Wednesday, October 31, 2012

6 Days to Go - Is the Map Contracting or Expanding?, Is Hurricane Sandy This Election's Defining Moment?

Days Until The Election: 6
Projected Popular Vote Total: Obama +0.6% (up 1.0% from yesterday)
Projected Electoral Vote Total: Obama 290, Romney 248 (unchanged from yesterday)
Current Betting Odds: Obama 67%, Romney 33% (Obama up 3% from yesterday)
Current Popular Vote Betting Odds: Obama 56%, Romney 44%

President Obama gains in my popular vote projection today on the tailwind of some strong polls released today.  There is still a large spread across the polls, all the way from +5% for Romney to +5% for Obama.  Clearly, pollsters have divergent views on the composition of the electorate that will actually show up in 6 days.  I thought it would be interesting to look at all of the polls as well as the accuracy of those polls that existed in 2008 to assess which scenarios are most likely.

Note that in my analysis, I exclude partisan-affiliated polls such as Public Policy Polling, as partisan motivations can obscure objectivity of polling, although PPP's poll currently shows an even race, largely in line with the other national polls.

There are a bunch of ways to aggregate this information to produce an average.  Taking a pure mean (adding up all the margins and dividing by the number of polls) yields Obama at +0.5%.  Taking a median (the middle number), yields Obama at +1.0%.  Taking a sample-weighted mean (applying more weight to polls with larger sample sizes) yields Obama at +0.3%.

Looking at the issue of historical accuracy, if we take only the 4 most accurate polls from 2008 (those that called it within 1% of the actual result), we get a mean of 0.0% (an even race) and a median of Obama +0.5%.

So if I distill it down, there are a ton of ways to average this thing, but just about any way you slice it, we are very close to the even line.  Having a polling spread of 10 points is actually not all THAT unusual...the final polls in 2008 had an 8 point spread.  The averaging techniques take the noise out of the system and the wisdom of the crowd usually pays off - consider if you will that if you aggregate all the polls, you are looking at results from over 16,000 surveys and inserting the wisdom of a dozen professional polling firms in determining turnout.

Of course it could all be wrong or change - but it's normally fairly accurate.

Relative to the states, I have been discussing what I view as a contracting battlefield recently.  Today, I think the true battlefield further contracts to 5 states: Florida, Virginia, Colorado, Ohio and New Hampshire.  You could make an argument for Nevada, especially knowing the history of poor polling performance, but bear in mind that poor polling performance has largely been biased against the Democrats - in 2008 President Obama won the state by about 4% more than the margin of the polls, ditto for Harry Reid in 2010, who grabbed victory from the jaws of defeat.

This runs contrary to what the Romney campaign has been saying and to some extent doing - they claim that they have expanded the map and the next 6 states are in play: Nevada, Iowa, Michigan, Wisconsin, Pennsylvania and Minnesota.  And they are spending money on ad buys in those states.

Don't trust necessarily what campaigns are saying - of course the Romney campaign is going to say the map is expanding.  And in an era of virtually unlimited political money, don't necessarily trust the fact that they are placing ad buys.  The 5 "true" battlegrounds are already so saturated with ads that spending more money there doesn't help.  What would be the true sign that the GOP believes they can provide an alternate path to victory for their candidate is if Mitt Romney himself starts appearing in those 6 states.

Of course, Romney is basically off the campaign trail for the time-being in the aftermath of Hurricane Sandy, so we don't know if he will put his time where his campaign's mouth is.  I just can't see him winning any of those states except in the event of an unexpected landslide, in which case they are irrelevant anyway.  They won't provide Romney with electoral vote number 270.

Assuming my scenario is right, Romney now has a must win in 4 out of the 5 remaining battlegrounds and those 4 MUST include Florida, Virginia and Ohio.

It is not an insurmountable task, but it's a tough road for Romney in 6 days.

Particularly in light of:

Presidents Look Presidential in Crises
President Obama is all over the news, looking Presidential, caring and responsive.  New Jersey Governor Chris Christie is lavishing praise on his administration's response, as are local officials in impacted areas. 

Unfortunately for Romney, presidential candidates have very little place in these crises.  He is trying to tour disaster sites, but it looks brazenly political, while the President's visits look useful. 

The news cycle is all about the disaster and the government's response, which has so far been good.

And there is that little clip of Mitt Romney in the primary debates all but saying that FEMA should be dissolved and responsibility sent back to the states.

Can he turn the tide and recapture the news cycle?  His number of days to work with to do so is going to be small.  Expect Hurricane coverage to dominate the week...Mitt will basically have only the weekend and Monday to work his way back in.

While a Mitt Romney victory is certainly not outside the realm of possibility (if you believe in the wisdom of crowds it has about a one third chance of happening), at this point, Obama winning a sound victory, perhaps reaching 332 electoral votes and losing only Indiana and North Carolina from his 2008 coalition, actually seems more likely

Advantage still Obama.

Tuesday, October 30, 2012

The 2012 Big Electoral Map - I Think We Are Down to 6 Real Battlegrounds

Days Until The Election: 7
Projected Popular Vote Total: Romney +0.4% (Romney up 0.3% in the past 2 days)
Projected Electoral Vote Total: Obama 290, Romney 248 (Obama up 9 in the past 2 days)
Current Betting Odds: Obama 64%, Romney 36% (Obama up 1% in the past 2 days)

Not a lot has changed on our map in the past two days (or frankly, the preceding three weeks), but time is running short for either camp to move the numbers.  In today's numbers, Romney moves up slightly in national polling and Obama picks up Colorado again, which has been right near the zero line and flipping back and forth. 

