Moderators: jcmanson, Sly Fox, BuryYourDuke
LUconn wrote:http://fivethirtyeight.blogs.nytimes.com/And when he's right, again?
Nate Silver has Obama's chances to win at 90.9%. Good example of a statistical formula putting garbage in, getting garbage out. What a joke.
LUconn wrote:The beauty is, nobody will ever know if he's right or wrong. It's quite a racket. He leaves open the possibility of either candidate winning. No matter who wins, he was right.He takes the available polling information and provides the the statistical likelihood of the overall result due to how the polls are moving. It's not a prediction, it's statistical analysis done right. It would be arrogant to claim that the polls "prove" one thing or another instead of showing them for what they are and analyzing where the trends are likely leading.
* 40 percent chance of no snow; just light rain showers, if anythingI'll take the second approach on something that we don't have systems in place to predict with surety. That's what Silver is doing. He's taking his background of statistical analysis in baseball and applying those principles to monitoring polls for elections. When combining the polling trends with margin of error, which reduces below the level of each individual poll with more polls being in play, it's statistically sound to provide the likelihood in a percentage format instead of pretending that one thing is sure and another isn't.
* 20 percent chance of snowflakes mixing in with rain, but no accumulation
* 20 percent chance of a slushy coating of snow, mainly on grassy areas
* 15 percent chance of a coating to 1” of wet snow, mainly on grassy areas
* 5 percent chance of 1-4” of wet snow
Humble_Opinion wrote:Scorch - I don't think Romney takes NV. If the Republicans couldn't overcome the union vote out there in 2010 when they tried to unseat Reid, then I don't see it happening this time around. I think Romney comes up just short of PA and WI.I think it depends on how much influence Steve Wynn and Vegas people like him can wield this time around. They've pulled no punches when addressing the effect (current and future) of the current administration on their businesses. If they can convince enough people in their employ (which may or may not happen), I think Romney can pull it out. However, Nevada also has a significant Libertarian population. Johnson may get enough of a vote there to change that outcome.
NotAJerry wrote:Yes, I am very familiar with statistics and math. But the flaw in his analysis is the fact that he can only analyze the numbers available.. If those numbers are worthless, he comes up with a worthless percentage. Which is why I said garbage in garbage out. 90% is cartoonish.LUconn wrote:The beauty is, nobody will ever know if he's right or wrong. It's quite a racket. He leaves open the possibility of either candidate winning. No matter who wins, he was right.He takes the available polling information and provides the the statistical likelihood of the overall result due to how the polls are moving. It's not a prediction, it's statistical analysis done right. It would be arrogant to claim that the polls "prove" one thing or another instead of showing them for what they are and analyzing where the trends are likely leading.
We need far more of that kind of approach and far less of people claiming to know for sure based on things that either aren't verifiable or haven't been verified yet.
Let's draw the comparison with weather forecasting. Most of the TV folks around DC aren't saying a word about potential snow tomorrow. Meanwhile, the Capital Weather Gang has put the following on their blog:
* 40 percent chance of no snow; just light rain showers, if anythingI'll take the second approach on something that we don't have systems in place to predict with surety. That's what Silver is doing. He's taking his background of statistical analysis in baseball and applying those principles to monitoring polls for elections. When combining the polling trends with margin of error, which reduces below the level of each individual poll with more polls being in play, it's statistically sound to provide the likelihood in a percentage format instead of pretending that one thing is sure and another isn't.
* 20 percent chance of snowflakes mixing in with rain, but no accumulation
* 20 percent chance of a slushy coating of snow, mainly on grassy areas
* 15 percent chance of a coating to 1” of wet snow, mainly on grassy areas
* 5 percent chance of 1-4” of wet snow
Hold My Own wrote:Oh Lord, the title on Drudge keeps changing. It went from "Boom for O" to "Exit Polls tight." Man I had a better feeling a while ago then I do now. They also posted that headline I was talking about in regards to Exit polls though
NotAJerry wrote:I do wonder what constitutes calling a state when the polling numbers are coming in. Right now CNN has WV called for Romney despite the percentage being 49/49, only 1% of precincts reporting, and Obama having 40 more total votes.
Humble_Opinion wrote:Morons on CBS are trying to say that the exit polls are looking good for Obama in NC, because the black vote turnout looks similar to what it was in 2008.Current vote count in NC has Obama within 8k with over 1.5 million votes counted. How are they morons for saying that it's looking good for him if he's very close with 34% of votes counted in a state that was considered a lock for Romney coming into today?
LUconn wrote:http://fivethirtyeight.blogs.nytimes.com/Looks like his garbage was pretty accurate after all.
Nate Silver has Obama's chances to win at 90.9%. Good example of a statistical formula putting garbage in, getting garbage out. What a joke.
4everfsu wrote:Their predictions have been accurate in every presidential election since 1980.0-1. That's the record of their system as an actual predictor instead of "predicting" things that happened in the past.
http://dailycaller.com/2012/08/23/unive ... omney-win/
- By LU Armchair coach
- By Sly Fox
- By TH Spangler