Log in

View Full Version : Election prognostication



The Storyteller
10-31-12, 9:39am
In another thread, Gregg responded to me about the following (paraphrasing here) that there are plenty of polls that say anything you want them to say.



Are you sure about that?

Quite sure Obama has a high probability of winning. Closest prognosticator has him with a 62 % chance of winning reelection. PEC gives him better odds at 9 to 1 probability. Nobody gives Romney anywhere near a 50% chance.

http://www.intrade.com/v4/home/

http://research.uvu.edu/DeSart/forecasting/october.html

http://election.princeton.edu/

http://fivethirtyeight.blogs.nytimes.com/

http://votamatic.org/

PEC had it dead on in the electoral college split in 2004.

But these are not polls, nor are they mere articles by mere shoot-from-the-hip journalists or commentators. These are professional statisticians or odds-makers whose job it is to predict the future based on current trends. Most polls are mere snapshots of the moment, telling you what would happen if "the election were held today". These sites take current and past polls, apply them to computer models based on history, and give percentages of probability of what will happen on election day. That's what I understand, anyway, because I just learned about them the other day and I'm still trying to understand how they work.

Of course, as one site mentions, these are predictions, not prophecies. For instance, one site gives Obama a 98% chance of winning. That sill gives Romney a 2% chance. It isn't 100%. That site is an outlier though. Most give Obama anywhere from 63% to 89% probability of winning reelection. I have yet to find one that gives Romney anywhere near a 50/50 chance of winning. This is based on where the race has been trending in a few battleground states.

My favorites are Intrade and the Princeton site. Princeton comes off as very methodical and scientific. Intrade is a betting site, so this one the odds makers set their odds based on how people worldwide are betting. Those folks are putting their money where their mouth is, sometimes a lot of it. I don't know that much about gambling, but seems to me those folks have a tendency to really know the nitty gritty behind probabilities of winning their bet.

These sites are not without controversy. Just google Nate Silver (538) in relation to Joe Scarborough or David Brooks and you will see what I mean.

Anyway, I thought with so many political junkies on this site, you might be interested in looking into these. I find them fascinating.

The Storyteller
10-31-12, 9:54am
Here is why these predictors are different from journalists, commentators, and polls. Polls want to get it right at the moment. The day before election, the closest poll wins and looks good. The one that is 5%+ off has egg on their face. Commentators have their own political axe to grind, so they look at the race through very tinted lenses, either red or blue. They get paid by people reading their column, so their primary interest is in entertaining while they comment from their POV.

Journalists want you to read their stuff, too, but their bent is to entertain through information. They have their biases, too, only different. Their bias is the exciting stuff. A split election is a reporters wet dream, especially a split White House (just google Romney Biden White House and you will see what I mean). They love comebacks, underdogs, and controversy. They love to see the race tight. They are also very lazy, picking up shorthand from their fellow reports. Romney is unlikeable and ignorant of the plight of the common man. Gore is stiff. Obama is aloof. None of which is true, of course. But they are mythologies that reporters hang their hats on that are very convenient. That is why the 47% comment and the Obama performance in the first debate played so well for them. They matched the reporters' bias.

They also aren't very good at math, and that is the main advantage of prediction sites. Those sites are interested in getting it right on election day. The prognosticator that is completely wrong and has been for weeks is probably out of the business, or at the very least has a great deal of egg on their face that they might never live down. There is also a race to see which of them is most right. In 2004 it was PEC, the Princeton site. In 2008, it was 538. Both called it almost exact, state by state, in their respective successes.

It will be interesting to see who is most right this time. Here is an old Wall Street Journal article on it from 2004:

http://synapse.princeton.edu/~sam/wsj_4nov04_pollcalc-followup.pdf

Gregg
10-31-12, 9:58am
When we found out DW was pregnant with our last DD I immediately looked her in the eye and said, "its a girl". I did have a little feeling about that, but mostly I figured it was 50/50 and if I got it right I could brag for the rest of my life and if I got it wrong everyone would forget all about it by the time the baby was a couple hours old. Kind of a no lose deal. That is the same position all the prognosticators are in.

