average would have dropped to 0.2%, but it would never have gone negative.

In any case, the five poll average now stands at an 0.8% Obama lead in Florida, so the state gets moved back into his column for the moment. I will caution, as I always do, that even leads up to 5% can be extremely ephemeral under the right circumstances. That is why I consider all states in that range as able to go either way pretty easily. A lead of less than 1%? That really is completely up for grabs. So nobody should read too much into Florida being just barely on one side of the line vs just barely on the other. Either way, Florida is still looking like a tossup.

This does put the “current” model right back to where it has been most of the year though:

In the most recent episode of Curmudgeon’s Corner recorded yesterday and released today, I mentioned that my “gut feel” was that we have seen Romney’s post-primary high water mark, and that we’ll start seeing Obama gaining ground going forward from this point. Despite today’s Florida change, the numbers don’t show that yet. We only have a small amount of post-RNC state level polling, and of course no post-DNC state level polling since that event isn’t even done yet. The right set of things hitting the news and things could break rapidly in Romney’s direction.

In recent times though, challengers have usually lost ground between their convention and the election. Nate Silver recently looked at this in detail. Silver of course points out that this is not a universal rule. 1996 and 2008 both did not fit this pattern. There is still lots of room for the daily ebb and flow of national and international events… and campaign rhetoric… to change the shape of this race. Generally speaking though, if Romney follows the typical pattern, he is done. To date he has never been ahead in this race. He needs to be gaining ground to win. He has no room for the sort of “normal” decline that Silver describes.

Romney needs something big to vault him into a clear lead. His convention was one of the biggest chances for that, and so far it looks like if he got any boost, it was minimal. Assuming we don’t see a very unexpected negative effect coming out of the Democratic convention that ends up helping Romney, this means Romney’s next big chances are the debates… which usually don’t have much of an effect unless someone screws up… or to just hope that Obama just gets pounded by bad news, economic or otherwise, that ends up eliminating Obama’s lead.

Romney could indeed win the debates decisively, or Obama could get hit by those waves of bad news… and that would change things… but like I said, my gut is saying the beginning of September will have marked Romney’s post-primary high water mark in my models. So write that down. I’m going out on that limb. You can all call me on it when it proves to be completely wrong. Which now that I’ve said it “out loud”, will almost certainly happen. :-)

Note: Chart and map from the Abulsme.com 2012 Electoral College Prediction page. Both assume Obama vs Romney with no strong third party candidate and show polling as it currently exists. Things will change before election day. On the map red is Romney, blue is Obama, gold states are too close to call. Lines on the chart represent how many more electoral votes a candidate would have than is needed to tie under several different scenarios. Up is good for Obama, Down is good for Romney.

I mentioned yesterday that several states were right on the edge of tipping if things went Romney’s way. Right on cue, a couple new polls and the biggest one, Florida, flips over to Romney…

For the past few months Florida has been leaning toward Obama, but only by the slightest of margins. Now the five poll average shows a Romney lead in the state for the first time since May.

Now, I will point out that the poll that pushed the five poll average over the center line was one from Foster McCollum that some places are treating as an outlier and therefore are not including in their averages. It has Obama up by 14%. By comparison no other poll has shown a Romney lead of more than 5% since May, and the biggest lead Romney has shown in any Florida poll in the last YEAR is 7%. Nate Silver actually specifically discussed this poll and the problems with it and why some people are excluding it in a recent post.

For my models, if a poll turns up in any of my several sources, I use it. I do not do any specific outlier removal. I let the five poll average do what the five poll average does, and if you have an outlier, it will still be balanced by the other four polls in the average, and soon enough as new polls come in, it will age out of the average anyway. With this poll, Florida shows up in my average with a 2.8% Romney lead. If you excluded it, you would have an 0.2% Obama lead. Either way, the polling shows Florida is very very close and could go either way.

I do count it though, and this means the first change in the “Current” line in my model since May (the last time Florida flipped).

This puts Romney back where he was in May in terms of the electoral college results if all the states vote according to current polling. That is, Obama wins 303 to 235. As I mentioned yesterday, in terms of his best case (where he wins all the close states and wins 317 to 221) this is the best position Romney has had since February.

Obama is still ahead in this race. But Romney has been gaining on him all month. Things look a lot closer now than they did a few weeks ago. And we’re about to head into convention week for the Republicans, so we should expect a further bounce toward Romney. (Although some analysts think it will be smaller than usual since so many people have already made up their minds.)

The things to watch will be if the bounce is wide spread enough and long lasting enough to actually put Romney in the lead for the electoral college for the first time ever. And if it does, can he keep any of those gains? The convention bounces are usually ephemeral and after the hubbub from both conventions die down, you end up pretty close to where you were before them.

