28 August 2012

Electoral Landscape and Predictions: End of August 2012

Bonncaruso's Electoral Landscape: Obama vs. Romney
August 28, 2012

This is my second major electoral landscape prediction-output for the 2012 election. The first one was on July 15, 2012, which you can see HERE or HERE. For those who know my analyses, they know that while I am a Democrat, my electoral analyses are non-partisan and brutally fair. I simply let the numbers speak for themselves.

There were a good number of polls in 2010-2011, you can see them all HERE (national), HERE (Alabama through New Hampshire) and HERE (New Jersey through Wyoming).

As of 2012, here are the stats on all 2012 polls, all of which are in one EXCEL-document HERE (through 28 Aug 2012).

Here the raw stats on polls, per state:

StatePolls through 2012-07-015Polls through 2012-08-028Change
District of Columbia011
New Hampshire8124
New Jersey9123
New Mexico682
New York11143
North Carolina17225
North Dakota231
Rhode Island000
South Carolina000
South Dakota121
Washington (State)7114
West Virginia110
Total STATE POLLS281399117
National 277400123
National – specialty polls14195
Total – without specialty polls558799241
Total – with specialty polls572818246

So, there were 399 state polls since the beginning of 2012 up through August 28th. All 818 polls are in the EXCEL document that is hyperlinked above.

There were roughly 400 national polls, but sometimes the Gallup and Rasmussen poll values over a weekend did not change and I did not input them until a change occurred.

The specialty polls (ACA, latino vote, swing state votes) are not per-se directly involved in the state calculations, but worth a look nonetheless.

As of August 28, 2012, the following „states“ have not been polled at all:

AlaskaDelaware, HawaiiIdaho, Kansas, Kentucky, Louisiana, Mississippi, Rhode IslandSouth Carolina and Wyoming. (11 states). The other 41 „states“ have been polled. It is pretty safe to assume that these 12 states will vote in 2012 in the same direction as they did in 2008. None of them are likely to be competitive.

Most of the states that have been polled the most are logically those considered the most competitive in the GE:

Florida (36, was 25)
Ohio (31, was 20)
Wisconsin (30, was 19)
Virginia (29, was 22)
Pennsylvania (24, was 17)
North Carolina (22, was 17)
Michigan (21, was 13)
Colorado (15, was 9)
Missouri (15, was 6)
California (14, was 11)
New York (14, was 11)
Massachusetts (13, was 11)
New Hampshire (12, was 8)
New Jersey (12, was 9)
Arizona (11, was 10)
Washington State (11, was 7)
Nevada (10, was 6)

Alone from these 17 states in the Union we have received data on 320 of those 399 state polls. 

Now, not all of these seventeen states are considered battlegrounds in 2012: California, Massachusetts and New York are all very safe DEM states, but CA and NY have a lot of CDs and therefore many electors, so the polling for CD primaries brought with it presidential match-up numbers as well. For this reason, so many polls of three rock-solid DEM states. Nonetheless, these values can also be very informative. More on this later.

Of the 17 states listed above, 15 were Obama wins in 2008. Missouri and Arizona are the two „McCain“ states from 2008 on this list of most polled states.

Based on the polling results, which, as mentioned above, you can find all HERE in one excel table, I come up with the following electoral landscape:

This is not the same EC predictions as the polling tracker from TPM, which currently shows the following:

Graphic source.

To-date, there are no special predictions for the congressional districts in either ME or NE, both of which do elector-splitting.

My methodology is similar to TMP's, but not identical. I take the average of all the last polls, no more than the last 7 to 9 polls within three weeks time, (in July, it was within one months time) and with no repeater pollsters. Closer to election day, this time-frame will be reduced to two-weeks and for the battleground states, one week. Any average at +2% or less is pure tossup. Otherwise, I designate a prediction.

What to make of this and the current polling data?

First, at the national level, we are seeing -essentially- a dead heat between President Obama and Governor Romney, but the national polling does not predict elections very well: state polling does, for the electors are decided on a state to state basis. Also, we have had extremely close National Popular Vote elections where the victor still came close to or 300 EV in the electoral college (see: 1948, 1960, 1968 and 1976). Furthermore, the national results are not really manifesting themselves at the state level – yet. This is the same prognosis as with the first electoral landscape on July 15, 2012. On the national level, nothing has changed.

