How Close Were US Presidential Elections?

Mike Sheppard writes "I'm a graduate student in Statistics at Michigan State University and spent some time analyzing past US presidential elections to determine how close they truly were. The mathematical procedures of Linear Programming and 0-1 Integer Programming were used to find the optimal solution to the question: 'What is the smallest number of total votes that need to be switched from one candidate to another, and from which states, to affect the outcome of the election?' Because of the way the popular and electoral votes interact, the outcome of the analysis had some surprising and intriguing results. For example, in 2004, 57,787 votes would have given us President Kerry; and in 2000, 269 votes would have given us President Gore. In all there have been 12 US Presidential elections that were decided by less than a 1% margin; meaning if less than 1% of the voters in certain states had changed their mind to the other candidate the outcome of the election would have been different."

Comments

Importance of protecting the process

[RichMan]

This shows how easy it would be to swing the election should one hack the voting in a few districts. The analysis can be used to show the regions to focus on.

This shows the importance of maintaining an open and audit able process if the system is to be protected from manipulation.

It also shows the importance of every vote and in protecting the rights of all to be able to cast their vote.

 

Re:99% off-topic question

[rthille]

I'm turning 41 in a week, and this is the 2nd election I listened to...even in 2000, while I 'listened' enough to make up my mind, I didn't think politics was really important. Even the Florida recount didn't seem to matter that much to me, I figured "how much more then the other one can one of these bozos screw things up?" After 9/11 and the other insane government fuckups of the first Bush administration, I got more involved. I figured there'd be no way 2004 would re-elect Bush, so I didn't donate too much or work too hard. Sure Kerry was wooden, but after the first debate my vote changed from "Anyone but Bush" to "Kerry, the guy who could articulate an intelligent position" (even if he could ramble on for days :)

Now in 2008 I'm working in a local campaign, donating money to Obama and Al Franken.

For an interesting picture about how much having the wrong guy at the top matters, read 'State of Denial'.

Marketing is an Engineering Problem

[Speare]

I have said in the past (since before 2000) that the very strong trend toward fifty-fifty splits between rivals only proves that Marketing is now an Engineering Problem.

To explain: all endeavors start as artforms, like "the tuning of these newfangled carburetors is a bit of a black art." Then you understand the general system well enough to call it a science, "we have found that if we measure the fuel mixture, we maximize combustion." Once the system is known very well, it is an engineering problem: "an electronic system monitors the mixture and adjusts for different conditions on the fly."

Just as the cola wars are in a well-settled detente, the business of national politics is a marketing endeavor. Whether you're Demopublican or Replicratic, whether you're a Preservative or a Libertine, your party system will simply apply the art, nee, the science, nee, the engineering methodology to ensure the candidates do the best they can. Of course, both sides have effectively infinite resources so the marketing comes out equal, and the course of history witnesses Gore/Bush 2000, too many 5-4 decisions to count, a roughly 50-51 Senate, and a dynamic but well-balanced electoral college.

We seem to be deadlocked into a 50%/50% world, regardless of the actual merits. Marketing is simply engineering the "choices" we have, and equally effectively on "both" sides of just about every political issue.

Re:Never changes

[Kjella]

It's an almost textbook example of optimization theory - assume there's two ice cream booths on a beach [0..1] with a uniform distribution of guests, now the optimal for the beach guests would having them at 1/4 and 3/4, but then each could steal customers by moving towards the center. End result you got two booths right next to each other in the middle, each serving half the guests. As long as any other booths can't enter (winner takes it all-system) that situation is stable. Any disturbance like the guest moving over to one side of the beach because it got better sun in the afternoon and the dividing line will move, again leaving half on each side. If you want clearer objective proof that having 40% of the votes it useless in the US, this is it. The politicians must redefine their politics so they're fighting for the majority, rather than stay true to anything.

Designed that way

[T.E.D.]

A "feature" (probably unintended) of the design of the Electoral College system is that most elections look like more of a blowout than they were. In theory, if someone manages to consistently get 50.5% in every state, they could win every state and the public will be told the next morning about the victor's huge landslide victory.

That's why after the 2000 election the Reps floated around those red state/blue state US maps with such glee. It made a squeaker look like a huge victory. (For a better picture, see the University of Michagan , which use some cartiographical tricks to adjust for population).
A better illustration are Regan's victories. Everyone knows Regan clobbered Carter and Mondale, right? Well, the true answer is not really, and sorta respectively. The electoral college turned his %50.7 victory in 1980 into a %86 state victory, and his %58.8 victory in 1984 into a %94 state victory.

It has been argued that this effect is actually good for the country, as it gives presidents more legitimacy from their elections.

What are the odds these were random?

[petes_PoV]

OK, we have some instances of small fluctuations causing major effects. Rather than just sitting back and says "wow, that was close", the next stage is to calculate the possibility of these events being statistically random.

Re:Thanks from the reminder

[jollyreaper]

FDR tried and failed to fix the 1930s recession..... it ultimately took a world war to bring-back full employment. Without the war, FDR would have been voted out of office in 1940, and the recession would have stretched through most of the 1940s.

Obama faces what FDR faced, and Obama's not going to be any more successful. (Unless a war saves him.)

Why is it we always praise wars for bringing full employment? I hate to use the cheesedick "war on x" phrases but seriously, what if we were literally do pull out all the stops and mobilize the population on the scale of total war but make the enemy be shoddy infrastructure or crappy housing or something. Instead of marshaling the entire industrial might of the nation towards turning out bombers and tanks, why not treat the whole war as a massive public works project? Make the government the employer of last resort. "If private industry cannot provide work for our good citizens, the government will employ them in something as close to their profession as possible, working towards the public good." It's unemployment benefits that don't keep you out of work and gives the government a tangible return for the money. When the economy picks up, the private sector can start hiring the workers back.

We've been cutting back on investing in infrastructure for decades, it'd be good to put some money back into our country again. Set a goal of getting us off fossil fuels over the next two decades, put government labs to work on seriously making a go of fusion power, green living, reshape our cities to be less energy intensive.

'Smallest number of votes' is misleading

[Ed Avis]

Yes, 289 is the _smallest_ number of voters that could be switched to change the result of the election. But that gives a misleading picture of how close it was. You should also consider the largest number of voters that could switch without changing the result: that is several million votes (for example, Texas voted for Bush; switch 24.999% of the votes Texas cast to Bush to Gore, and the result does not change). In other words huge numbers of people (outside Florida and other swing states) could have decided to vote for Gore (or Nader) instead of Bush and it wouldn't have made the slightest difference.

Perhaps the fairest measure of the closeness of an election is: what is the smallest number N of votes such that if you picked N individual votes at random across the whole country and flipped them, there is more than a 50% probability that the result would change?