Friday, October 22, 2004

Election predictions, anyone?

Anyone care to take a guess as to how the presidential election is going to turn out? I'll post my prediction as a comment so as not to influence your guesses.


Blogger Mwal said...

The md5sum of my prediction text is:


10/22/2004 12:50:00 AM  
Blogger Vincent said...

Electoral map (using the LA Times utility)

Split electoral votes: None. All of Nebraska's CDs are safe Bush. I'll predict that Maine's CD2 will go to Kerry. The ballot referendum will fail in Colorado.

There's no clear trend in the state polls for Iowa, Wisconsin, Ohio, and Florida. Assuming the rest of the states go as shown, either Kerry needs to win Florida, or he needs Ohio plus one of Iowa and Wisconsin.

Biggest upset potential: Arkansas.

10/22/2004 12:52:00 AM  
Blogger Vincent said...

I see Mwal is being secretive (and fast), as usual. :-)

Incidentally, this suggests a campaign strategy for Kerry. The priority order of states (for campaign visits and advertising) should be approximately as follows: OH, WI/IA (tie), FL, PA, AR, WV, MN/MI, NM, VA.

Kerry should go all out for the three borderline states critical to victory. Next priority should be to try to poach FL, which would render OH, WI, and IA irrelevant. Kerry needs to make some move to counter Bush's long-shot gambit in PA, whose 21 electoral votes are critical to Kerry, but he can most effectively pressure Bush by making a sneak attack on AR (weakest Bush state) and WV (which would coordinate well with his OH campaign). Bush will push hard for MN and/or MI, but both should be relatively safe for Kerry. NM is leaning more pro-Kerry by the day. VA may just barely be within reach, and its proximity to PA, WV, and OH justifies its inclusion as the lowest worthwhile priority.

Notably absent from this list are hopeless states (NV, CO, NC, TN, and MO -- if any of these fall, Bush will have lost anyway) and New England states that should be tenable with minimum effort (NH, ME CD2).

10/22/2004 01:12:00 AM  
Blogger finou said...

This comment has been removed by a blog administrator.

10/22/2004 10:24:00 AM  
Blogger finou said...

So an interesting comment was made in my lab:
If the Red Sox win the world series at game 7, then the victory parade in Boston will be on election Tuesday. Now that probably means that a bunch of Bostonians won't vote so what if Kerry ends up losing the election because he loses Massachusetts because the Red Sox won the world series :)-

10/22/2004 10:25:00 AM  
Blogger Justin said...

The thing that has been most interesting to me from the "horse-race" perspective is the degree of inaccuracy in the various polls, where different polls are making vastly different predictions simultaneously. My hunch is that we are rapidly apporaching another "Dewey defeats Truman" election, where the pollsters discover that their methods have some sort of systemic bias.

You have a lot of younger people, particularly younger and techie, that use a cell phone as their home phone. Pollsters cannot call these phones, so these folks aren't represented in the samples. Many other people are screening their calls through an answering machine and/or caller ID and don't bother to pick up to answer a poll. This skews the polls toward those people that are lower income (no money for cell phones, caller ID, etc), and the unemployed and stay-at-home Moms who are lonely/bored enough to take a poll.

On the turnout front, the democrats are wagering way too much on their ability to turn out the freshly registered voter and have been awful at mobilizing their base. I'm betting that a large majority of these folks will sit out this election just as they have sat out earlier elections, particularly when Bush starts his "warm and fuzzy" TV campaign in major urban centers in the next week or two. Kerry has been ineffective at mobilizing his base because he refused to go after Bush on the war until too late in the game and has failed to articulate any sort of real message. Bush's attack on the Kerry health plan in the last debate, for example, was savage and almost completely unrebutted. If Kerry explained his plan, there were enough subjunctives and dependent clauses that I couldn't understand it.

Despite my desire to see Kerry win, I'm wagering an electoral college landslide for Bush.

