Based on the currently counted ballots (which are not 100% complete still as there are still absentee and military overseas ballots to count in some states, but is about 98% complete nationally), here is what happened state by state versus my projection:
On a national level, it appears that Obama will finish at approximately +2.7% (50.6% for Obama, 47.9% for Romney and 1.5% for minor third-party candidates), an error of 1.8% and a bias of 1.8% for Romney.
To remind those of you who aren't statistics junkies - error is how wrong the projection is, bias is how much the projection favored one candidate versus reality. In the case of a single point of data, like the national vote, they will be the same, in the case of the 50 states, they will be different since if I project two states and in one case I under projected Obama's vote total by 2% and the other I over projected his vote total by 2%, my average error will be 2% (the absolute amount I missed each state by), but my bias will be 0% (since I missed equally in both directions.)
As you can see, my average absolute error was 3.9% for all 50 states plus DC with an average bias of 1.9% for Mitt Romney. The average error is higher than my national error because the average weights all states equally, while the national vote is influenced by a number of states that were called much more closely.
In the relatively close states, those decided by 10% or less, my average error was 2.3% and my average bias was 2.2% for Romney. It makes sense that I called these states a lot more closely since there was a wealth of polling data in those states, versus virtually none in a lot of the non-competitive states. The fact that the bias and error were so close speaks to the fact that with the exception of Ohio and Indiana, all of the other states showed a bias for Romney.
In total, I give myself a grade of about a B+ on calling the race. On the plus side, I got 49 states right and an absolute average error of 3.9% across all 50 states is actually quite good given the limitations of public polling. I ding myself some because clearly there was an almost across-the-board pro-Romney bias to the projections. This speaks to the fact that while I had been saying and most pollsters had been polling a race somewhere between 2004 and 2008 in terms of turnout model, the actual turnout closely resembled 2008, showing the strength of the Obama team's ground game.
I don't see these results and find a need for a radical shift in my methodology.
But I do see the need to make one tweak.
Scott Rasmussen has generated a great deal of controversy over the past 8 years by publishing polls that are out of the mainstream of other pollsters.
This year:
* He missed the national vote by about 4%, with a Pro-Romney bias, among the worst of the national polls
* In Ohio, he missed by 2% - Pro-Romney (worst of all polls), Colorado by 5% - Pro-Romney (worst of all polls), Wisconsin by 7% - Pro-Romney (worst of all polls), Iowa by 7% - Pro-Romney (worst of all polls) and on and on
In every close state, dropping Rasmussen from the averages would have improved the accuracy of my calls.
In 2008, his misses weren't as outrageous (he missed the national vote by about 1.5% Pro-McCain and missed most of the swing states with a Pro-McCain bias, but not as severely), but clearly Rasmussen is not equally likely to be wrong on either side.
Accordingly, going forward in 2016, if Rasmussen continues polling, I will be considering the Rasmussen poll to be a partisan poll rather than a non-partisan poll for purposes of my averaging.
First Look - 2016
Yes, I know, I should probably let President Obama enjoy the moment for a few days before I start writing about the 2016 race. But, it's been 5 whole days since we had an election! Plus, the President doesn't have to worry about 2016, since he won't be in the mix.
First, let's have a look at the map and what we learned Tuesday night about the states that may be in play in 2012. Let's first assume that states that were D+10 or R+10 (that is, voted 10% more Democratically or 10% more Republican than the nation as a whole) are more-than-likely not going to be in play in 2016 either. They could flip if there is a landslide election, but are probably irrelevant to a close race. We'll take a look at the ones between 10-15% for those rare exceptions such as Indiana in 2008, but anything above 15% is solidly out of play.
These states give us the following:
Safe Republican (>+15% R)
Utah, Wyoming, Oklahoma, Idaho, West Virginia, Arkansas, Kentucky, Nebraska, Alabama, Kansas, Tennessee, North Dakota, South Dakota, Louisiana, Alaska, Texas, Montana.
Strong Lean Republican (+10-15% R)
Mississippi, South Carolina, Arizona, Missouri, Indiana, Georgia
Lean Republican (+5%-10% R)
None
Toss-Up (Between +5% R - +5% D)
North Carolina, Florida, Ohio, Virginia, Colorado, Pennsylvania, Iowa, New Hampshire, Nevada, Wisconsin
Lean Democratic (+5% - +10% D)
Minnesota, Michigan, New Mexico, Oregon
Strong Lean Democratic (+10-15%D)
Washington, Maine, Illinois, New Jersey
Safe Democratic (+15% D)
Connecticut, Delaware, California, Massachusetts, Maryland, New York, Rhode Island, Vermont, Hawaii, District of Columbia
As you can see, including all the states that are either greater than D +5% or greater than R +5% yields an electoral map that is 217 Democrats, 191 Republicans.
Looking at the swing states, we can understand where the fights may take place. We'll look at 2012 position relative to the national vote, the history in the last 7 election cycles (which saw 4 Democratic wins and 3 Republican wins nationally) and the trend the past 8 years:
North Carolina:
2012 Rating: R +4.9%
History Last 7 Elections: 6 Republican, 1 Democrat (Obama in 2008)
Trending: Democratic - (R +10.0% in 2004, R +6.9% in 2008, R +4.9% in 2012)
My Notes: North Carolina had never been a swing state until 2008, when Barack Obama pulled off a shocker in the state. It's trend was similar in 2012 to 2008. It is still more conservative than the nation as a whole and its socially conservative demographics are changing. If the trend continues in the next 4 years, it will be a swing state in 2016, but one that still favors the GOP in a close race.