Now, to the tossups (7 states, 83 EV):

CO (Obama +1.60)
IA (Obama +1.5)
MI (Obama +1.68)
NV (Obama +2, only one poll, SUSA)
NC (Obama +0.67)
OH (Obama +1.67, one write-in poll)
VA (Obama +1.00)

Notice that every single tossup state has the ever so slightest Obama advantage. Now, some of this is simply because older polling from before July 15 has long fallen out of the statistic and some of it is just simply because the season is starting to heat up:

1.) There are only two polls in the mix in Iowa right now, a RAS poll from 10.08 and a PPP from 17.07, roughly three weeks apart from each other. That is very sparse data for a state that looks to be incredibly competitive.

2.) Ohio has been reliably Obama leaning and is actually continuing to do this, but a write-in poll for the Columbus Dispatch has blurred the statistic. Without that poll, OH would not be a tossup. Even so, the average would be lower than in July, so there has been erosion for the President in the Buckeye State – not a good sign for him. Since Ohio and Nevada are the two most reliable bellwethers in the nation at this time, impending Obama wins in both of those states would pretty much assure a national victory. Likewise for Governor Romney: if end polling shows him ahead in BOTH states, then most likely Governor Romney would be the next President of these United States.

3.) In Nevada, there is only one poll that is recent enough to be used and it shows Obama at +2, therefore in the tossups. But Obama has won 9 of the 10 polls conducted in NV this year, so his lead in the Silver State is more resilient than it currently looks on paper.

4.) Michigan keeps appearing as a possible battleground and then disappears back into Obama territory, and as is the case with Florida, one very likely outlier poll with very bad demographic distribution is polluting the statistics. More on this poll and the pollster later. MI is a 5-for-5 Democratic state and incumbent Bush 43 was unable to peel this state away from the Democrats in 2004. Wait and see, but the probability is actually very high that at the end of the day, Michigan will vote Democratic. However, at this time, it is the only core Democratic state in play.

5.) North Carolina is exactly as it was in 2008: a bitter battleground. This is actually a statistically postive change for Obama, for in the last electoral landscape, NC was Romney +1.20%, now it is Obama +0.67% in a state that Obama won by +0.29% in 2008. We are talking about micro-numbers here in relation to the nation and we are talking about GOTV on Election Day. Of all the states polled, North Carolina is without a doubt the closest one and I expect this will remain so, when all is said and done. I believe that from the flurry of activity on the ground in North Carolina, we will learn alot about the race: if President Obama feels confident that his lead in VA is holding, then he will sink lots of resources into North Carolina.

6.) Virginia has narrowed since the last Electoral Landscape analysis. a must-win state for the GOP, which Obama won by +6.30% in 2008. Of 29 polls in 2012, Obama has won 22, Romney has won 5 and there were two absolute ties. His current margin average is +1.00, it was +3.25% in July. In a very large analysis of Virginia from January 2009 (which you can read in full starting HERE), I wrote the following conclusion:

However, in VIRGINIA, there was real resistance to Obama, but in counties that are „emptying out“, so to speak. We see a large poli-demographic shift in VA, with the north and the SE gaining greatly in political strength for the Democratic party. Here there were obviously far fewer GOP defections, if at all (McCain scored more raw votes in VA than Bush from 2004), but far more newly registered and democratic dedicated voters. This poses a far larger problem for the GOP than either Indiana or Ohio, for Obama's 1.03% margin in Indiana can be overcome and Ohio is expected to be a battleground state is virtually every cycle, but the addition of more than 500,000 voters to the democratic rolls in just one cycle is much harder for the opposition to overcome. The best case scenario for the GOP is that Virginia becomes a bitter battleground state. However, +6.30% is hardly a battleground margin. It is a better margin that Obama scored in OHIOFLORIDA, INDIANA and NORTH CAROLINA. It is a lean winning margin, but a comfortable one and will require a minimum 12.60% shift back in order for the GOP to regain the state, and I doubt that this shift will come from those 500,000 new voters. The worst case scenario for the GOP is that Obama cements Virginia into the Democratic column in his first term, adding the state to core Democratic territory and thus making the electoral math for the GOP more difficult. „

And I believe this is exactly what the Obama team has been doing: working to „cement“ Virginia into the Democratic column as a new blue state. The fact that, in spite of a bad economy, Obama has been able to maintain the upper hand until now - and I expect he will visit the Old Dominion quite often. Look what Virginia does for the electoral math. If Obama retains all the western states and cements Virginia into the Democratic column, then were we to consider OH, MI, FL, NC and IA as pure tossups, Obama would have 263, Romney would still still at 191. Obama would only need to win one of the five to get over 270.