Now it is certainly not impossible in the last week of the election that some profound (or not so profound) news could break that changes the shape of the race very late.  It happened in 2000, when late-breaking news of George W. Bush's youthful DWI conviction appeared to have at least a 2-3 point impact on the national popular vote late as it reinforced the image of him as an immature, irresponsible blue blood.  He won anyway, after a long fight in Florida, but the revelation definitely caused him to lag his prior poll numbers. 

In the last two election cycles, we've seen no such late break in either direction and the actual results on election night mirrored pretty closely the perspective a week before the election.

In 2008, 7 days before voting, I was projection Obama to win the popular vote by 7.4% and win the electoral vote 375-163.  In actuality, 1 week later, Obama won the popular vote by 7.2% and won the electoral vote 365-173.  My "lag 7 days" vote projection was off by 0.2% and at that time I had only 1 state incorrectly called: Missouri (Obama wound up losing Missouri's 11 electoral votes by 0.1% but picked up 1 by winning a Nebraska congressional district.)

Of the closest states, the difference between my projection 7 days before the election and the actual result in the closest states was as follows:
Missouri: Projected: Obama +0.7%, Actual McCain +0.1%           (0.8% error)
North Carolina: Projected: Obama +1.3%, Actual Obama+ 0.3%  (1.0% error)
Indiana: Projected: Obama +1.6%, Actual Obama 1.0%                 (0.6% error)
Montana: Projected: McCain +1.8%, Actual Obama +2.2%           (0.4% error)
Florida: Projected: Obama +3.8%, Actual Obama +2.8%               (1.0% error)
Ohio: Projected: Obama +6.1%, Actual Obama +4.6%                  (1.5% error)

So, without an October or November surprise, none of the 6 closest states were projected more than 1.5% wrong 1 week out and the 5 closest states were all projected within 1%.

Note: You may notice that all of the "1 week out" projections I had in the close states were more favorable to Obama than the actual result.  You might conclude that this is because of some sort of methodology bias on my part, but in actuality, the race tightened in the final days somewhat, as I noted the next day in my blog.

The point of all of this is that barring an October or November surprise this time around, I feel confident that the states where a candidate has a 2% lead or greater in our aggregation methodology will be won by the candidate currently leading.  That still leaves us with the 6 outlined in purple: Florida, Virginia, Colorado, Iowa, New Hampshire and Ohio.

Putting these as "toss-ups", we get an electoral count of 253 for Obama and 206 for Romney, leaving a very narrow path for Romney to navigate.

If Obama wins Ohio, it's over.  271.  Game done.

Ditto that Florida.

If Romney takes both Florida and Ohio, he still has work to do as that just evens the map at 253 apiece.

The map then becomes simple.  The candidate that wins Virginia has to win 1 other state to win.  The candidate that loses Virginia has to get all 3 of the remaining states to win (Iowa, New Hampshire and Colorado.)  Unless Romney takes an elector in Maine, which is improbable but not impossible.  That could create some messy tie scenarios.

Of course, I could be a lot less accurate than last time - maybe the polls I use are a lot less reliable this cycle, maybe my methodology just got lucky, maybe turnout will be way up or down.  Maybe Romney or Obama will have a late surge that does not resemble 2008 (although there is no indication as of now that anything is moving in the race.)  Maybe the hurricane will change something that I can't foresee (although logic would seem to indicate that it would tend to freeze the race where it is rather than move people to one candidate or another.)  Maybe lots of things, but I doubt it.

I think we are down to a 6 state race.  And Romney needs the biggest 3 plus one more to win (or everything but Virginia)

Advantage: Obama.

If you like this site, tell your friends.

Sunday, October 28, 2012

The Big 2012 Map: Stable But Divergent National Polling, Could There Be a Hurricane Sandy Effect?

Days Until The Election: 9
Projected Popular Vote Total: Romney +0.1% (Romney down 0.1% from 4 days ago)
Projected Electoral Vote Total: Obama 281, Romney 257 (Obama up 4 from 4 days ago)
Current Betting Odds: Obama 63%, Romney 37% (Obama up 3% from 4 days ago)

As I alluded to in my shorter write-up yesterday, the numbers really stopped moving after the effect of Romney's first debate performance was fully factored into the race.  The last two Presidential debates and the VP debate, along with all the associated news cycle, are not moving the race much.

Today, we see some positive developments for both sides.  Barack Obama flips New Hampshire back into his column, along with its 4 electoral votes.  On the plus side for Romney, Minnesota has inched closer over the past few weeks and now falls into our "Leans Obama" category, at least theoretically putting it in play.  Neither campaign has events planned in Minnesota that include members of the ticket, although the Obama campaign did recently make a $500K ad buy in the state, although they claim it is in border cities and designed to reach Wisconsin voters, where both campaigns are fighting hard.  I don't seriously consider the possibility of a Minnesota victory for Romney unless the national race breaks hard in his favor in the waning days of the campaign, at which point, it won't be that relevant anyhow.  In other words, Minnesota is unlikely to be a "tipping point" state, as Romney will only win it if he has already comfortably won key states like Ohio and Wisconsin.

The divergence of national poling continues.  On one wing, we have the Gallup poll, which continues to show Mitt Romney with a 5% national lead.  On the other extreme, you have the RAND poll, which shows President Obama up by 6%.  Neither of those results seem particularly likely, as the breadth of other polling shows a much closer national race.  The remaining 6 national polls show somewhere between a 3% lead for Romney and a 2% lead for Obama, with our weighting methodology giving Romney a razor-thin lead in the national vote.

