2016 Postmortem
Related: About this forumNBC/Marist has Obama up by 8 in Iowa & PPP has Romney up by 1...Why?
Both show big numbers for Obama with early voters, but obviously NBC/Marist is showing a larger number of early voters (34%) than PPP is. Even though the early voting sample PPP tweeted says that they are showing early voters for Obama with 66%--about the same as NBC/Marist.
So Why does One poll (NBC/Marist) give Obama a +8 lead in Iowa and 67-33 margin with early voters while another, PPP, gives Romney a +1 even though they show as staggering a number of early voters for Obama as NBC/Marist?
I think the answer is easy--NBC/Marist polls both landline and cell phones while PPP polls only landlines. They are missing a lot of people (especially younger people--and maybe not so young people, I'm 48 and only use a cell phone) who have voted for Obama and use cell phones.
hrmjustin
(71,265 posts)Mr.Turnip
(645 posts)You should look at the averages more than any single poll, discounting big outliers of course.
Im guessing right now Obama is ahead in Iowa by 3 or 4 points.
WI_DEM
(33,497 posts)call cell phones, too.
LisaL
(44,974 posts)All the party affiliation percentages look similar now to 2008.
So why in the world would Obama be behind a point?
It doesn't make sense.
Drunken Irishman
(34,857 posts)woolldog
(8,791 posts)WI_DEM
(33,497 posts)woolldog
(8,791 posts)the polls are all over the place. Could be robocalling, could be cell phones, I dunno.
But I feel comfortable we have a decent lead in IA right now.
Jennicut
(25,415 posts)Which is why Rasmussen's polls are off sometimes. NBC/WSJ/Marist was probably more thorough in their polling. With those early voting #'s, Romney has real problems in IA, OH, and WI. You have to make up the difference on election day. Which is why early voting in NC is so important. We probably won't win NC but we can certainly give Romney a headache there and force them to keep some kind of presence in that state with the early voting #'s.
fugop
(1,828 posts)Is anyone worried at all about a possible Bradley Effect? I was so relieved we didn't seem to have one in 2008 (tho many on the right were hoping). I hope that isn't rearing its ugly head this year.
Not at all because we already got over the whole "voting for the first African American President". If it was going to happen then 2008 would have been it. That first great barrier was already crossed. I don't know why it would come up 4 years later and people would be dishonest about voting for a black man now, you know? I think the racists are all in the tea party group and want to wear that with badges of honor.
regnaD kciN
(26,045 posts)Last edited Sat Oct 20, 2012, 03:12 AM - Edit history (1)
...and I don't see it has made their results any more sane.
LisaL
(44,974 posts)Drunken Irishman
(34,857 posts)It's how they're coming up with their model for LV. Check out the RV, which is more inline with what other polls are showing. However, their LV model is really off.
regnaD kciN
(26,045 posts)I think they have problems with both their methodology and their LV model. As Nate pointed out a couple of days ago, leaving aside the LV/RV split, Gallup trackers tend to show more and bigger swings that other trackers, despite the whole point of a three- or seven-day tracker is to smooth out the variances. There's just something wrong with the way they conduct their polls, aside from the LV screen.
Welcome_hubby
(312 posts)What we should do is compare all pollsters who poll cell phones against pollsters who don't poll cell phones. And that comparision has always been made here: http://www.tnr.com/blog/electionate/107446/obama-pulling-ahead-in-live-interview-surveys
Nate Silver spoke about it as well.
Cicada
(4,533 posts)If you want to predict who will win why not use the superior method? And the most accurate way to use polls to predict results has been found to be this - toss out the best poll, toss out the worst poll, and use the median of the other polls. That's more accurate than using single polls. You want to be accurate, right? Then stop focusing on single polls.
thatsrightimirish
(1,391 posts)According to PPP Obama has a 66/32 advantage among early voters in Iowa.
gorekerrydreamticket
(427 posts)n/t