7.) Colorado has also narrowed, but again, because of a dearth of polls. And the two polls showing it extemely close or with Romney ahead will fall out of the statistics next week. So, as was the case with Nevada, Colorado is more solid than it currently looks on paper.

That all being said, this is the lowest EV level for Obama since I have been following the statistics for 2012, and the best level for Romney, who has had problems getting above 191 EV for a long, long time.


But there is a caveat in that as well:

Florida is currently in the Romney column, but only because of one wild outlier poll from the same firm that did the one wild outlier MI poll. That firm, Foster-McCollum / Baydoun, put out a poll showing Romney suddenly at +14.58% over Obama in a state where the two have been neck-and-neck for the last 16 months. Nate Silver at 538 tore this poll apart because the projected demographics are just so far away from the actual demographics of the Sunshine State and even farther off from the demographic break-down from 2008 that this poll is very likely to win OUTLIER POLL of the year award in my book. With this poll in the statistic, it is now Romney +2.97, but without this poll, it is Romney +0.06. Yes, +0.06, which would make Florida look eerily like Florida of 2000. This poll will fall out of the statistic in little over one week, so more likely than not, Florida will return to the tossups in the next electoral landscape analysis.

Key states currently NOT on the tossup list:

Pennsylvania was a state that was looking very iffy and dangerous for the President in 2011, but statistically looks more solid right now. President Obama is holding a +7.67 average over Governor Romney. In July was a +7.33 average, practically unchanged. The only problem with this is that Pennsylvania has a new voter ID law that may really make things rough for about 800,000 voters, which could indeed swing those electors to Romney. So, for now, Pennsylvania looks safe, especially in consideration of its narrow-margin electoral history. But this can change, and is a major warning sign for the Obama team, no doubt.

For most of 2011, New Hampshire was looking like a very solid GOP pick-up in 2012, but since then, President Obama has retaken a lead in the Granite State (average: +5.00, it was +2.67% in July, an improvement for Obama). More importantly, in NH polling, the President is now hitting 50%, an important mark. In the case of a very close election, it is very conceivable that Governor Romney can overtake the lead in New Hampshire once again.

With its very turbulent recall elections, Wisconsin has become a hotbed of partisan activity, but in reality, President Obama is still holding an average lead of +2.71% over Governor Romney in the Badger State, in spite of Romney's selection of Paul Ryan to be his running mate. That being said, Obama's average in Wisconsin was +6.67% in July, so there has been measurable erosion here and a strong possibility that Wisconsin will enter the tossup zone in the next two weeks. Obama won Wisconsin by a landslide +13.90% in 2008, the largest presidential winning margin for any candidate since 1964. Therefore, Obama has lost a lot of ground here. However, Wisconsin is a 6-for-6 Democratic state with enough of a solid Democratic base to pretty much insure an incumbent Democratic win in 2012, even if just a narrow win. The only Democratic incumbent to not win WI in the last 64 years is Jimmy Carter, who lost the Badger State to Ronald Reagan by 4.72% in 1980, far under Reagan's +9.72% national average. This means that most likely, if Romney is winning in WI, then he is winning by close to a landslide nationally, which the national numbers are currently not predicting.

What about some other battlegrounds from 2008?

Missouri was the closest race of the night in 2008 and it took 15 days to finally declare a winner: John McCain, by +0.13%. For months on end, I complained about the sparse polling coming out of Missouri. No more, however. In the last 6 weeks, 9 polls have been taken in Missouri, in no small part due to the „legitimate rape“ statements of GOP senatorial candidate Todd Akin, which has really caused a shake-up in the numbers in MO. That being said, the state is trending more and more conservative from year to year and is likely not going to be a battleground unless Obama is nationally around 8 points ahead of Romney, which is not the case right now.

Likewise, Indiana, which Obama won by +1.03% in '08, was polling in upper single-digits for Romney and they showed Romney at about +7.5 over Obama in the Hoosier State at the time, a kind of margin that is not even close to the margins that the GOP once enjoyed in the absolute RED bastion. However, the last Rasmussen poll showed Romney with a very healthy +16 point lead in Indiana. I did a massive analysis of Indiana in 2009 (which cuts through a lot of propaganda), which you can read starting HERE. The latest

One state that is very noticeably NOT on the battleground radar this time around is New Mexico, which, as a minority-majority state, firmly in President Obama's column. He won NM by +15.13% in 2008 and current polling shows him between +11 and +14. Also to note is that NM is one of the states from 2008 where Obama vastly exceeded the final polling average, again a sign that the Latino vote in many states in 2008 was not correctly weighted into the polling results.

