As you might have heard, the presidential election is tomorrow. As I've said, I believe I'm alone on ScienceBlogs as supporting anyone but Obama. But this is Built on Facts, not Built on Wishful Thinking, and so let's have our official quadrennial Election Prediction Contest!
Here's the plan. We're predicting three quantities of interest.
1. Electoral Vote
2. Senate Seats
3. Popular Vote (in percent)
We're not going to bother with the house, since raw numbers aren't as important. A house with 59% Democrats is only marginally different than one with 61% Democrats. In the senate on the other hand, a 59 seat majority versus a 61 seat majority can make a world of difference. So if you'd like to participate in the predictions, this will be the plan. The electoral vote number is the number of electoral votes Obama gets. You can include McCain's number as well especially if you think a third party will get any votes. But if you do, make sure you note which number is which. The senate seats category is the number of seats held by the Democrats. Assume the Independents continue to caucus with the Democrats. The popular vote category is the percentage of the popular vote held by Obama and McCain respectively. They shouldn't add to 100% because of the smaller parties, so you'll probably want to take that into account.
My predictions (try to format yours at least roughly along these lines):
1. 338 Obama EVs.
2. 58 Democratic senate seats
3. 49.0% Obama, 45.0% McCain, 6% other.
Thus each prediction involves five numbers, one for each of the first two categories and three for the popular vote. Scoring will work like this. Once the actual numbers are known, for each of your guesses, take the actual result and subtract your guess. Your goal is for the following expression to be as small as possible:
It only uses your first two guesses, the popular vote is there as a tiebreaker. Note I've weighted the senate score somewhat on the grounds that the possible spread in EVs is much higher than the possible spread in senate seats. On the off chance more than one person has the same guess for all three (for instance, if someone copies) the tie is broken in favor of the earlier poster (I disqualify myself, by the way). Sadly there's no prize other than the knowledge of a job well done.
Where does my 338 number come from, by the way? Start with the 2004 map and give Obama these states: CO, FL, IA, NV, NM, OH, VA. I don't think he gets Missouri or North Carolina, but those are the two states I think I'm mostly likely to be wrong about.
What's the chance the McCain pulls out a victory? According to Intrade yesterday, it's about 1 in 8. According to FiveThirtyEight yesterday, it's about 1 in 26.
So let's have your guesses! If possible try to keep mostly on topic about what you think will happen. There's the rest of the internet for arguing about which result you think is better. I'm going to close comments on this post at around midnight tonight so there can be no late entries. Good luck!
1. Obama 291 EVs (McCain wins both OH and FL)
2. 58 dem senate seats
3. 51% Obama, 45% McCain, 4% other (are we really going to get 10ths of a percent in the final results? if so, assume all those are #.0)
Three predictions: Tuesday will be an infrastructural Grand Guignol of Brobdingnagian proportions. An 04 November Obama victory will be met with an immediate Supreme Court appeal contesting early voting and Obama's birth upon US soil/citizenship. Gears are lubed when the Treasury Department withdraws Wall Street support, followed by a 2000 point Dow plunge.
Departing Clintonistas supergluing keyboard letters was small change. In what away were mechanical voting machines - throw a lever - insufficient?
1. Obama 322
2. 52 Dem seats
3. Obama 51%, McCain 44 % Other 5 %
1. Obama 322
2. 52 Dem seats
3. Obama 51%, McCain 44 % Other 5 %
1. Obama 338 EV
2. 57 Dem Seats
3. Obama 53%, McCain 43%, 4% Other
I see you all have not lost your fight :
GOOD! Because we have a lot to do. You! (the American people) are going to have to take back control of your elected government at every level, and set your government back on the right path of service to you, and the greater good of the World.
Barack Obama and the democrats are your best hope of doing that now. Tell your family, friends, and everyone you know to support them as best they can. Because the Bush McCain vote fraud, vote cheating, vote buying, vote manipulation machine is already hard at work to cheat you again. And we all know what a disaster that has been the past 8 years of Bush McCain.
Barack Obama and the democrats will need all the power you can give them at every level of government (Federal, State, County, and local City elected governments). Obama and the democrats will have an enormous mess to fix for the American people, and the rest of the World. A mess caused by the corrupt Bush McCain administration.
You see, starting back in 2000, and before 911, it was mostly the Republican governors, Republican legislatures, and county elected Republican officials that conspired with the corrupt Bush McCain administration to raise college, and university tuitions by the fastest, and highest rate increases in American history. Some state tuitions went up by as much as a WHOPPING! 30% in one year.
The reason the Bush McCain administration did this was to force struggling working class kids into the military to pay for the sudden jump in tuition. Which was forced on them by the corrupt Bush McCain administration, and their corrupt Republican Governors, and republican controlled state legislatures.