10/22/2004 01:02:00 PM  
Blogger Vincent said...

Good to hear from you, Justin! How are you doing?

I think you're absolutely right that there is a large systematic error that the polls can't get a handle on. Unfortunately, I don't think anyone knows how large it is, how variable it is from state to state, or who will get the advantage from it. There are many unanswered questions. Will people without a land line favour one particular candidate, and by how much? Despite partisan claims by each side, how many actual new voters (no duplicates or in-state moves) have been signed up in each state? How likely are new or first-time voters to vote? Will undecided voters break for the challenger as usual, or did 9/11 change everything? Or will they all stay home? How successful will partisan get-out-the-vote drives be this election?

These systematic errors are at least as significant as the quoted random margins of error and don't average out as sqrt(N) like the random errors.

I may be proven wrong by the election results, but I'd guess that the Democrats' strategy is a sound one. The base has been mobilized (mobilizing) for four years now, and their extreme distaste for the current president will be motivation enough to get them to the polls in large numbers. The critical voters for Kerry will be the "soft" Democrats, undecideds, and disenchanted moderate Republicans. Bush's approval ratings show that if Kerry can win this group, the election will be his even if the Republican base turns out in large numbers. Kerry's problem is that many people in this critical group are either uninformed as to Kerry's stand on issues or believe (incorrectly, I think) that Kerry will be softer on terrorism, a claim that that Bush and especially Cheney hammer incessantly on the campaign trail. The challenge is twofold: let the swing voters know what Kerry stands for, and inspire them to get out and vote. I'm not sure that Kerry's been all that successful on either count.

. . . . . . . . . . . . . . . . . . . . . . . . . . .

Okay, so I'm clearly an informed Kerry supporter, but I thought Kerry did a great job in the third debate on health care. If I recall correctly (and I may not, since that was a while ago), Kerry detailed an optional plan for Americans to buy into government-sponsored health insurance of the kind given to government workers, with subsidies for those unable to afford it. Bush countered that it was 1) an expansion of government and 2) too expensive, and then offered that his plan would be for medical malpractice litigation reform. Kerry's responded that his plan was optional and will expand health care coverage, an answer enough to satisfy all but the Republican base on Bush's first criticism. He never really addressed the second criticism, which I think was a real missed opportunity. As for medical malpractice, Kerry pointed out that this is a very small component of health care costs. He proposed allowing Medicare to negotiate for lower drug costs (good idea) and allowing the reimportation of prescription drugs from Canada (bad idea, but politically popular). As far as I can tell, Bush's only positive contributions to the issue were the Medicare drug discount card (which gets mixed reviews) and health savings accounts.

10/22/2004 01:44:00 PM  
Blogger Justin said...

I've been well. Working on a contract in the Detroit area now... Unfortunately, I realized yesterday that I won't get to vote this year. I was scheduled to fly back home for the weekend a few weeks ago, when I planned on picking up my absentee ballot. At the last minute, some code blew up, so I volunteered to stay in MI to fix it, so no chance to pick up a ballot :-(

Bush's attack on the expense and complexity of Kerry's plan dovetailed very nicely with the ads that he's showing now and with the old "tax and spend" label that has been applied to liberals. Bush's point that if the government were subsidizing "the same care Senators get", employers would be idiots to continue offering coverage to their employees, which makes Kerry's claims that the program was optional ring hollow. Frankly, I think that the Democrats are being eaten alive trying to advocate government-run health care without coming out and admitting that's what they're advocating.

I think pundits are vastly over-estimating the capacity of average voters to even follow Kerry's explanations, even when the content is adequate. There were two or three times in the last debate where Kerry stopped talking when he realized that he was on a tangent six layers of predicate clause deep with no idea how to get back to his point. Bush, on the other hand, speaks in short, clear sentences that are simple to understand.

I'll agree with you if you say that something as complex as health care lends itself to complex analysis, I'll agree with you. When it comes time to debate, though, if you can't explain why the economics of your plan works in a simple manner, 99% of the people will assume that the economics don't work.