Analyzing the swing states, I find it is instructive to look at several factors - the amount of state polling that has been taking place, the consistency of the results of that polling, the historical reliability of polling, the current betting odds in the state and the investment of resources by the campaigns in those states.

Let's run down the states within 5 points and analyze:
Breadth of polling - very low - the last poll released had a sample end date of October 10th - in other words, we have no poling data from the past two weeks and scant data even before then.
Consistency of polling - low - the last three polls have an 11 point spread - all the way from Romney +9% to Obama +2%
Historical polling reliability - medium - Arizona polling has been about average in terms of projecting actual results - population and demographic shifts happen at a moderate pace and the large non-English speaking population presents some polling challenges
Campaign resource investment: low - neither campaign has events planned in Arizona, meaning they probably believe it is relatively safer for Romney
Intrade Odds - Romney >90% (not enough betting volume to get an accurate exact number)
Prognosis - the campaigns are more likely to be right than not that this one is in Romney's camp, but the scant and divergent polling give Arizona at least some shock potential on election day.  I'd like to see some more polls here.  My degree of confidence in Romney's lead is moderate in the absence of fresh data.

Breadth of polling - high - there have been 10 polls conducted here in the past two weeks
Consistency of polling - high - polls from the past week show only a 3 point spread, from Romney +2% to Romney +5%
Historical polling reliability - medium - Florida's demographics and turnout variability present some challenges, but polling has historically been within a few points of the actual result.
Campaign resource investment: medium - both campaigns are spending a lot on TV in key markets, but campaign events have been decreasing, perhaps recognizing that Florida has been moving to Romney and the tipping point is elsewhere.
Intrade Odds: Romney 74%
Prognosis - my degree of confidence in a small Romney lead is high.

Breadth of polling - high - there have been 6 polls conducted here in the past two weeks
Consistency of polling - medium - polls from this week have a 6 point spread, from Romney +4% to Obama +2%
Historical polling reliability - high - Virginia's demographics are stable, turnout is consistent and poll results generally get it pretty close on election day.
Campaign resource investment - high - Virginia is drawing a lot of top-of-the-ticket campaigning and both campaigns clearly view it as very much in play.
Intrade Odds: Romney 52%
Prognosis - a true toss-up, but likely one where Romney has a slim lead

Breadth of polling - high - there have been 6 polls conducted here in the past two weeks
Consistency of polling - low - polls from this week show an 8 point spread, from Romney +4% to Obama +4%
Historical polling reliability - medium - there are significant transient population factors and turnout variability to contend with in Colorado.
Campaign resource investment - high - both campaigns are investing a lot of time and money in the state.
Intrade Odds: Even (50%/50%)
Prognosis - another true toss-up and one that is hard to call.

Breadth of polling - high - there have been 6 polls conducted here in the past two weeks
Consistency of polling - medium - polls from this week show a 5 point spread, from Romney +1% to Obama +4%
Historical polling reliability - high - stable demographics and consistent turnout make this a relatively easier state to project
Campaign resource investment - moderate - both campaigns are spending in the Des Moines market and dropping in on the way to other mid-west swing states.
Intrade Odds: Obama 67%
Prognosis - it appears likely Obama has a modest lead here

New Hampshire
Breadth of polling - high - there have been 7 polls conducted here in the past two weeks
Consistency of polling - low - polls show an 11 point spread, from Romney +2% to Obama +9% in the past week
Historical polling reliability - low - New Hampshire is demographically stable, but the large contingent of independents have made it very tough to project (think 2008 Democratic primary)
Campaign resource investment - high - for a 4 electoral vote state, New Hampshire is getting disproportionate attention.
Intrade Odds: Obama 62%
Prognosis - very hard to pick a leader here and definitely more subject to a late swing than other states.

Breadth of polling - high - there have been 7 polls conducted here in the past two weeks
Consistency of polling - high - there is only a 3 point spread across the entire two week period, from Obama +1% to Obama +4%
Historical polling reliability - low - Nevada's large transient population, especially in Las Vegas, makes this a very hard state to call - most of the polls had Sharon Angle beating Harry Reid by a few points on election eve 2010 and got it wrong.
Campaign resource investment - medium - there is a lot of TV spend but few campaign visits, perhaps as Nevada is not particularly close to the rest of the battlegrounds.
Intrade Odds: Obama 83%
Prognosis - all the evidence seems to suggest a consistent Obama lead, but historical inaccuracy of Nevada polling gives some pause.  Bias has tended historically to be Pro-Republican in polling, so I still give Obama the edge here.

Breadth of polling - very high - there have been 14 polls conducted here in the past two weeks
Consistency of polling - high - of all 14 of those polls, there is only a 5 point spread, from even to Obama +5%
Historical polling reliability - high - Ohio has stable demographics and fewer independents than most swing states - polling tends to be highly reliable.
Campaign resource investment - very high - 4 of the 10 most spent markets for TV are in Ohio and it is the most-visited state by both Presidential and VP candidates.
Intrade Odds: Obama 68%
Prognosis - Obama clearly has a small but meaningful lead here in my opinion.

Breadth of polling - medium - there have been 5 polls conducted here in the past two weeks
Consistency of polling - medium - there has been a 6 point spread among the 5 polls, from even to Obama +6%
Historical polling reliability - high - similar to Ohio in many ways, pollsters generally get in right in Wisconsin.
Campaign resource investment - high - Romney has upped his game here and both campaigns are making lots of stops and ad buys.
Intrade Odds: Obama 72%
Prognosis - Obama seems to hold a small but meaningful lead here, but I would like more polls to validate.

