IA PrimR: Iowa State University: Likely outlier poll has Gingrich at only 5%

[Tender Branson]

New Poll: Iowa President by Iowa State University on 2011-11-01

Summary:

Cain:	25%
Paul:	20%
Romney:	16%
Bachmann:	8%
Perry:	8%
Other:	15%
Undecided:	8%

Poll Source URL: Full Poll Details

[Mr. Morden]

You entered this in the database as being Nov. 1, but doesn't it go by the last date the poll was taken (in this case Nov. 13)?

[Tender Branson]

You entered this in the database as being Nov. 1, but doesn't it go by the last date the poll was taken (in this case Nov. 13)?

Yeah, but it would scew up the average. There's no way Gingrich is at 5%, when all other polls have him between 15-20% during the Nov. 13 week.

Similar to this VA poll that was not entered, because Romney had a 30% lead over Cain, when other polls showed Cain ahead of Romney or tied.

At least, this poll is in the database now, but it won't screw the latest average.

[Ⓐnarchy in the ☭☭☭P!]

>Paul:   
20%

Well, here's the evidence of the previous poll not being completely out there.

What do you call something with a low roof that keeps rising? An elevator! Therefore, I shall call this the PAULAVATOR

[Eraserhead]

You entered this in the database as being Nov. 1, but doesn't it go by the last date the poll was taken (in this case Nov. 13)?

Yeah, but it would scew up the average. There's no way Gingrich is at 5%, when all other polls have him between 15-20% during the Nov. 13 week.

Similar to this VA poll that was not entered, because Romney had a 30% lead over Cain, when other polls showed Cain ahead of Romney or tied.

At least, this poll is in the database now, but it won't screw the latest average.

Stop playing God, Tender!

[Meeker]

>Paul:   
20%

Well, here's the evidence of the previous poll not being completely out there.

What do you call something with a low roof that keeps rising? An elevator! Therefore, I shall call this the PAULAVATOR

The poll that has Newt Gingrich at 5% is your validation for that wacky Bloomberg poll? Lulz

[Verily]

It's not an outlier. It's just old. Look the date, concluded Nov. 1.

[© tweed]

It's not an outlier. It's just old. Look the date, concluded Nov. 1.

bro, you gotta read the intervening posts, lest you come out with this.

[Tender Branson]

In 1 hour we should know more, because Rasmussen will release their latest Iowa poll.

[emailking]

Yeah, but it would scew up the average. There's no way Gingrich is at 5%, when all other polls have him between 15-20% during the Nov. 13 week.

It doesn't matter. Throwing out outliers screws up the average. It doesn't matter if he's at 5% or not. The existence of this poll is evidence that the average you are trying to preserve is on the high side. This is how statistics works. You can't throw out data because you think it may be a statistical outlier. Those matter!

I know it got included anyway, but I've seen this logic espoused on this forum before and its just wrong.

[Swing low, sweet chariot. Comin' for to carry me home.]

I know it got included anyway, but I've seen this logic espoused on this forum before and its just wrong.

take it from your own personal jmfcst, this poll isn't worth the bandwidth it's consuming.

[California8429]

Yeah, but it would scew up the average. There's no way Gingrich is at 5%, when all other polls have him between 15-20% during the Nov. 13 week.

It doesn't matter. Throwing out outliers screws up the average. It doesn't matter if he's at 5% or not. The existence of this poll is evidence that the average you are trying to preserve is on the high side. This is how statistics works. You can't throw out data because you think it may be a statistical outlier. Those matter!

I know it got included anyway, but I've seen this logic espoused on this forum before and its just wrong.

You certainly can throw out crappy polls from polling sites that rarely ever poll (like Vanderbilt which is complete bs) that have the leading candidate at 5% when a respected polling site (for republicans has them 30 points higher.

[Mr. Morden]

There's no reason to not include this in the database with the proper date.  I mean, if you're going to include ARG polls in the database, then the running average in the database is kind of a joke anyway.

[Tender Branson]

There's no reason to not include this in the database with the proper date.  I mean, if you're going to include ARG polls in the database, then the running average in the database is kind of a joke anyway.

Someone corrected the poll anyway and put in Nov. 13, something I would have done anyway after there were 3 polls out showing the Gingrich surge that is actually taking place there right now.

[Eraserhead]

There's no reason to not include this in the database with the proper date.  I mean, if you're going to include ARG polls in the database, then the running average in the database is kind of a joke anyway.

Exactly. Either you throw out everything suspicious or you have to keep them all basically (barring internet polls, of course).