How about the rock solid states?

Well, one would think that there should not be much ado about the rock-solid core states for both parties, but they would be wrong. The margins coming from polling of those states can indeed tell us a lot about the direction of the race.

For instance, Obama won New York by +26.86% in 2008: he is at an average of +24.5 from June and the July and August polling that has now come in shows him at +23 and +29, the average of which (were the polls in close enough proximity to each other, which they are not) would be +26, which shows absolutely no erosion of his base in such a large blue state.

In California, there is some statistical erosion, not enough to make the state even close to competitive, but enough to affect the national margin in a possibly close race. In California, where Obama won by +24.03% in 2008, the current average shows Obama at +18.3 from June and +16.95% at this time.

On the Republican side, in Utah, the only poll thus far for 2012 shows Governor Romney ahead by a whalloping +42 over the President. This is not a big surprise, as UTAH has the largest Mormon population in the Union and before 2008 for 8 cycles in a row, it was the most conservative state in the Union. John McCain won UT by „only“ +28.08% (still a massive margin, ala Obama in New York), but Bush 43 won UT by +45.45% in 2004. The chances are very strong that Governor Romney will smash all records in UTAH in 2012. To date, there has been no polling from either Wyoming or Idaho (both states with sizeable Mormon population),but as Utah goes, also goes Wyoming and Idaho, to be sure.

In Arkansas, a state the Bill Clinton won easily in both 1992 and 1996, the one poll for 2012 shows Governor Romney up on President Obama by +24. This is important as data, for Arkansas was one of the five bible belt states in the Union to buck the Democratic trend in 2008 – and of the five, Arkansas bucked the trend the hardest. McCain won AR by +19.85%, and a +24 for Romney shows that the state is moving even more to the Right.

However, in neigboring Tennessee, which also bucked the Democratic trend in 2008, polling shows a leaner race between Governor Romney and President Obama. The last two polls showed Governor Romney up by +6 and +7, respectively, far under McCain's +15.06% win in 2008.

There has been individual motion in polling in other solid states worth noting:

In Vermont, the latest Castleton poll has the President at +37 over Romney. Obama won VT with +37.01%, identical to this polling average. This is a massive improvement for the President in Vermont since the first Castleton in February, which showed him at „only“ +25 over Romney.

Conversely, in Connecticut, the President is still ahead and still over the 50% mark, but his margin average is +7.67 in a state that he won by +22.37% in 2008. There is large and measurable erosion for the President here – not enough to cause him to lose the state, but interesting to note. Likewise, in neighboring MA, Obama's current margin average is +13.5 over Romney, again, in a state that he won by +25.81% in '08. That being said, the margin is being affected by a Republican poll showing him up by only 11, which is pretty unlikely here. Nonetheless, Romney IS having an effect on some states on the East Coast.

But not in New Jersey, where the last two polls have Obama at +13 in a state that he won by +15.53% in 2008. And in Maine, the President is holding at between +14 and +15 in a state he won by +17.32% in '08. So, the „Romney-effect“ is not evenly spread in the NE.

In Minnesota, the President's lead has gone from +14-15 to +6, if the SUSA poll from July is to be believed. But remember, the North Star State is a 9 for 9 Democratic state since 1976, the ony one to resist the Reagan landslide from 1984. And were it not for the Nixon landslide of 1972, it would be a 12 for 12 Democratic state. Minnesota is NOT in play.

In Montana, Gov. Romney's lead over Obama has increased, to +17 from +9 (Rasmussen, August vs. June). But in Oklahoma, Romney has gone from +35 in the Sooner Poll from May to +30 in the Sooner Poll from July.

So, on both sides, in the so-called „safe“ states, we are seeing some growth and some erosion. Pretty typical stuff for a election that is looking like it is going to be a "base" election.

What would this mean in the case of an impending landslide win for either Obama or Romney?

It would surely mean that the margins for the losing side would be DRASTICALLY reduced, and that is simply not happening right now. Most of the Obama double-digit states from 2008 are still double-digit states in 2012. Most of the McCain double-digit states from 2008 are still double-digit states for Romney in 2012. And the subtle changes we are seeing in many polling values is more evidence of the theory of slow electoral shift than of an impending change in 2012.

