Guestimating the election with twitter(election.setfive.com)
election.setfive.com
Guestimating the election with twitter
http://election.setfive.com/
2 comments
Biased sampling.
Yeah, basically you should take most of these polls with a grain of salt. Other highly biased polls:
http://www.7-election.com/
First of all, apparently 7-11 doesn't have locations in most of the "red states", which already shows bias. And there's a huge difference between voting for a candidate and buying a cup with their face on it to advertise your choice to the world. Nice marketing gimmick though.
http://www.alleyinsider.com/2008/11/in-landslide-john-mccain...
Apparently McCain supporters are behind the times with their AOL accounts. I haven't come across anyone who uses AOL in years.
Even the "real" polls have all sorts biases, like the "cellphone effect":
http://www.fivethirtyeight.com/2008/11/cellphone-effect-cont...
Again, McCain supporters tend to still use landlines over cellphones.
http://www.7-election.com/
First of all, apparently 7-11 doesn't have locations in most of the "red states", which already shows bias. And there's a huge difference between voting for a candidate and buying a cup with their face on it to advertise your choice to the world. Nice marketing gimmick though.
http://www.alleyinsider.com/2008/11/in-landslide-john-mccain...
Apparently McCain supporters are behind the times with their AOL accounts. I haven't come across anyone who uses AOL in years.
Even the "real" polls have all sorts biases, like the "cellphone effect":
http://www.fivethirtyeight.com/2008/11/cellphone-effect-cont...
Again, McCain supporters tend to still use landlines over cellphones.
Very interesting AOL poll...
Another link: twitter users can “vote” on twitter using http://twitvote.twitmarks.com/ - Currently 15002 for Obama, 2681 for McCain.
Another link: twitter users can “vote” on twitter using http://twitvote.twitmarks.com/ - Currently 15002 for Obama, 2681 for McCain.
It's even worse than that.
Most polls "weigh" the respondents based on, well, judgment. So a poll might be weighed 60-40 dems to pubs. Or 65-35. It varies.
And then the polls each have their "likely voter model", which means all the questions they ask to figure out if you're going to vote or not. Many of these are untested.
I think Obama's going to win by 3-4 percent nationwide, but quite frankly the polls could be way out of whack in a lot of spots. He could pull out a ten-point lead, or it could get really close in spots.
Polling for future behavior is a very inexact science. Heck -- even exit polling isn't that great.
Most polls "weigh" the respondents based on, well, judgment. So a poll might be weighed 60-40 dems to pubs. Or 65-35. It varies.
And then the polls each have their "likely voter model", which means all the questions they ask to figure out if you're going to vote or not. Many of these are untested.
I think Obama's going to win by 3-4 percent nationwide, but quite frankly the polls could be way out of whack in a lot of spots. He could pull out a ten-point lead, or it could get really close in spots.
Polling for future behavior is a very inexact science. Heck -- even exit polling isn't that great.
While it does seem it is definitely a very biased sampling, we have decided not to rework the points system to compensate. We just watched New Mexico go from a dark blue(meaning it had a 20% democratic lead) go to white which mean it just broke dead even with the republicans. We'll post up time lapse of the map after today. It may be interesting to see if there any trends related to time for the map too.
Technohipsters favor Obama!
It's true, too bad technohipsters don't run this country (hint: their white haired parents do)
It is ugly and open to bias but we're hopping it might show something interesting.
We're also planning to take snapshots of the map and assemble a time lapse for Wednesday.