Ohio (18 ev), Colorado (9 ev) and Iowa (6 ev) have Obama leads less than 2% at the moment. Will Romney pick off one of them next? If he is going to, the next couple of weeks are when you would expect it to happen first. So keep watching.

Note: Chart and map from the Abulsme.com 2012 Electoral College Prediction page. Both assume Obama vs Romney with no strong third party candidate and show polling as it currently exists. Things will change before election day. On the map red is Romney, blue is Obama, gold states are too close to call. Lines on the chart represent how many more electoral votes a candidate would have than is needed to tie under several different scenarios. Up is good for Obama, Down is good for Romney.

Chart and map from the Abulsme.com 2012 Electoral College Prediction page. Both assume Obama vs Romney with no strong third party candidate. Both show polling as it currently exists. Things will change before election day. On the map red is Romney, blue is Obama, gold states are too close to call. Lines on the chart represent how many more electoral votes a candidate would have than is needed to tie under several different scenarios. Up is good for Obama, Down is good for Romney.

One state changes status today, and it is a big one!

In my five poll average Florida moves from Obama up by 0.6% to Romney up by 0.2%. In either case, the bottom line is that Florida is too close to call and could go either way. The recent trend has been toward Romney though. We’ll see if Romney can pull Florida closer toward him, or if it remains a super close swing state. (Which is kind of what everyone expects.)

Anyway, a big state flips to the Romney side. This effects the “current” line by moving 29 electoral votes in Romney’s direction, but the “best cases” representing the range of possible outcomes stay the same:

I had said I wouldn’t mention national level polls much, but there has been a big enough move there lately to warrant another mention. In the last couple of weeks there has been a substantial move toward Romney in the national polls, with the RCP Average dipping to an Obama lead of as low as 0.2% on May 8th, although it is back to 1.3% as of today. So why do the state numbers still show a massive Obama lead?

There are two answers, one is of course that the election is based on the states, not the national popular vote, and they do not always move in lockstep. But frankly, they DO usually move together, and only tend to diverge much in really close situations. The electoral college just usually shows an “amplified” version of the separation between the candidates in the national popular vote.

The more important factor right now is the second answer. State polls are still slow. There are new national polls every day. Meanwhile even important close states like Florida are still getting polled only once a week or so. Most states are polled even less frequently. This means that any state by state analysis will take longer to react to changes.

If the race narrows significantly and stays narrowed, the state numbers will start showing that soon enough. (And state polling in close states will get increasingly frequent as we approach the election, so this lag will decrease dramatically later in the year.)

Edit 2012 May 10 15:56 UTC – Moved summary box earlier in post.

Map and chart from the Abulsme.com 2012 Electoral College Prediction page. Both assume Obama vs Romney with no strong third party candidate. Both show polling as it currently exists. Things will change before election day. On the map Red is Romney, Blue is Obama, Gold States are too close to call. Lines on the chart represent how many more electoral votes a candidate would have than is needed to tie under several different scenarios. Up is good for Obama, Down is good for Romney.

Remember how a few days ago I said things were starting to improve for Romney? Well, that didn’t last long…

Obama’s lead in the five poll average for Florida increases to over 5%. It leaves the swing states and moves to “Weak Obama”. My “best case scenarios” give one candidate or another all of the swing states. Florida makes a big difference. So what do things look like now?

Yes, that is right. With current state polling, if Romney wins ALL of the states that are close…

He still LOSES 260 to 278.

There are major caveats to give:

• November is still far away. Romney hasn’t geared up his campaign machine. Important states are not yet being blanketed by advertisements. We haven’t had the conventions yet. Most people aren’t paying attention yet. News events will change the agenda. The situation in April is not a good predictor of what will happen in November. Obama has a head start. That is all that can be said.
• State polling is still sparse. In Florida, five polls goes back to January. Florida is more frequently polled than many others. Some states actually haven’t been polled at all. Until polling ramps up, this projection will lag events.

Despite this, it is remarkable for Obama to be showing this kind of lead. Romney has a lot of work to do. But wait…. haven’t there been polls in the news lately showing Romney ahead? What is going on?

First, never pay attention to individual polls. Instead look at one of the places that aggregate. Lets look at RCP’s Average. Obama’s lead was usually between 4% and 6% from February until April. It was 5.3% on April 10th. It declined to 2.4% on April 14th. This dip is eight days old. Too short to be picked up in state polls. The projection here is consistent with the lead Obama showed until the last few days. If the new trend is sustained, it will show up in the state results soon.

Second, the president is not elected by popular vote. It doesn’t matter if Obama is ahead in California by a little more or less. Obama will win California. But that will show up in national polls. Popular vote and electoral college will only diverge in close elections, but the election is decided by the electoral college. Look at the popular vote polls as leading indicators, but in the end, look at the electoral college. That is how Presidential elections work.