As for the sampling errors, I think we do have a pretty good understanding of the biases they introduce. A lower- or lower-middle income family is, more or less, just as likely to answer a phone poll today as they were 10 or 20 years ago. An upper-income family is far less likely to answer, since they know that it is a pollster calling.

Newly registered voters, other than African Americans in the 1960's, never vote in any number. If someone doesn't care enough to register when they can do so at the DMV, city hall, etc., why would you think they will pay enough attention to the race to care who wins, figure out where their polling place is, and take the time to drive over there to vote? When polls look at likely voters, rather than registered voters, Bush gains significantly.

10/22/2004 02:36:00 PM  
Blogger Qian said...

I would like to be able to predict a Kerry victory, but it's just too close to call and it shouldn't be. I think that the Kerry campaign has done a very poor job of framing the debate. An election with an incumbent should always be (and usually is) a referendum on the performance of the incumbent and this incumbent has done a mediocre to poor job in all areas. And that is the one message we should have heard from the Kerry camp all along. Instead there's been far too many different messages from Kerry, starting with the stupid slogan "stronger at home, respected in the world" that's since been abandoned. Then they started coming out with plans like they were breeding rabbits. I just got a flyer today from the Kerry campaign about his plan for "Rural America." I know I don't live in NYC or LA, but I don't think I live in freakin' rural America! Why the h*ll does Kerry even have a plan for rural America?! All these Kerry plans take the focus away from the Bush record. Do people know anything about Bush's plan for the next 4 years other than more tax cuts? Heck No! So Kerry has all these plans out there with huge bullseyes painted on them for the Republicans to shoot down. And at the same time we're NOT talking about what Bush has done. If you watched the presidential debates you might have noticed that every time Kerry criticized Bush on something he went on to say "I'll spend x billion dollars more on it." Then he promised to have bigger tax cuts for the middle class than Bush. The end result is that people think he's just making empty promises, which he probably is. All the while the undecided voters are thinking "Kerry's plans don't sound feasible" instead of "Bush really screwed up on Iraq and the economy's not so hot and the environment's not that great and a lot of jobs have been lost and how come gas is so expensive and...."

Why have the Democrats forgotten everything Clinton knew when he ran again Bush 1.0? Clinton focused on the economy because Bush 1.0 was solid on foreign policy, but attacking his record still worked. Does anyone remember Clinton's plan for rural America during the 92 campaign? No! And if he had one he made sure it was locked away in a vault protected by freakin' lasers so nobody could see it and talk about it. It's about the incumbent, stupid!

In contrast, the Republicans, who are far more focused and effective campaigners, have successfully made it a Kerry v. Bush battle without much resistence from the Dems. And Bush is a charming, likable guy to a large portion of the population. Had the campaign been about Bush's record instead of a personality contest, I think Kerry would be in a much better position. The Republicans know their strengths and weaknesses. Sadly, the Dems have no clue. Still, the election's going to be very close. Now imagine if Kerry had ran the right campaign.

10/22/2004 03:36:00 PM  
Blogger finou said...

What I don't get is how people can find Bush trustworthy, likeable, and especially charismatic.

Without looking at the content of what he is saying (which is what a lot of people seem to do anyway): Bush reminds me of those sleazy guys that come up to you at bars with horrible pick up lines... or a used car salesman. Either way, based on how he speaks and delivers his messages (which is something you could totally change so it must be a conscious choice, right?), I wouldn't trust him or find him likeable or relate to him in any way. I'm not saying this based on how I think he did as a prez or what I think of his policies or anything of actual content. I'm just saying I don't like his manerisms, how he answers questions, and that sort of thing.

So why does his way of talking and stuff work with such a large portion of the US? (maybe they all like used car salesmen? or maybe they are all sleazy bar guys?)

I mean not that I find Kerry all that charismatic either... but no one ever accuses him of that!