Breadth of polling - high - there have been 7 polls conducted here in the past two weeks
Consistency of polling - high - over the entire two week period, the spread is only 4 points, from Obama +3% to Obama +7%
Historical polling reliability - high - Pennsylvania fits the same mold as Ohio and Wisconsin
Campaign resource investment - low - Romney does not appear to be making a serious play for Pennsylvania
Intrade Odds: Obama >90% (volume too low to get accurate odds)
Prognosis - Obama seems highly likely to win here, barring some major national shift.

Breadth of polling - medium - there have been 4 polls conducted here in the past two weeks
Consistency of polling - medium - there is a 7 point spread in the polls, from Obama +3% to Obama +10%
Historical polling reliability - medium - Minnesota has some of the independent streak of New Hampshire (see Jesse Ventura), leading to some late shifts that are hard to poll for, but in aggregate polls have generally been about as accurate here as nationally.
Campaign resource investment - low - Obama's small ad buy aside, there isn't much going on here.  Romney seems to be conceding this one to the President.
Intrade Odds: Obama>90% (volume too low to get accurate odds)
Prognosis - Obama appears highly likely to win here.

Based on all of this, in my estimation, Arizona and Florida appear highly likely to go to Romney, Minnesota and Pennsylvania seem highly likely to go to Obama and Virginia, Colorado, Iowa, New Hampshire, Ohio and Wisconsin are the 6 states where the final days of the election are likely to be fought.

If we give Romney Arizona and Florida and give Obama Minnesota and Pennsylvania and apply the intrade odds to the remaining states, we get a base of 237 electoral votes for Obama and 235 for Romney.  Applying Intrade odds for the remaining states in a random model, we get the following outcome:

Obama Wins Electoral College: 74%
Romney Wins Electoral College: 24%
Electoral College Tie (Likely Romney Victory in House): 2%

The possibility of a tie means that we cannot ignore Maine and Nebraska, the two states that split their electoral vote by congressional district.  Last time around, Obama won 1 of Nebraska's 5 electoral vote by taking its most moderate district.  He appears to be well behind this time.  This time, Romney is hoping to take 1 of Maine's two congressional districts, the 1st district.  He appears to be behind as well, but within striking distance, possibly trailing by only 3% or 4%.  This will only matter in a very, very close electoral college, but as you can see from this model, that is a possibility.

Hurricane Sandy Effect?
It's a good thing that Hurricane Sandy is projected to hit this Monday and Tuesday instead of next Monday and Tuesday.  If it had hit next week, it would have had a dramatic impact on voter turnout and created a major disruption at just the wrong time to people making an important election choice.

As things stand, the effect will likely be very little unless there are still major power outages on election day.  While this is possible, it appears unlikely, as utilities will have a month to get the power back on.  Stay tuned - I'm right in the path of the storm in Southern New Jersey, so if you don't hear from me for a while, it's because I don't have power to post my blog.

If you like this site, tell your friends.

Saturday, October 27, 2012

Could We Be Headed for 2000 All Over Again?

As I watch the new polls roll in each day, I'm noticing an increased stability in the race.  If I could distill down what is going on to a few major events in this year's campaign, it would be as such:
(1) Out of the Primaries - Solid Obama Lead
Coming out of the Republican nominating contests, once Mitt Romney secured the GOP nomination, he trailed President Obama nationally by a little over 3%.  Our map from back in June showed the President at 3.2% with a solid electoral college lead of 332-206.  Basically, the President was leading in almost everything that was expected to be a swing state - he led modestly in Florida, Virginia and New Hampshire and trailed only modestly in North Carolina.  It basically looked like Mitt Romney was going to have to run the table everywhere to win and even that might not get him to victory.

(2) Paul Ryan is Selected
While little has been written about it in the mainstream media of late (the media has since moved on to the next narrative), the selection of Paul Ryan as Romney's running mate re-energized the GOP ticket and tightened the race.  By late August, our map showed Obama leading by a mere 1.1% and his electoral advantage tightened to 294-244.  Romney had picked up Florida and Colorado and was within striking distance in Virginia and closing in Ohio and Wisconsin.

(3) Bill Clinton Stomps Clint Eastwood
Do you remember the most memorable moments from the RNC?  Clint Eastwood talking to an empty chair.  The most memorable moment from the DNC?  Bill Clinton picking apart the GOP platform with a skill that only the master has.  The poll results showed a big bounce for President Obama coming out of the two conventions with the President's national lead at +5.0% in early September and an electoral lead of 332-206, identical to his lead back in June.

(4) Romney Owns Obama in Debate #1
If Mitt Romney finds a way to win this election, the President's team will have a single event to look back on in the post-mortem, the way he got utterly devastated in the first presidential debate.  100 million people watched a disinterested, disengaged President get obliterated by an energized, suddenly more moderate Mitt Romney.  The polls quickly reflected this reality.  By early October, Mitt Romney was up 1.2% in national polling and had closed the electoral gap to 281-257, having retaken the lead in all of the solid south as well as Colorado.

And that is pretty much where we have stayed.  The second and third debates failed to move the needle in either direction as did the VP debate.  Mitt Romney continues to sport a modest lead in national polling (a somewhat larger lead in Gallup and Rasmussen, trailing in the Investor's Business Daily Poll and the Ipsos/Reuters poll, but leading modestly in aggregate) and trailing slightly in the electoral college. 

Of the 7 events that I said could change the race (Romney's VP selection, the two conventions and the four debates), it is clear that 3 did (Romney's VP Selection, the DNC and the first debate) and four had a less measurable impact, if any.

