Thursday, October 18, 2012

Romney's big lead in the Gallup Poll

On Wednesday (10/16/12) Gallup reported a large 50%-46% Romney advantage over Obama in a national poll, much to the excitement of Romney supporters. Although reflecting a national lead for Romney, the Gallup result is aberrant because other polls show the race somewhat tighter. For instance, Rasmussen has Romney up 2% while Ipsos/Reuters has Obama up 3%.

All three of these polling agencies are reputable, so the reality is probably somewhere close to the average. Indeed, Pollster has Romney up 0.4% nationally (47.2% to 46.8%) while Polltracker has Romney up 1.1% (48.4% to 47.3%). Both Pollster and Polltracker use all possible polls to estimate support among likely voters. As a result, we expect Pollster and Polltracker estimates to be more accurate than just the individual polls, like Gallup, Rasmussen, and Ipsos/Reuters.

But for the sake of understanding what a best-case scenario might mean for Romney, let's assume the Gallup Poll released 10/16 reflects what will be the "true" support for the candidates on election day.

The geographic breakdown from this poll tells the story for Romney:
  • East: Romney down 4%
  • Midwest: Romney down 4%
  • South: Romney up 22%
  • West: Romney down 6%

This breakdown reveals that Romney is solidifying his support among those who were already leaning in his favor (the very red South). That's good for Romney's campaign, but a deeper look at the poll shows that he still struggles in swing-states, wherein all hopes of victory reside.
 
For this election, only two swing states are in the South: North Carolina, which seems to be in the bag for Romney, and Florida, which leans to Romney by less than +2%. In other words, six of eight swing states are in areas of the US where Obama has a solid lead.

The certainty of Romney's Southern support means that Romney could be up 45% in the South and it wouldn't affect the results of this election—although it would give him an eye-popping lead in a national poll like Gallup.

Chances for Victory Update

Here are the Chances for Victory if the election were held on 10/19/12:
  • Obama - 99.35% (303 Electoral College votes most likely)
  • Romney - 0.64%
  • Tie - About 1 in 22600

Chances for a tie in the Electoral College.

Based on last night's debate, I believe that today is the closest this race will be—unless the third debate or some unpredictable event fundamentally changes the dynamics. This belief, along with some speculation from around the web about the possibility of a tie in the Electoral College, has prompted me to calculate exactly how much we should worry about a tie between Obama and Romney.

The election currently has nine swing states:
  1. IA - Iowa
  2. NV - Nevada
  3. OH - Ohio
  4. WI - Wisconsin
  5. NH - New Hampshire
  6. CO - Colorado
  7. VA - Virginia
  8. FL - Florida
  9. NC - North Carolina
I list these states in the order that they are likely to go to Obama. For example, Iowa is more likely to go to Obama than Nevada, and Nevada is more likely than Ohio, etc.

Having nine swing states means that there are exactly 512 different scenarios for the election, most of them highly improbable. An example of an exceptionally improbable scenario is one wherein Obama wins North Carolina, Florida, and Virginia while Romney wins the rest of the swing states (1 through 6).

Of the 512 scenarios, exactly five will result in ties. In order from most to least likely, these scenarios are where (from the swing states) Obama wins only:
  1. NH, VA, CO, IA
  2. NH, WI, OH
  3. NH, VA, CO, NV
  4. VA, CO, WI
  5. NH, VA, NC
So only 5 of 512 scenarios result in a tie. That's less than 1% of the scenarios. Note however that these are very unlikely scenarios. Why are they unlikely? Take the most likely tie scenario as an example. In this scenario, Obama would have to win Virginia (VA), where he is currently trailing, but lose Nevada (NV), where he currently has a nearly sure chance of winning.

Because these scenarios are unlikely, the chance of a tie is much less than 1%. In fact, the chance for a tie is 50 times less than 1% (or 0.02%). Put another way, that is only a 1 in 5000 chance of a tie in this election as it currently stands.

Chances for Victory Update

Here are the Chances for Victory if the election were held on 10/18/12:
  • Obama - 98.6% (290 Electoral College votes most likely)
  • Romney - 1.4%
  • Tie - 0.02%

Tuesday, October 16, 2012

Impact of the second presidential debate on the election.

To understand the influence of the second Presidential debate tonight, we should begin by assessing the impact of the first debate on October 3.
  1. Mitt Romney was declared the winner of the first debate by the media and seemingly by public opinion. This win improved Romney's national polling averages by at least 3%. At the same time, Obama has lost about 2%. Inspection of the Pew poll taken after the debate reveals that the movements were largely a consequence of a flip of enthusiasm, with Obama's supporters losing enthusiasm while Romney supporters gained it. These changes in the polls resulted in Romney's ultimately enjoying about a 1.5% lead in national polling among likely voters. But it is important to note that the swing didn't come principally from undecided voters but from voters who already have an allegiance to their candidates.
  2. Stylistically, Romney gave a solid performance in the first debate while Obama was somewhat lackluster, probably in an attempt to avoid confrontation which might have alienated undecided voters. While Romney's better stylistic performance helped him significantly in the polls, it also burdened Romney with higher expectations in the second debate. In other words, he had few Republican voters to motivate in the second debate because they were already motivated.
Thanks to these points, it is not necessary to declare a winner to predict how tonight's debate affects the general election. Instead, I only have to estimate how each candidate's performance is perceived by his own supporters.