10/22/2004 06:08:00 PM  
Blogger Qian said...

Indeed, the world seems to be divide into people who think Bush is "folksy" and people who find him totally insufferable. I'm no psychologist, but maybe the difference does have a lot to do with what you think of the content of his communications. I'm sure a lot of Republicans find Kerry's mannerisms totally off-putting. Maybe I dislike the way Bush talks because I know he's lying to me. Maybe if I thought he was telling me the truth, his mannerisms would seem "charming". Shudder.

10/22/2004 06:19:00 PM  
Blogger Justin said...

I think Kerry tends to come across, at best, as the smartest kid in the class that kept reminding the teacher that he forgot to assign homework. 99% of the country wasn't the smartest kid in class.

At worst, Kerry comes off like he's trying to act a lot smarter than he is. His speech has no sense of flow, and he seems allergic to periods. By the time he finishes a sentence, I feel like I need to go back to junior high English to diagram it to figure out what he said. Bush was a horrible speaker during the 2000 primaries, but he at least was direct. Today, he is far more polished-- I haven't seen nearly the same growth from Kerry. His handlers really should have had a heart to heart with him during the primaries that he had to lose the Senate-speak.

I've been begging Kerry for months to come out with some reason to motivate me to vote for him. His policy on Iraq has been, at best, tortured. I don't believe that the president has a lot to do with the short-term performance of the economy, and I don't think that his economic policies, over the long term, will be sufficiently different from Bush's to warrant any excitement. Most of the rest of his plans look to be DOA in Congress.

Getting back to the mobilization of the base, the Democrats sold themselves short in the primaries by going for someone that looked electable over someone they actually liked. Yeah, those that are totally fed up with Bush will vote, but I don't see any sense of excitement among core issue voters.

Frankly, I'd be voting against Bush because I think he's telling the truth and for Kerry because I think he's lying. For example, he has a history in the Senate of being a free trader, though he has become a protectionist in the campaign. I don't see him making significant changes in Iraq-- just like Bush's criticism of Kosovo ended the moment he had any power to change the situation.

10/23/2004 02:21:00 AM  
Blogger Emily said...

Hi, You should Check This Out. Get a Free Flatscreen TV or LCD Monitor. Just click on link to find out more. FreeFlatScreen

10/23/2004 02:28:00 AM  
Blogger Vincent said...

Today's has an interesting look at what may be this year's equivalent of the butterfly ballot.

10/24/2004 02:04:00 PM  
Blogger Eric said...

I don't understand the criticisms of Kerry for being hard to understand. From what little I heard, it came off as easy to understand and also easy to see through his propaganda.

I agree that it's easy for "optional" plans like medicare and no-student-left-behind, to become virtually mandatory for all but the rich folks. And that's especially true, if you don't have the option of not opting to not pay to subsidize the program for other people. Personally, I'd be more interested in a real debate over whether there should be a massive, extremely inclusive, government run health insurance program and what it should (not) cover. But instead Kerry chose to try to further his politican ambitions by avoiding that debate and trying to trick Americans into thinking they could have it all.

Yes, I worry that the cost of the current program is too high. And I don't think either candidate has even (credibly) pretended to have any idea of want to do about it. Bush added benefits to medicare. Kerry wants to add more. I'm suprised that more young people don't object to the massive fiscal burden that is being placed on them. The senior citizens groups are quite effective in lobbying for more benefits for seniors, but there is virtually no representation of young people who don't want to pay for it.

Given that Kerry's campaign seems to be a constant attack on Bush, I think it would be more effective for him to play clips of Bush campaining in 2000 followed by clips of him campaigning in 2004. Demonstrating that Bush has changed his position on several issues (bordering on isolationist to Iraq, CO2, steel tarifs) would simultaneously make Bush seem less trustworthy and like flip-flopping was the norm for politicians.