There are no major scheduled events left between now and the election in 10 days.  The race has more or less been stable since early October.  It would seem we are where we are unless an October (or November) surprise breaks in the next week and a half.

But this does remind me of another race where there was discussion of a Republican candidate leading in the national polls but a Democrat potentially winning the electoral college - 2000.

Going into the election, many believed that split result was the most likely outcome based on the polling data.  Of course, the reverse wound up happening, very narrowly.

Now I wasn't in the prognostication business back then and I like to think we've learned something about polling and analysis in the intervening 12 years.  But is it close enough that another Year 2000 surprise could be in the offing?  It certainly isn't impossible.

If you like this site, tell your friends.

Wednesday, October 24, 2012

The Big 2012 Electoral Map - Why It's Really Close and Why I Believe Obama Will Probably (But Not Certainly) Win, Foreign Policy Yawn, The Split Vote Scenarios

Days Until the Election: 13
Projected Popular Vote Total: Romney +0.2% (up 1.3% in the past 4 days)
Projected Electoral Vote Total: Obama 277, Romney 261 (Romney up 9 in the past 4 days)
Current Betting Odds: Obama 60%, Romney 40% (Romney +1% in the past 4 days)

It's close.  It's very close.  Certainly closer than 2008.  Whether it winds up closer than 2004, when President Bush won re-election on the basis of a 2.1% margin in Ohio (New Mexico and Iowa were actually narrower, but winning both of those would not have put John Kerry over the top) remains to be seen.  It would require a rare twist of fate for it to be as close as in 2000, when President Bush won his first term by officially less than 0.01% of the votes cast in Florida.

But it's close.  Close enough that even if this were election eve, I wouldn't be 100% confident that my model would correctly project the winner.  Certainly close enough that with nearly two weeks left until the election, I can't predict with any certainty.

But President Obama is still ahead.

Consider the case for Obama's re-election:
(1) President Obama has not trailed on my electoral map the entire election cycle.  Not for a day.
(2) President Obama has not trailed in the betting odds on Intrade once.  Not for a day.
(3) The electoral map favors Obama - even when he trails modestly in the national polls, he maintains his electoral lead.

I'm not saying Romney couldn't win.  He could peak at exactly the right time - on election day.  Democratic voter turnout could be much lower than the pollsters are modeling.  We could still have an October surprise that could wildly swing the race. 

But in spite of Romney's surge following the first debate, Obama is still where I would place my bet today.  We'll see if I'm singing a different tune in two weeks.

What should give Republicans some encouragement is that Romney's paths to 270 have clearly widened.  Ohio is still by far the easiest path and expect that to continue to be the most fought over battleground.  But Romney could also win by holding what he has (which now includes a razor-thin lead in Colorado) plus picking up Iowa and Nevada.  Or he could lose all 3 of those states but win by picking up Wisconsin.  Or, in a very low shot, he could pull off a huge upset in Pennsylvania.  But all those paths are less probable than the one showing in the chart above.

Who Schedules These Debates?
What the heck was the commission on Presidential debates thinking?  Foreign policy as the theme for the closing debate?

I'm not saying that foreign policy doesn't matter.  Certainly how we deal with trade with China, Iran's nuclear ambitions, Israel's place in the middle east, Iraq, Afghanistan and a whole host of other issues matter.

But recent polling indicate that 6% of voters consider foreign policy their top priority in this election.  6%.  Why would we make the final debate that voters see something that only 6% care about.  How about a debate solely on the economy?

And what a horrible set of questions on foreign policy.  How can you have a foreign policy debate without a single question on Mexico, our immediate neighbor that has been ravaged by cartels?  Not a single question on the Eurozone crisis, the single greatest threat to the world economy?  90 minutes on a topic virtually no voters prioritize and they don't even ask the most important questions.

At any rate, the poor topic choice essentially made the debate of very minimal impact.  Polling indicated that President Obama modestly won the debate in the eyes of the majority, including myself.  But it won't be a poll mover - Romney passed the bar as Commander-in-Chief with his command of the facts on the stage and beyond that, few will be swayed by competing answers on Libya.

Split Scenarios
Most Presidential elections end the same way - one candidate gets the most popular votes and wins the electoral vote.

In very close races, however, interesting scenarios emerge.

(1) The Popular Vote and Electoral Vote Split
This is quite a plausible scenario this year.  I show Mitt Romney up by 0.2% and President Obama up in the electoral college.  Even Romney were up 2%, assuming that 1.8% movement was evenly spread across the country, he would pick up only Iowa and would still lose the electoral vote 271-267.

I am personally licking my lips at that prospect.  Following the split result in 2000, when President Bush lost the national popular vote but narrowly (quite narrowly) won the electoral vote, there was an outcry on the left for electoral reform.  Since then, 9 states, all blue states, representing 132 electoral votes have passed laws enacting the National Popular Vote Interstate Compact, which would award that state's electoral votes to the national winner if states representing 270 electoral votes sign on, meaning that if states totaling 270 electoral votes signed onto the compact, the winner of the national popular vote would win the election.

The reason no Republican state has signed on is obvious - the electoral college was a structural advantage to the GOP in 2000.  If that ceases to be the case, we would have a shot at real electoral reform.

Wouldn't it be nice if candidates had to campaign somewhere besides Ohio, Florida and Virginia to win an election?  Shouldn't there by campaign rallies in Boston, Chicago, Los Angeles, Dallas, Atlanta and all the other major cities that get shunned each year because they aren't in swing states?