Storyteller, I agree that it is interesting to see how all these different people come up with their predictions. Fox and all the other conservative media have tracked down plenty of sources with legitimate data who are predicting a Romney victory. Truth is none of them have a crystal ball. It's a close race. I don't know who will win and get no comfort from a third party source telling me what I want to hear. I've decided to campaign for a couple candidates I believe in. On election day I will know I was engaged in the process even if my candidates don't win.

Just for the heck of it I googled "predicting a Romney win". It returned "about 120,000,000 results". I'm guessing that somewhere on just the first few pages there are multiple groups analyzing all kinds of historic data and using that to predict a Romney victory. Did you know that 98% of the time when the temperature on election day in Springfield, MO is above 52* F. the Republican candidate will win?

The Storyteller
10-31-12, 10:11am
I'd love to see those sites Fox has found, Gregg. It interesting to see the different takes so far. Intrade I have heard has a Republican lean. I did read that one Chronicle article, and found they author was a pilot who writes right leaning columns sometimes. Hardly a mathematician or statistician. I'm sure there a plenty of people predicting a Romney win, but who are they and what are their qualifications?

One of the Romney campaign's tactics seems to be to perpetuate the myth of momentum, I guess to keep the stories positive rather than 47% and FEMA, so I'm sure that could be where Fox is coming from. Supporting their candidate's campaign.

Gregg
10-31-12, 10:34am
I've never met anyone who is actually qualified to predict the future. No one with a working crystal ball. I know several people that do a pretty good job of figuring out what the most likely possibilities are, but that's about it. As I said above, I get no comfort googling until I find sites that will tell me what I want to hear. Out of the 120 million hits I feel fairly sure there are some sites that reflect the opposite results of Princeton or InTrade. I just really don't have anything that motivates me to dig around for them.

Storyteller, you're obviously an ardent Obama supporter. When you post a series of links showing your candidate besting the opposition my only real reaction is to think that's about what would be expected (same goes for most all of us). The methodology used in various predictions is interesting, we agree on that. I just don't see much value in guessing the outcome when we will know the real story in less than a week. There just isn't any benefit to it. Myself, I'm perfectly content doing what I can to help insure the outcome I desire and then sitting back and watching as the results come in. YMMV.

creaker
10-31-12, 10:51am
People watch a close race where the runners are jockeying for position. It's in the media's interest to keep it close and have each candidate look as if they are just pulling ahead or just falling behind. I'm guessing that the polls get cherry picked to make that happen.

CathyA
10-31-12, 11:00am
I do wonder about the predominantly democratic east coast states that have such damage. If I lost my house and community, the last thing I'd be thinking about was the election.

freein05
10-31-12, 11:13am
I think a lot of the problem is these polls are using nation wide polls and forgetting we elect the president by the electoral collage. Obama is planning on winning by the number electoral collage votes he gets which is 270 not the popular vote.

The Storyteller
10-31-12, 11:16am
Storyteller, you're obviously an ardent Obama supporter. When you post a series of links showing your candidate besting the opposition my only real reaction is to think that's about what would be expected (same goes for most all of us).

I am very much an Obama supporter. I was very worried that the things I most value (the environment and eco-agriculture) would not do well under a Romney administration. Plus his plan to dump FRank/Dodd and deregulate Wall Street would be disastrous for the economy.

I was believing the media that the race was a tossup, so I was quite worried. I then heard on NPR about the science of election probability prediction, which has been around for a lot longer than polling, since the 1800s in fact. In all that time, they have only called three presidential elections wrong.

Most of these services I posted about are relatively young. As I said, I haven't found any that come close to giving Romney an edge. That gives me personal comfort.