Now, on to some bad pollsters:

At the beginning of 2012, a fly-by pollster group that called itself the „GS strategy group“ put out way out there, wild polling results for 6 states. Here are the results. I lambasted that pollster in this THREAD at politicalhotwire. At the thread, I listed ten reasons why this was bad polling, and those reasons still hold. By the way, GSSG never again published any results in the last 7 months. Not a peep out of them.

Enter Foster-McCollum / Baydoun, which put out a wild poll of Florida (HERE) and also a wild poll of Michigan (HERE). In both cases, this pollsters has crunched the numbers based on demographics that don't even come close to being representative of the state. Nate Silver at 538 did scathing reports on both polls, HERE for the Florida poll, HERE for the Michigan poll. It is outstanding reading. Some small excerpts:

The poll was weighted to a demographic estimate that predicts that just 2 percent of Florida voters will be 30 or younger. It’s a decent bet that turnout will be down some among younger voters this year, but that isn’t a realistic estimate. In 2008, according to exit polls, 15 percent of voters in Florida were between 18 and 30.

The poll also assumed that 10 percent of voters will be between the ages of 31 and 50. In 2008, the actual percentage was 36 percent, according to the exit survey.

The poll projected Latinos to be 7 percent of the turnout in Florida, against 14 percent in 2008. And it has African-American turnout at 10 percent, down from 13 percent.

If the turnout numbers look something like that in November, then Mr. Obama will lose Florida badly. He’ll also lose almost every other state; his electoral map might look a lot like Walter Mondale’s.“

and from the Michigan poll blog of his:

The head of Mitchell Research, Steve Mitchell, wrote a long memo accompanying his poll release on Monday. In that poll, he weighted the survey to assume that African-Americans would make up only 8 percent of Michigan’s turnout. By contrast, black voters represented 12 percent of the turnout in Michigan in 2008 according to exit polls, and 14 percent according to another source, the Current Population Survey. Blacks also made up 13 percent of Michigan’s vote in 2004 and 11 percent in 2000, according to exit polls. African-American participation is sometimes lower in midterm election years, but blacks were 12 percent of Michigan’s electorate in 2006, the exit poll reported that year. (There was no exit polling in Michigan in 2010.)

So why only 8 percent this year? Mr. Mitchell wrote that he simply doesn’t believe the exit polls:

African-American participation in this poll is 8%, not 12%, which is the percentage of the population but not likely voters. I do not believe blacks represented 12% of the vote in 2008 and I don’t believe they will in 2012. Having polled this state for 26 years, blacks have represented about 7%-8% of all votes cast in every statewide race. At best, it went to 9% four years ago. It will not reach that level this year.”

What is the evidence for Mr. Mitchell’s claim? He didn’t present any of it in the memo. The exit polls certainly aren’t perfect, but they’ve been consistent from year to year and also seem to agree with Census Bureau data. (Another fact that Mr. Mitchell cites — that African-Americans represent 12 percent of the overall population in Michigan — is also slightly incorrect; they made up 14 percent of Michigan’s population as of 2011, according to the Census Bureau.)“

I noted this information, for this is also the time of an election year where fly-by pollsters will try to influence the „market“ in poll numbers, so to speak. The GSSG polling came right at the beginning of the Primary season, and now this McCollum / Baydoun is putting out wild results right as things are heating up.

Now, I could decide to throw out all outlier polls, but where does one stop? That could be very problematic, depending on your views. In 2008, I recorded EVERY poll and took the averages and at the end of day, on election night, the averages were pretty darned close to the actual results. Outlier polls fall out of the statistics at some point in time and as election day comes around, more and more respected pollsters will be putting out more results with much more regularity.


Uncomplicated answer: easy to guage if the conventions are at least a week apart from each other, hard to guage if the conventions are back to back, as they were in 2008 and as they will be this year. Normally, both sides get a bounce, which then fades, so likely we will be looking as similar national and state numbers when I put out the next electoral landscape in the middle of September...

FACIT: a lot of things can happen until election day, but in the Electoral College in August, it is without a doubt slight advantage Obama.


Update: just as this thread went to press, PPP put out a poll of Iowa, showing Obama 47 / Romney 45, which makes Iowa, combined with the last Rasmussen, poll at perfect +0 tie among the two. That poll is not in the EXCEL document that was just published, for this reason, the link here.

No comments:

Post a Comment

Constructive comments and critique are always welcome. Please keep it polite and respectful.