See, Bush McCain had plans to get us into all these immoral, foolish, criminal, and unnecessary wars from the start. So they could use these wars to seize power, and later to get reelected. But, for their evil plan to work they needed more volunteer soldiers struggling to pay for an education whose blood they could spill to help them seize more power. Remember Bush McCain's "MISSION ACCOMPLISHED!" theatrics.
The exploitation, and lost lives of these finest Americans is despicable, disgusting, immoral, corrupt and criminal. And it makes me SICK, and ANGRY!
You will have to vote for Obama, and the democrats in overwhelming numbers to overcome the Bush McCain vote fraud machine. Vote early if you can. Then help your fellow Americans cast their votes on through election day. Vote for Obama, and the democrats like your life, and the lives of your loved ones depends on it. Because it does. You will not survive 4 more years of "Let Them Eat Cake" Bush McCain, and their republican allies.
Just look at the mess we have now.
You can fix this mess with your votes for Obama, and the democrats. But you must continue to surge forward for Barack Obama, and the democrats, and for your-selves most of all. The children, and the World are counting on us.
It's in your hands now. And I know you will get it done.
God bless all of you.
JACK SMITH - WORKING CLASS... :-)
1. Obama 360 EV
2. 57 Democratic Senate seats
3. Obama: 54%, McCain 45%, Other 1%
What if the Georgia Senate race goes to a runoff, how will you count that towards the total, or will you wait for that to be decided?
Here is my prediction:
EV Obama 353, McCain 185.
Senate, Dem 55, Rep 42, Ind 2; with GA too close to call
Pop Vote: Obama 52% McCain 45 other 3%
Well I see someone didn't read the last paragraph! Discussion is great, but do try to keep the electioneering to a minimum.
where are you guys getting "other" votes above 1%? i just don't see the third-party candidates this year approaching that figure.
personally, i'm guessing Obama getting the popular vote by a margin of 10 percentage points or better (he'd better; any less and he'll lose, even if the GOP have to drag the court cases out into February to do it), and no less than 300 electoral votes. the senate, i don't know nearly enough about the politics of the various contested seats to guess.
EV: Obama 329, McCain 211
Senate: Dem 58 (including Sanders), Rep 41, Lieberman 1.
Popular vote: Obama 52%, McCain 46%, other 2%
I'm giving Obama all the Kerry states plus CO, IA, MT, NV, NM, NC, OH, and VA. My gut feeling is that Ron Paul will play spoiler in MT, tipping that state to Obama, while the remaining tossup states (currently AZ, FL, GA, IN, MO, and ND) will go to McCain. I'm most likely to be wrong about FL, MT, and ND. If Obama wins Georgia, look for a Dem landslide.
In the Senate, Dems will pick up AK, CO, NH, NM, NC, OR, and VA. GA and MN are too close to call, so I will guess that the Dems will get one of those two (but I'm not going to guess which one). Since that gives 59 with Lieberman, I wouldn't assume that they will need Lieberman in the Dem caucus (however, they will keep Sanders)--but if they do get to 60 including Lieberman, they will keep him.
EV: Obama 311
Senate: 57 Democratic
Obama 52%, McCain 47%, Other 1%
EV:
Obama 353
McCain 185
Senate:
57 D
41 R
2 Independents caucusing as Dem: (Lieberman from CT, Sanders from VT)
Popular Vote:
52% Obama
46% McCain
Given your interest in sites like InTrade, as well as the shape of your Sunday Function this week (the ArcTan) I thought it would be really nice to see a post regarding this site:
http://electoralmap.net/polling.php
A comparison of polling data to InTrade prices for every state.
The general shape makes perfect sense. As soon as the polls start tipping by 5 or 10 points, the market becomes convinced. But the sites author chose the logistic function. What function would you choose? ArcTan? Error Function?
I'd be interested in your thoughts.
--Ed
Hey, that's very clever. Thanks for the link! I think there's two possibilities for improvement.
First, I wouldn't force the intercepts of the best-fit function to zero and one at the ends. Intrade the company makes its profits as a percentage cut, and so that may affect the prices people pay especially near the edges. In a fee-free version of Intrade the intercepts should be zero and one, but we have no a priori reason to believe it should be thus as it is now.
Second, the logistic function seems to fit well. But I'll bet the error function would fit better. There's a lot of random factors involved, and the more there are the more I suspect a normal distribution or something like it will appear. The difference is slight, but the normal distribution function is more closely bunched about the mean than the logistic distribution which results in a steeper cumulative distribution functions. Since this one doesn't seem to be steep enough (apart from the whole intercept thing), that leads me to think the error function is the way to go.
1. 357 Obama EV's
2. 58 Dem
3. Obama 52% McCain 44% Other 4%
Ed Brayton has previously said he will vote Libertarian, though he seems to have mixed feelings about Bob Barr. (Ed does not live in 'battleground' state.)
1) Obama 360
2) 59 Dem
3) Obama 53.5% McCain 45% Other 1.5%