Alan
10-31-12, 11:42am
I've long believed with a relatively stable percentage of R's & D's in the country, the key to any federal election is the enthusiasm level displayed by the individual groups. Enthusiasm brings voters to the polls which we saw in 2008 and 2010. I think it will be the deciding factor in 2012 as well. According to today's CBS/NYT/Q-Poll (http://www.cbsnews.com/8301-250_162-57542715/poll-obama-holds-small-ohio-edge-fla-va-tight/?pageNum=3&tag=page)in the most hotly contested swing states (which I believe carries over to all others as well):


http://media.hotair.com/greenroom/wp-content/uploads/2012/10/cbs-nyt-enthusiasm1.jpg

The Storyteller
10-31-12, 11:43am
I should mention one thing I learned while studying these sites... the notion that when there is an incumbent running, undecideds at the end break for the challenger isn't true. Statistically, they tend to break about even.

The Storyteller
10-31-12, 11:46am
I've long believed with a relatively stable percentage of R's & D's in the country, the key to any federal election is the enthusiasm level displayed by the individual groups. Enthusiasm brings voters to the polls and we saw that in 2008 and in 2010. I think it will be the deciding factor in 2012 as well. According to today's CBS/NYT/Q-Poll (http://www.cbsnews.com/8301-250_162-57542715/poll-obama-holds-small-ohio-edge-fla-va-tight/?pageNum=3&tag=page)in the most hotly contested swing states (which I believe carries over to all others as well):


http://media.hotair.com/greenroom/wp-content/uploads/2012/10/cbs-nyt-enthusiasm1.jpg

One thing the prognosticators emphasize over and over again... never look at just one poll. You have to look at all of them.

As Gregg said, you can find a poll somewhere that will say just about anything. Like Gallup giving Romney a 5% lead nationally, for instance.

Alan
10-31-12, 11:54am
One thing the prognosticators emphasize over and over again... never look at just one poll. You have to look at all of them.


I agree. And they all say pretty much the same thing regarding the enthusiasm advantage.

http://www.cbsnews.com/8301-250_162-57542715/poll-obama-holds-small-ohio-edge-fla-va-tight/?pageNum=3&tag=page
http://www.realclearpolitics.com/articles/2012/10/01/poll_shows_republicans_with_enthusiasm_advantage_1 15625.html
http://gop12.thehill.com/2012/09/poll-suggests-enthusiasm-favors-romney.html
http://www.washingtonpost.com/blogs/election-2012/wp/2012/10/30/new-polls-give-gop-enthusiasm-edge/
etc., etc.

The Storyteller
10-31-12, 12:19pm
I agree. And they all say pretty much the same thing regarding the enthusiasm advantage.

http://www.cbsnews.com/8301-250_162-57542715/poll-obama-holds-small-ohio-edge-fla-va-tight/?pageNum=3&tag=page
http://www.realclearpolitics.com/articles/2012/10/01/poll_shows_republicans_with_enthusiasm_advantage_1 15625.html
http://gop12.thehill.com/2012/09/poll-suggests-enthusiasm-favors-romney.html
http://www.washingtonpost.com/blogs/election-2012/wp/2012/10/30/new-polls-give-gop-enthusiasm-edge/
etc., etc.

Fair enough. Conservatives just LOVE Romney. ;)

I'm sure that is factored in to these models, though.

Gregg
10-31-12, 12:20pm
I was very worried that the things I most value (the environment and eco-agriculture) would not do well under a Romney administration. Plus his plan to dump FRank/Dodd and deregulate Wall Street would be disastrous for the economy.

I have decided to support Romney simply because I think he will be more likely to establish an environment in which jobs are created and revenues rise. In other words, I put the economy as my top priority and after running the candidates through my own filter the arrow pointed to Romney. The government needs to bring in more money, period. I do not believe either party will do anything to reduce spending so increasing revenue is the only way this country will remain solvent.