I also agree that democrats would have been better served by nominating someone who people could like and beleive was honest, instead of someone who was "electable". IMHO, by nominating Kerry, democrats reinforced Repulicans' views that Democrats want to play slimey politics, tax and spend, perpetuate class warfare, etc.

10/24/2004 02:09:00 PM  
Blogger Vincent said...

This comment has been removed by a blog administrator.

10/24/2004 04:26:00 PM  
Blogger Vincent said...

New state polls show Bush taking the lead in Hawai'i, formerly considered a safe Kerry state. A poll shows Bush +5 in Michigan, yet another released today shows Kerry +5. Arkansas is tied, and Kerry is pulling to within the margin of error in Nevada. (See, e.g., Real Clear Politics for summaries of state polls.)

Are these polls indicative of national trends (perhaps undecideds breaking in opposite directions on a state-by-state basis), or are the polls especially noisy this weekend? What Election Day surprises await us?

10/24/2004 05:03:00 PM  
Blogger Mwal said...

Wow, I didn't realize you were going to be so specific in your predictions. Though it is more interesting that way.

My prediction was: "I would bet at 53:47 odds that Kerry will win."

i.e., I was inclined to think Kerry would win, but not very confident in my ability to make a prediction.

This is still mostly what I think. Though I am a little nonplussed by the behavior of the polls.

Anyway, I thought I might as well reveal my prediction before the election. If I revealed it after the election, it would probably just get lost in all the legal challenges, rioting, insurrection, extraterrestrial invasions, etc. :-)

10/28/2004 03:06:00 AM  
Blogger Vincent said...

For some insight on why poll numbers vary, Mystery Pollster investigates the Gallup methodology in depth. The same analysis could apply to other poll organizations as well, but many others don't give the same level of detail regarding their data collection and analysis techniques. Information on how these may vary between polls is given in this Slate piece.

10/28/2004 07:55:00 PM  
Blogger Vincent said...

Excellent Florida news: a Quinnipiac poll shows Kerry ahead by 17 points among early voters in Florida.

10/28/2004 08:06:00 PM  
Blogger Vincent said...

Okay, one final set of predictions:

I stand by my initial map, with a couple of exceptions. Florida will go to Kerry, but New Mexico is a real tossup that has the potential to be the closest state for the second consecutive presidential election. Add the potential for a Kerry victory in formerly red Nevada, and I'll guess that Kerry's final EV count will be 311. While the outcome in at least one state will be within the margin of litigation, the nationwide result will not.

Nader's effect on Kerry will be negligible. But Badnarik may draw enough votes from Bush to swing a state Kerry's way (most likely NM or NV).

Voter turnout will flirt with 130 million nationwide. Kerry will win the popular vote and break 50% of votes cast (but just barely).

Republicans will be overzealous in challenging minority voters at the polls, and there will be at least one incident in which well-deserved physical violence occurs.

10/31/2004 12:32:00 AM  
Blogger Mwal said...

Ok, here are some specific estimates, produced in my brain via a mysterious process of dynamical symmetry breaking:

Kerry gets 285 +/- 10 +/- 17 +/- 5 electoral votes. (Bush gets the rest.) Errors are: statistical, systematic, and electoral model, respectively.

Turnout is 120 +/- 9 million votes.

Kerry gets (49.5 +/- 0.7 +/- 0.7) percent of the popular vote.

If Kerry wins, Bush will concede before 2004-Nov-03 23:59 UTC.

If Bush wins, then Kerry will concede by the aforementioned time if and only if a swing of more than 10,000 votes in a state would be required to change the result.

If there is a tie in the Electoral College then a whooooole lot of people will complain about it. (Both the tie and the College.)

(Note: Just to be totally clear, the quoted errors describe the estimated variation between the prediction and the actual election, rather than the deviation of the election itself from a 'perfect election'. For that, I'd use the names "resolution" and "bias"... but such an analysis is beyond the capabilities of the instrument used to produce these estimates.)

11/02/2004 02:48:00 AM  

Post a Comment

<< Home