On a related note, to validate to divergence of the state and national polls, I ran my current state projections through the 2008 vote model to see how the national vote might come out if the state projections are right.  The results of that were that if all of my state predictions were exactly correct and turnout exactly mirrored the 2008 election, President Obama would win by 0.8% in the national popular vote, or a 1% divergence from my current actual popular vote projection.

This is well within the margin of error and also bear in mind that for states that are not close, the projection is based on scant polling data, increasing the propensity for error in those states.  A 1% divergence fundamentally confirms my projection:
* The national vote is close to even
* President Obama is ahead in the electoral college

If you like this site, tell your friends.

And if you think my predictions are full of it - bet on Mitt Romney on Intrade - you can get good odds.

Saturday, October 20, 2012

The 2012 Map: Still Advantage Obama Despite Gallup Poll Results, Celebrating Milestones

Days Until the Election: 17
Projected Popular Vote Total: Obama +1.1% (Obama up 2.1% since last week)
Projected Electoral Vote Total: Obama 286, Romney 252 (Obama down 8 since last week)
Current Betting Odds: Obama 61%, Romney 39% (unchanged since last week)

The fundamental situation in the Presidential race didn't change in the past week and that's bad news for Mitt Romney, as, in my estimation, he is still behind in the electoral college.  Romney picked up a very narrow lead in Virginia and New Hampshire in my projection and lost Colorado, netting him 8 electoral votes, but lost ground in the popular vote polling.

First, a few notes on the popular vote.  The Gallup poll has been the subject of a lot of press of late, with it showing a 7% lead for Mitt Romney on Thursday, a 6% lead for him on Friday and a 6% lead again today.  Gallup is among the most established and respected polling firms in the country, so it would be foolish to ignore their results, but they do cause me to scratch my head in light of the other polling results that we are seeing from established tracking polls.  The Investor's Business Daily poll (which many have pointed out over the past 8 years was the closest to projecting the outcome in 2004, although it was mid-pack in accuracy in 2008) shows Obama up by 2%.  Rasmussen, which has long been accused of being a right-leaning polling firm, shows Romney up by 1%.  The Reteurs/Ipsos tracking poll shows Obama up by 3%, Rand has Obama up by 3% as well and Romney is up 2% in the UPI poll.

In short, of the 6 tracking polls covering the race, 5 of them have the range somewhere between Obama +3% and Romney +2% and Gallup has Romney at +6%. 

Additionally, several non-tracking national polls came out in the past week that validated the range of the other 5 tracking polls - a Battleground poll this week showed Obama +1%, an ABC News/WP poll showed Obama +3% and a Hartford Courant poll showed Obama +3%.

Clearly, one of two things is going on:
(1) The Gallup Sample is dramatically different from the other polls in terms of their assumption around likely voters
(2) Gallup had a few days which were just a statistical anomaly that will even itself out over the course of the next week

In the case of #1, we would have to determine which likely voter model we believe in order to determine if Gallup is right or if the rest of the polling universe is right.

In the case of #2, the effect of weighting and averaging the polls that I use in my statistical model would largely take care of the outlier.

So is the Gallup poll dramatically different in sample selection?  It does not appear so from its internals.  One of the things that is observable is that the Gallup poll appears to be a lot less stable than other national polls which leads me to believe that they normalize their data a lot more.

Bear in mind, the Gallup poll has a margin of error at a 95% confidence interval of +/-3%, meaning that 6% is actually within the margin of error (if you add 3% to Obama's total and subtract 3% from Romney's, which is the correct way to measure these margins, you get an even race.)

As far as my projection goes, I see no reason to discount the results from the Gallup poll in my measures but also no reason to give it more weight than any other national poll.

At the state level, I think it is safe with scarcely more than 2 weeks and 1 debate to go to narrow the field to states that truly have a real chance of switching.  Let me do the rundown of all the competitive states:
All the solid/strong states - assume these are gone.  Indiana, South Dakota and Missouri are out of reach for the Democrats - Obama won't even try for them.  Ditto that New Mexico, Washington and Connecticut for the Republicans.  So, let's start by striking the left and right hand columns from consideration.

Of the states within 10 points, here is my rundown of the real prospects:
Montana - nobody is spending any money here or campaigning here and Obama lost it last time.  I consider it out of reach for him.
North Carolina - the Obama campaign is cutting back on media buys here and isn't making campaign stops.  I think the crown jewel of his 2008 victory and the site of the 2012 DNC is gone for the Dems.
Arizona - a state that was utterly uncompetitive in 2008 given John McCain's home field advantage keeps flirting with being competitive.  But Obama isn't campaigning or spending much there.  I think this one will stay in GOP hands.

Florida - still hotly contested by both parties with tons of ad spending and candidates criss-crossing the state.  This one is still up for grabs.
Virginia - both campaigns are fighting hard here and it is extremely close.  Still up for grabs.
New Hampshire - given it only has 4 electoral votes, I'm shocked at how much both campaigns are investing here.  Definitely still competitive.
Colorado - among the closest in the nation and still very much up-for-grabs.  A large Mormon population helps Romney but a large Hispanic population helps Obama.
Wisconsin - I will leave this on the competitive list since the state still appears very close in the polls.  It doesn't seem that the Romney camp is fighting too hard for this one though.
Ohio - definitely the lynchpin of the campaign and one fought heavily over by both sides.  Obama has a small but clear edge here, but this is still competitive.
Nevada - similar demographics to Colorado create a similar campaign dynamic.  Also, the hard-hit economy makes it a prime Romney target.  Still competitive.
Iowa - still up for grabs and attracting campaign dollars and visits.