I am very nervous about the effect a Romney victory might have on several other critical factors (including the environment and related subjects). I don't think he would deliberately set out to rape and pillage the environment, but charging ahead for near term gains without considering all the effects can be disastrous. My priorities are where they are for what I believe are good reasons, but I admit this could end up being a short sighted decision.

The Storyteller
10-31-12, 12:36pm
I think a lot of the problem is these polls are using nation wide polls and forgetting we elect the president by the electoral collage. Obama is planning on winning by the number electoral collage votes he gets which is 270 not the popular vote.

Exactly. While some of these folks make predictions about the popular vote (most of them give Obama the edge, but only slightly... see Intrade for example), their focus is on the electoral college. So, they look very in-depth at state-by-state polls, especially (some almost entirely) at the swing states. For instance, does it really matter if Obama wins California by 32% and Romney Texas by 24%? They only need to win them 50% + 1.

iris lily
10-31-12, 12:42pm
I don't give a ch*t about predictions, I'm interested in the real deal. But you guys, carry on discussing it, no skin off me. And certainly the pollsters that do the best predicting will be able to command higher fees for the next Presidential election. It's important to them. Not to me.

redfox
10-31-12, 12:57pm
Here's my prediction:
Some day this entire mess will be OVER!! I cannot wait. As Molly Ivins called it, Our Quadrennial Circus. Oy.

Alan
10-31-12, 1:24pm
Here's my prediction:
Some day this entire mess will be OVER!! I cannot wait. As Molly Ivins called it, Our Quadrennial Circus. Oy.
You're not the only one to feel that way.


http://www.youtube.com/watch?v=OjrthOPLAKM&feature=player_embedded

loosechickens
10-31-12, 4:10pm
Surfing the net on the road in AZ, on my iPhone. I like fivethirtyeight.blogs.nytimes.com. I know many conservatives are dismissing him as "pro liberal" because he predicted Obama win in 2008 and today. However, they are conveniently forgetting that liberals said he was "pro conservatives" when he steadfastly and almost alone predicted the Republican 2010 sweep.

He crunches statistics, just as he used to do for baseball team owners. He wants to have his predictions be accurate. And that is way more important to him than biasing his predictions. Which is why I trust him. I trusted him in 2008. I trusted him in 2010 even though I didn't like what he said, and I trust him now.

In end, we will see if he is correct this time, but I will bet on him.

ApatheticNoMore
10-31-12, 6:02pm
Really que sera sera.

redfox
10-31-12, 10:21pm
My brother posted this US News article:

http://www.usnews.com/opinion/blogs/robert-schlesinger/2012/10/31/why-obama-will-beat-romney-state-polls-and-the-electoral-college

and, yeah, what you said, ApatheticNoMore!

The Storyteller
10-31-12, 10:29pm
You're not the only one to feel that way.


http://www.youtube.com/watch?v=OjrthOPLAKM&feature=player_embedded

Cute!

The Storyteller
11-1-12, 9:15am
Really que sera sera. The same can be said of the weather, but that doesn't mean I don't check the forecast.

Especially when the outcome can have a profound effect on my farming.

Alan, that vid is EVERYWHERE now. Saw it on GMA this morning and it was last night's Moment of Zen on the Daily Show.

The Storyteller
11-5-12, 9:36am
Okay, after reading this, I am slightly less optimistic of an Obama win...

In Defense of Nate Silver, Election Pollsters, and Statistical Predictions (http://www.wired.com/opinion/2012/11/why-predictions-and-statistical-models-are-necessary-and-good-for-democracy/), by Zeynep Tufekci

A one in five chance of a Romney win sounds a lot closer than a 20% chance.

And here is a blog post from PEC that compares the various sites and explains their approaches...

Comparisons among aggregators and modelers (http://election.princeton.edu/2012/11/04/comparisons-among-aggregators-and-modelers/#more-8408), by Sam Wang.

Later today, I'm going to post screen shots of each prediction site and compare their predictions with what happens Tuesday.

