- by Lyndon_Henry, Blogger
- 10/20/2012 1:12:38 PM
... I just glanced over at Election Analytics, which now has its forecast up for how the election would turn out if the votes were cast today. If we're to believe this data, Obama would still win, but not as neatly as Election Analytics showed him doing earlier this month.
Beth, first I wanted to say I'm glad you found my earlier comments from another thread interesting enough to quote in your blog article.
Re: the recent Gallup poll finding of a 6-7 percentage point national lead for Romney, Nate Silver has a very interesting analysis in his 538 blog:
He finds the Gallup poll results to be a significant outlier from all the rest and has some interesting analysis about it. Also gives a peek of insight into how his model works.
The blog has another entry that focuses on some swing state polling results:
His analyses provide lots of insight into both polling issues and political analytics.
- by SethBreedlove, Data Doctor
- 10/19/2012 6:00:32 PM
I believe in the anaylitical tools, however one has look at analytically. Where was the poll taken, how were people selected? An online viewer submitted poll on Fox news is going to be very different than on MSNBC.
Also polls are just a snap shot of a particullar time. What is accurate today at 5 P.M. can radically change by 10 P.M. in the flash of a news story.
A good article on this is Confused by all the polls? Pollsters explain the variation
Also no matter what is going on, one should never underestimate the stupidy of the masses.
- by louisw900, Blogger
- 10/19/2012 4:59:54 PM
Really nice to learn that the All Analytics community for the most part does not believe in these tools for election prediction, now if we could spread the word to the major news and cable channels.
- by Nnanci, Blogger
- 10/17/2012 5:51:01 PM
Beth, - IMO the mood tipped in favor of Obama in that debate especially at the last minute when he brought in the attack on Romney over the closed door comment about 47%. However the win itself may not be as strong for Obama because the debates shake him as well, he's tackled over issues such as the Libya one and on jobs as well...some people may change their minds along the way. Also if these polls are based on uncommitted voters.... the result may have to deviate from what was polled under different circumstances.
- 10/17/2012 5:32:13 PM
@kichecko, I just glanced over at Election Analytics, which now has its forecast up for how the election would turn out if the votes were cast today. If we're to believe this data, Obama would still win, but not as neatly as Election Analytics showed him doing earlier this month. I'm not sure what this says about the debate results, but it's interesting to watch! On Oct. 2, Obama would have had a 1.000 probability of winning with 347.2 expected electoral votes, according to Election Analytics. Now the tool shows Obama with a 0.975 probability of winning (if election held today) with 293.5 electoral votes.
- by Nnanci, Blogger
- 10/17/2012 4:40:32 PM
So true voter sentiments shift every other minute. Its amazing just how influential the presidential debates actually are. For sure if these debates are held like a day or two to elections they will influence the result.
However because there will be a time lag from the close of campaigns to the voting day, voters will have got over the wind of these debates and may rely on some event or other that will occur towards the end (maybe even a poll)...unless a given candidate is decisively so popular that no event can influence the people otherwise.
- by Jeff, Data Doctor
- 10/17/2012 4:17:44 PM
I guess i can't imagine the set of inputs that would give a an accurate prediction of any election. What would they be? Dem v. Rep , Demographics like race, income etc. But then what?
- by Noreen Seebacher, Blogger