Pennsylvania - the prospect of winning here for the first time since 1988 is quite a sexy idea for the GOP, but they aren't investing heavily here and I think this one stays Democratic unless Obama collapses in the next couple of weeks.
Michigan - this one once appeared competitive but keeps slipping away for the GOP.  It will stay Democratic.
Oregon - Romney is investing nothing here.  It stays in Obama's column.
Minnesota - same as Oregon, Minnesota is competitive in theory only.  It will stay Democratic.

So we are left with a battlefield of 8 states: Florida, Virginia, New Hampshire, Colorado, Wisconsin, Ohio, Nevada and Iowa.

Obama starts with a base of 237 Electoral Votes, Romney 206.
The true battlegrounds have the following electoral votes (in rank order):
Florida - 29
Ohio - 18
Virginia - 13
Wisconsin - 10
Colorado - 9
Nevada - 6
Iowa - 6
New Hampshire - 4

Let's look at the path's to victory for each candidate:
For Obama:
(1) Win Big
The easiest path is win the two biggest prizes.  If he takes Florida and Ohio, it's over.  He gets 284 electoral votes.

(2) Win Easiest
The states where he holds the strongest leads are Wisconsin, Ohio, Nevada and Iowa.  They have a collective 40 electoral votes, which would give the President 277 electoral votes.  So, he could even afford to lose either Iowa or Nevada and still have 271 electoral votes.  Wisconsin, Ohio +1 seems like Obama's easiest path

For Romney:
(1) 3 Biggies Plus 1
Florida, Ohio and Virginia would get Romney to 266.  Win any other state and he is at 270.  New Hampshire would seem like the easiest state to claim in this electoral map.  Maybe that's why he is investing there.

(2) Win the Closest
If you go down the rank order, to get to 270, he needs Florida, Virginia, New Hampshire, Colorado and Wisconsin to put him over the top at 271 electoral votes.

So what strategy are the campaigns investing in?

Both candidates have been criss-crossing Florida and Virginia, states that Romney will need in both strategies and ones that Obama needs in strategy #1.

We'll see where they spend time as the campaign closes.

Site Milestones
My last post was my 500th on this blog.  September was the most-read month ever for this site and October is only 100 viewers away from surpassing September, something it seems almost sure to do today.

I'm hardly Nate Silver in terms of fame in projecting elections, but I would note that we were almost as good as Nate in 2008 (this site predicted 48 out of 50 states correctly, Nate picked 49 of them right.  We called the popular vote exactly correctly as well.)  But it's nice to know that thousands of people every month are reading and enjoying what I do.

Thanks for reading and for telling your friends about us.

Sunday, October 14, 2012

The Big 2012 Elecotral Map - A Stable Picture of a Changed Race, Should Obama Abandon North Carolina and Florida?, Veep Candidates Clash, The Battle for Congress

Days Until the Election: 23
Projected Popular Vote Total: Romney +1.0% (down 0.2% from last week)
Projected Electoral Vote Total: Obama 294, Romney, 244 (Obama up 13 from last week)
Current Betting Odds: Obama 61%, Romney 39% (Romney up 1% since last week)

As the aftermath of the first Presidential debate settles down in the polls, we see the "new normal" for the race beginning to stabilize.  Romney actually lost a little ground this week, with his national polling falling back fractionally and Virginia flipping back to Obama.  But the basic picture is still the same - a pick 'em race nationally with a slight structurally electoral advantage to Obama.

Romney needs to add a minimum of 26 electoral votes from here to win the Presidency.  If he picks up the 3 closest states (New Hampshire, Nevada and Virginia), it leaves him just short at 267.  This means that Romney still needs one of the larger states - either Ohio, Wisconsin or Pennsylvania, or to pick up those 3 plus Iowa.

The betting odds continue to tighten a little but still show an Obama advantage, largely, I believe, because of this structural electoral advantage that we have been discussing for some time.

Should Obama Bail on North Carolina and Florida?
The Democratic National Convention was held in Charlotte and North Carolina holds a special place on the map for Barack Obama as a state he'd like to have.  Florida obviously holds special significance for Democrats as the site of the epic recounts 12 years ago.

But, as a matter of strategy, if I were advising President Obama, I would urge him to abandon his campaign in North Carolina and Florida in the waning days of the campaign.

Sure, winning one of those two states would seal the deal for a second term.  But they seem to be slipping out of reach and he doesn't need them.

Certainly, I would contest Colorado, which is basically just one media market and still seems very winnable.  Other than that, I'd focus on holding the states with leads.  As I described above, Romney, if he takes Florida and North Carolina, still needs New Hampshire, Nevada, Virginia, Colorado and 1 other state.  Firewall Ohio, Wisconsin, Iowa, Pennsylvania and you win the Presidency.  Make Romney feel heat by rallying Hispanic voters in Colorado and Nevada.  Make the fight take place on your turf and focus resources.

Likewise, if I were advising Romney, I'd be telling him to pour everything into Ohio and Wisconsin.  Those are two big states that appear "gettable".  Michigan looks like too far a reach.  I think Pennsylvania is a long shot.  Nail Ohio and Wisconsin and your paths to the Presidency are many.  Miss them both and it is an almost impossible map.

VP Candidate Debate to an Essential Draw
The VP debate is typically the least significant of the four national debates in terms of moving the polls and that certainly appears to be the case this year - in part because almost everyone is voting based on the top of the ticket and in part because, in my opinion, Joe Biden and Paul Ryan essentially fought to a draw.

