I’d like to see their sample. If most of the calls were made in Northern California with all those hippies, I can see how Obama would be winning. Not to mention men versus women.
As they say, there are lies, damn lies and statistics.
Gallup isn’t polling any primaries, so you aren’t reading this correctly.
Zogby is doing a tracking poll in 4 states with primaries: CA, NJ, GA, and MO.
Gallup is doing a national tracking poll.
Both polls seem to trend an uptick in Obama’s popularity over the last few days.
I think by now we can conclude the poll numbers rarely reflect what will actually happen.
Ahh yes I was under the impression the Gallup poll was covering specific states, my bad.
Really? The folks at Field Research Corporation would take issue with you.
(Field Poll in CA: Clinton 36, Obama 34, 12% to Edwards and others out of the race, and 18% undecided.)
I’m pretty sure that poll is pre-Edwards dropout. Because there’s no way 12% vote for him tomorrow.
Partially yes. The Field methodology has them polling a very specific sampling over the course of the week, and Edwards dropped out on day 3 of 6 days of getting their specific sample.
The actual .pdf of their poll points out that half of the 12% “other” mark is early ballots already cast for Edwards or another candidate besides Clinton/Obama. They point out that California race clearly rests with undecideds and late deciders.
WASHINGTON (CNN) – Democratic Sen. Hillary Clinton is losing ground to Sen. Barack Obama in a national CNN/Opinion Research Corp. poll released on the eve of critical Super Tuesday presidential primaries and caucuses.
The two are virtually tied in Monday’s survey, which shows the New York senator has lost a comfortable national lead she’s held for months over Obama and other rivals.
The survey also shows Arizona Sen. John McCain as the clear Republican front-runner.
Obama, who trounced Clinton in January’s South Carolina primary, garnered 49 percent of registered Democrats in Monday’s poll, while Clinton trailed by just three points, a gap well within the survey’s 4.5 percentage point margin of error.
“Coming out of his overwhelming victory in South Carolina and followed quickly by his Kennedy family endorsements, Obama clearly has the momentum in this campaign,” said Bill Schneider, CNN’s senior political analyst.
The poll is consistent with other national surveys during the past few days. A CNN averaging of five national polls conducted in the last few days – a “poll of polls” – puts Clinton at 45 percent and Obama at 43 percent. Those five surveys were done by CNN/Opinion Research Corp., Gallup, Pew, ABC and CBS.
If the die-hard conservatives who listen to Hannity, Limbaugh and the like decide to come out in full force, McCain could have some trouble.
So one polling organization in one state beats the odds and that invalidates my entire statement?
Yeah!
er,
No!
er,
Nevermind! (I used you as a springboard to discuss the Field Poll, which I find kind of fascinating to look over…how dare you blow up my illogical opportunism!)
Heh, no worries. But I admit, that was a cool link. I wonder what makes their methodology so much more accurate then everyone else’s?
Rasmussen Reports has Obama now pulling nearly even with Clinton for the nomination. Two days ago, this split was about 60/40 Clinton. Last night it evolved to about 55/45. At the rate it’s going, Obama will have pulled ahead by the time polls open tomorrow.
He has eclipsed her in this “prediction market” (not a survey or poll, mind) for the California primary, 56.9 to 48.5.
The proper daily California survey has not been released yet. Yesterday, they were in a dead heat.
How accurate is that market approach?
It depends on the number and predictive skill of the participants. Since those figures are not public, your guess is as good as mine, frankly.
However, here are the latest polls for all Super Tuesday primaries:
http://www.realclearpolitics.com/epolls/2008/president/democratic_primaries.html
You may note a trend here: The more recent poll almost always seems to give Obama better odds; with the exception of New York and Tennessee, he is either very close or leading.
All that remains to be seen is if it’s the Tom Bradley Effect again.
Ask Iowa.
Not very.
edit: The markets aggregate information that’s already public, and that’s it. They don’t offer any particular insight into who might actually win any given contest. Check out the markets before NH - they had Hillary far below Obama, just like everyone else.
/cynic on
And in response, the day before Super Tuesday, Hillary publically got choked up and let a tear slide down her eye.
Hey, worked in New Hampshire.
/cynic off (ok, so that switch doesn’t work so well. ;) )
A reasonably decent political market, in theory, can be a good way of taking the pulse of a broad number of polls, with adjustments for momentum and likely errors in those polls. Such a market would also have a relatively small amount of observer bias built in (i.e. it wouldn’t be overly skewed to a pro-Obama or pro-Clinton view).
I’ve followed the InTrade markets a bit over the last few months, and I’m not convinced they meet such criteria very well. Perhaps the other markets are better, but Slate seems to highlight InTrade’s results.
Things that could make political markets results weak:
- Few participants
- Low volume (therefore, creating little incentive for participants to make accurate judgements)
- Structural issues (IIUC, InTrade’s method of settlement discourages folks from taking the other side of long-shots, thus meaning InTrade tends to overstate the probability of long-shot candidates).
- Uninformed participants
- Participants trying to game the system. If the value of a high InTrade rating is greater than the cost to achieve that (by indiscriminately buying shares of your favored candidate), then there’s an incentive to game the system.
There’s probably other issues as well.