I watched the political talking heads Sunday morning for the first time in over a year. After following these sites, it really stood out just how stupid they are. They make statements based on little or no facts at all. Among the dumber things I heard... "I don't care what the polls say. When you get on the ground out there, it just FEELS like Romney is going to win." --David Gurgen on CNN and CBS.

Alan
11-5-12, 10:35am
I watched the political talking heads Sunday morning for the first time in over a year. After following these sites, it really stood out just how stupid they are. They make statements based on little or no facts at all. Among the dumber things I heard... "I don't care what the polls say. When you get on the ground out there, it just FEELS like Romney is going to win." --David Gurgen on CNN and CBS.
I have little confidence in the various poll's results but I do care about what they say. For example, over the weekend CNN's Ohio poll showed a 49-49 tie in the state using a D+10 methodology. For the life of me I can't figure out where the +10 came from. It's not bourne out by the 2008 actual vote and certainly not by the 2010 mid-term vote.

The only two possibilities that come to mind are: a) The media has a vested interest in forcasting a close election in order to keep people tuned in, or b) The media is hoping to suppress one side's participation in what they're representing as a potential lost cause.

One thing I'm certain of, the Obama 2012 campaign cannot match the enthusiasm and energy of it's 2008 "historic" campaign, while the passion exhibited by conservatives during the 2010 mid-terms is still readily apparent in my little part of Ohio, and I suspect the entire nation. So, on that point, I'll agree with Mr. Gurgen.

Gregg
11-5-12, 10:47am
I am now officially in the vote, wait and see camp. Going to watch the Food Network until Tuesday evening and just do quick checks of headlines on the web in the mean time. I can't say I feel anxious, but I am looking forward to Wednesday.

The Storyteller
11-5-12, 12:18pm
I realize it is just one poll, Alan, but I did see one this weekend that showed Dem enthusiasm higher than Rep.

But the aggregators consistently warn about looking at individual polls.

The Storyteller
11-5-12, 3:29pm
Okay, found one that projects Romney.

http://www.colorado.edu/news/releases/2012/10/04/updated-election-forecasting-model-still-points-romney-win-university

It is based on economic factors and over a month old, but still, there you have it. We'll see which who prevails tomorrow. Polls in the battleground states would have to be very, very wrong for Romney to pull this off...

http://www.simplelivingforum.net/attachment.php?attachmentid=1042&d=1352143640

This is getting more fun than watching the election itself. Like Moneyball for politics.

oldhat
11-5-12, 4:14pm
Surfing the net on the road in AZ, on my iPhone. I like fivethirtyeight.blogs.nytimes.com. I know many conservatives are dismissing him as "pro liberal" because he predicted Obama win in 2008 and today. However, they are conveniently forgetting that liberals said he was "pro conservatives" when he steadfastly and almost alone predicted the Republican 2010 sweep.

He crunches statistics, just as he used to do for baseball team owners. He wants to have his predictions be accurate. And that is way more important to him than biasing his predictions. Which is why I trust him. I trusted him in 2008. I trusted him in 2010 even though I didn't like what he said, and I trust him now.

In end, we will see if he is correct this time, but I will bet on him.

I agree--Silver is an analyst of polls who seems to know his stuff. He has an impressive track record of accuracy. I just heard him being interviewed on NPR's "On the Media," and he admitted that one of these days he's going to be wrong. But I got the feeling he doesn't think this is going to be that time.

Gregg
11-5-12, 4:24pm
Perhaps the most storied prediction method of all. It's not a poll, but is 94.44% (17/18) accurate since 1940...

http://www.thepostgame.com/features/201211/washington-redskins-world-series-presidential-election-forecast-obama-romney

The Storyteller
11-7-12, 10:39am
Great article on my new heroes...

Nate Silver-Led Statistics Men Crush Pundits in Election (http://www.bloomberg.com/news/2012-11-07/nate-silver-led-statistics-men-crush-pundits-in-election.html)