Biden was very good on substance - he was quick on his feet, aggressive in countering Ryan's attacks and came across likable, as he virtually always does.  He has received some criticism for smirking and laughing during Ryan's responses, but I don't think in context that those responses will hurt him.

Ryan appeared capable, cool and collected.  He was also aggressive on the attack and showed credibility and understanding on foreign policy. 

In short, I don't expect that the VP debate will do much to change the race.

Tune in next week for the second Presidential debate, where it is essential for Obama's chances that he significantly outperform his first debate performance.  Look for Romney to be aggressive to maintain / support his positive momentum.

The State of the Congressional Races
It's been a while since I've looked at the state of the races.  It appears more likely than not that the Democrats will retain control of the Senate at this point and that Republicans will retain the House.  Here are the latest numbers:

In the Senate, there are 30 Democratic and 37 Republican seats that are not up for election.  Of the balance, here are where things stand:

Safe or Likely Independent Seats - 2
(both likely to caucus with Democrats)
Vermont, Maine

Safe or Likely Democratic Seats - 12
California, Maryland, New York, Delaware, Minnesota, Rhode Island, Hawaii, Michigan, New Jersey, Washington, West Virginia, New Mexico

Safe or Likely Republican Seats - 6
Mississippi, Tennessee, Texas, Utah, Wyoming, Nebraska

Close Races - Projected 10 Democrats, 3 Republicans
Florida - Nelson (D) +7.6%
Ohio - Brown (D) +5.6%
Missouri - McCaskill (D) +5.2%
Pennsylvania - Case (D) +5.0%
Connecticut - Murphy (D) +4.2%
Wisconsin - Baldwin (D) +3.0%
Indiana - Donnoley (D) +2.0%
Massachusetts - Warren (D) +1.8%
Arizona - Carmona (D) +1.5%
Virginia - Kaine (D) +1.0%
North Dakota - Berg (R) +0.1%
Montana - Rehberg (R) +0.5%
Nevada - Heller (R) +3.0%

Projected: 52 Democrats, 46 Republicans, 2 Independents
(effective control 54-46 Democratic)

So, Democrats have the lead, but also have more close races to defend that they are currently leading.  Republican control of the Senate is not impossible, but looks to have become increasingly unlikely as the races have played out.

On Intrade, the odds of Republicans winning at least 50 Senate seats is currently pegged at 34%, and keep in mind that 50 seats only gives them the majority if they also win the Vice-Presidency, otherwise they would need 51 to get control.

In the House,
Current generic polling has the Democrats at +1.3%.

Based on this, projecting based on the newly redistricted House (which structurally favors the GOP) would give us:
Republicans 220 Seats, Democrats 215 Seats

I don't generally do seat-by-seat analysis of the House, but other sites do, so here is there perspective: (splitting the toss-ups evenly): GOP 239, DEM 196 GOP 240, DEM 195

Republicans have a 90% chance of retaining the House, based on the latest Intrade odds.

Obviously, the seat-by-seat analysis yields a broader spread than the generic polling data would indicate.  This may well be the case because of the candidates in the close races.  But I'm inclined to believe the truth is somewhere in between.  Either way, the GOP appears well-poised to retain the House.

If you like this site, tell your friends.

Tuesday, October 9, 2012

Oh How The World Can Change in 2 Weeks

Days Until the Election: 28 (that's right, just 4 short weeks!)
Projected Popular Vote Total: Romney +1.2% (Romney up 2.8% since 2 weeks ago)
Projected Electoral Vote Total: Obama 281, Romney 257 (Romney up 66 since 2 weeks ago)
Current Betting Odds: Obama 62%, Romney 38% (Romney up 8% since 2 weeks ago)

Ladies and gentlemen, we have a horse race. 

On the heels of the one of the most one-sided debate performances of my lifetime, the complexion of the 2012 Presidential race has shifted significantly.

Romney has surged to a narrow lead in the national polls and while he still trails in the electoral total, his paths to 270 suddenly seem numerous.  He could win the states he leads plus Ohio, where he is a mere 0.4% behind.  Alternatively, he could pick up Pennsylvania or Michigan and win it all.

Now, let's be clear - there is good reason why President Obama is still the betting favorite to win re-election despite trailing in the national polls.  The structural advantage he holds in the electoral college make Romney's path tougher even if he wins nationally.  Romney still has to hold all 5 of his close states plus take one of the big 3 others and I'm not candidly sure that Pennsylvania and Michigan are realistic targets unless Romney posts a 5%+ national lead (at which point the electoral map almost becomes irrelevant as he will win in almost any imaginable scenario.)

So Obama can basically play 3-state ball - go after Florida, Virginia and Ohio and all he needs to do is take 1 of the 3.

Plus, there are still 3 more significant events in the political season...the 3 remaining debates.  If we review the 4 that have occurred so far from my list of 7, it has been a mixed bag:
1.  Selection of Republican VP Candidate - minor boost for Romney
2.  Republican National Convention - weak or zero boost for Romney
3.  Democratic National Convention - moderate boost for Obama
4.  First Presidential Debate - moderate-to-large boost for Romney

I find it almost inconceivable that Obama will turn in 2 more performances as flat as the one last week or that Biden vs. Ryan will go as poorly as the first Presidential debate.  Romney doesn't need victories that large, but I think even 3 even debates is probably a big plus for the Obama camp since it takes some of the luster off Romney's first victory.

Obama is still correctly the betting favorite.  But certainly not as big a favorite as he was two weeks ago.  On to the VP debate.

If you like this site, tell your friends.