First, both candidates had solid performances, with the only noticeable stumble coming from Mitt Romney, who appeared to misunderstand the content of Obama's statements in the aftermath of the Benghazi attacks. Given that Romney's performance tonight was no stronger than his previous performance, it is unlikely that he gained any support that he didn't already have. The Benghazi stumble rattled Romney a little, but was not catastrophic in the eyes of his supporters. So it is unlikely that Romney will see much of a drop in the polls in the wake of tonight's performance.

Second, Obama was on attack and he hit every key talking point, like Romney's "47%" comments and some of Romney's stated positions on women's issues. While Obama's attack strategy may not guarantee his gaining of many undecided voters, it is very likely that he did charge up his own supporters. For example, the NPR snap poll, which has a strong liberal bias, had the President winning tonight's debate 90% to 10%. This poll indicates that Obama overwhelming pleased his supporters tonight. Consider that the NPR snap poll after the first debate gave Romney a slight advantage.

With these observations in mind, it's straightforward to predict how tonight's debate will influence the polling and consequently our expectations for the general election: Obama should gain the 2% back that he lost after the last debate plus about 0.5% that Romney will lose for (1) his stumble over Benghazi and (2) for Obama's serendipitously getting the last word and using it to jab Romney with reminders of the "47%" comments.

This predicted 2.5% swing in the poll numbers will likely solidify both Pennsylvania and Ohio for Obama and push both Colorado and Virginia to lean towards Obama over the next week. Florida may even come back into play.

Chances for Victory Update

Here are the Chances for Victory if the election were held on 10/17/12:
  • Obama - 96.9% (290 Electoral College votes most likely)
  • Romney - 3.1%

Colorado Flips and Ohio Moves Towards Dark Blue

Monday (10/15/12) was the first day where we could see some of the aftermath from Joe Biden's crushing Vice President debate performance on Thursday. Most of the action was in Ohio, where the pollster PPP reported a 51% to 46% advantage to Obama in a poll taken over the weekend.

Other national polls that involve polling after the Vice President debate have driven some changes that affect the swing state map. Most notably, Colorado and Virginia are likely in the midst of moving back into Obama's favor.

While Romney still has about a 1.5% advantage in the general electorate, these small changes in the swing states improve Obama's chances significantly. Obama is poised to make a decisive comeback in Virginia and Colorado if he can muster a strong debate performance tonight. With Ohio in his pocket for now, Obama has a rare second opportunity to solidify his chances for reelection.

Tonight's debate promises to be one of the most important in history.

Here's the bottom line chances for victory if the election were held on 10/16/12:
  • Obama — 99.1% (290 Electoral Votes)
  • Romney — 0.9% (248 Electoral Votes)

Sunday, October 14, 2012

Introduction to the Chances of Presidential Victory

Almost two weeks ago, Barack Obama led the 2012 general election for President of the United states with a virtually unassailable advantage in the most important swing states:
  1. Pennsylvania
  2. Ohio
  3. Wisconsin
  4. Colorado
  5. Virginia
  6. Florida
  7. North Carolina
On October 3rd, Obama met his Republican challenger, Mitt Romney for the first of three Presidential debates. While one's judgement of the winner of these debates is greatly influenced by personal perspective, Mitt Romney was declared the victor by the media. In the subsequent 10 days, we have seen Mitt Romney surge from roughly 44.5% support among likely voters to close to 48%. At the same time, Obama's support has fallen from over 48% to nearly 46%. A difference of about 1.5% separates the candidates today (Saturday 10/13/2012).

More importantly, Romney now leads in Colorado, Virginia, Florida, and North Carolina. With leads in these states, Mitt Romney's easiest path to victory is to take all four of these swing states along with three in which Obama leads: Nevada, New Hampshire, and Iowa. To say that Obama "leads" in these latter three states means that Obama has a better chance to win these states than Mitt Romney. Here's the breakdown as of today:
  • Nevada : 16% chance of Romney winning
  • New Hampshire : 16% chance of Romney winning
  • Iowa : 12% chance of Romney winning
(These chances come from Pollster.)

Clearly, for Mitt Romney to win the Presidency, he must overcome all these poor chances at once. To calculate Romney's chance of winning these three states we just multiply these three probabilities together to get an 0.3% chance. Here's the math:

     0.16 * 0.16 * 0.12 = 0.003072

As you may recognize, 0.3% is a very slim chance. You might be thinking: "Clearly, that's not Romney's chances for winning this election. After all, he is leading the national polls by about 1.5%."

Indeed, this thinking is correct because there are other ways for Romney to win the Presidency. He could simply win Ohio where he has a 5% chance of victory. If Mitt Romney wins all the swing states in which he has better chances and if he also wins Ohio, then Romney will win the general election. The reason is that Ohio is worth 18 votes in the Electoral College while Nevada, New Hampshire, and Iowa are worth 16 votes in aggregate.

It's obvious that there are many paths to victory for both candidates, so calculating the chance of victory for a candidate requires one to calculate the chances for all paths to his victory.

Using this approach, here are the chances for the candidates if the election were held today:
  • Obama: 96.25% chance of winning
  • Romney: 3.75% chance of winning
Additionally the most likely electoral vote count would be:
  • Obama: 290 votes
  • Romney: 248 votes