Florida:
2012 Rating: R +1.8%
History Last 7 Elections: 4 Republican, 3 Democrat (Clinton in '96, Obama in '08 & '12)
Trending: Mixed (R +2.6% in 2004, R +4.4% in 2008, R +1.8% in 2012)
My Notes: Florida is definitely a swing state, but one that has consistently favored Republicans in very close elections. While 2012 saw the closest match to the national vote since 2000 (when it was R +0.5%), there is no clear trend here and it always appears to be on the Republican side of the ledger. Florida seems unlikely to be decisive in 2016.
Ohio:
2012 Rating: R +0.8%
History Last 7 Elections: 4 Democrat, 3 Republican (matched national results)
Trending: Mixed (D +0.3% in 2004, R +2.5% in 2008, R +0.8% in 2012)
My Notes: Ohio is the swingiest of swing states, picking the Presidential winner every year since World War 2. It has no clear trend in either direction and is likely to mirror the national vote again in 2016, favoring whoever wins the national vote.
Virginia:
2012 Rating: D +0.3%
History Last 7 Elections: 5 Republican, 2 Democrat (Obama in '08 and '12)
Trending: Democratic (R +6.8% in 2004, R +0.8% in 2008, D +0.3% in 2012)
My Notes: The growth of Northern Virginia has swung the balance of power in Virginia. It could certainly be winnable by a strong Republican candidate in 2016, but all of the demographic trends favor it going Democratic in a close election.
Colorado:
2012 Rating: D +2.0%
History Last 7 Elections: 4 Republican, 3 Democrat (matched national results except went for Dole in '96)
Trending: Democratic (R +2.3% in 2004, D +1.9% in 2008, D +2.0% in 2012)
My Notes: Hispanic population growth have turned this traditional swing state from a lean Republican to a lean Democratic orientation. If Republicans can't cut into the Democrats 3:1 edge with Hispanic voters, they will struggle here in 2016.
Pennsylvania:
2012 Rating: D +2.5%
History Last 7 Elections: 6 Democrat, 1 Republican (Bush in '88)
Trending: Republican (D +4.9% in 2004, D +3.2% in 2008, D +2.5% in 2012)
My Notes: The last several cycles, Republicans have invested heavily against this large electoral prize only to come up significantly short. Republicans are chipping away at the Democratic edge here, however, as growth in Democratic Philadelphia slows and Republican Pittsburgh is resurgent. I expect Pennsylvania to be legitimately in play in 2016.
Iowa:
2012 Rating: D +2.9%
History Last 7 Elections: 6 Democrat, 1 Republican (Bush in '04)
Trending: Democratic (D +1.7% in 2004, D +2.5% in 2008, D +2.9% in 2012)
My Notes: Iowa has been more Democratic than the nation every election cycle since 1988 and is trending that way now. Agriculture is still strong but other sectors of the economy are starting to eclipse it. I see it as favoring the Democrats in 2016.
New Hampshire:
2012 Rating: D +3.1%
History Last 7 Elections: 4 Democrat, 3 Republican (matched national results)
Trending: Mixed (D +3.8% in 2004, D +2.5% in 2008, D +3.1% in 2012)
My Notes: New Hampshire keeps picking winners. It is a swing state, but one that consistently favors the Democrats by a few points. It is also only 4 electoral votes, which makes it less likely to be decisive in a close race...although it certainly was in 2000.
Nevada:
2012 Rating: D +3.9%
History Last 7 Elections: 4 Democrat, 3 Republican (matched national results)
Trending: Democratic/Mixed (R +0.2% in 2004, D +5.6% in 2008, D +3.9% in 2012)
My Notes: Growing Hispanic population and growing Democratic Hispanic margins pushed the state from a true swing state to a lean Democratic state in 2008. That lead has moderated somewhat, but demographics will continue to swing to the Democrats unless the GOP can make inroads with Mexican Hispanics.
Wisconsin:
2012 Rating: D +4.0%
History Last 7 Elections: 7 Democrat, 0 Republican
Trending: Mixed (D +2.8% in 2004, D +6.9% in 2008, D +4.0% in 2012)
My Notes: No Republican has won Wisconsin since Ronald Reagan ran the table in 49 states in 1984. Paul Ryan no doubt helped the GOP some in 2012 and the Scott Walker revolution showed that the state is swingable, but this seems like a tough climb for the GOP in 2012.
Of note, if you take the 2012 and make the national popular vote exactly even, keeping the same leanings of the states, the Republicans pick up North Carolina, Florida and Ohio, giving them 253 states. Democrats pick up the balance of the states, giving them 285 electoral votes. This is the structural advantage the Democrats have, that I spoke of often during the 2012 race.
The easiest pick-up in a close race from there would appear to be Virginia, but there are two problems. One, winning Virginia would still give the GOP only 266 electoral votes, meaning they would also need Colorado. Second, Virginia is trending Democratic. A more plausible scenario is to pick up Pennsylvania, which is trending their way of late, which would give the GOP 273 and the win.
Next time, I will look at the potential 2016 candidates for both parties since, for the first time since 2000, both parties will have a more-or-less open field.
In December of 2008, when I first looked at the possible candidates in 2012 for the GOP, I said I thought the front-runners were Sarah Palin, Newt Gingrich, Tim Pawlenty and Mitt Romney, with Romney having the edge as the "next guy in line", a distinction that has historically been very important in GOP nominating contests.
I was wrong about Palin running, obviously. The other 3 did run and Romney did win, although I, along with virtually everyone else, missed projecting the insurgent campaign of Rick Santorum in the race.
If you like this site, tell your friends.
Posts
No comments:
Post a Comment