The most accurate pollsters

Wow, according to Markos, by my giving Research 2000 the "Zogby Award" for their 2008 polling, I've committed a "serious accusation" in that I "claim that Research 2000 cooked its numbers" in their polling. Yikes, no fun with that guy until he gets his needed vacation...

One of his commenters gives a good definition of why Research2000 gets the award:

On October 25, R2K had the race at 52-40, while the RCP average (which didn't include the R2K tracker) was at 50.4-42.4.  Then R2K suddenly tightened when the cumulative trend didn't.  R2K was down to 50-45 on October 30 -- a 7 point drop in Obama's lead in 5 days.  But the RCP average on Oct. 30 was 49.7-43.8 -- only a 2 point drop in Obama's lead.  

Then from Oct. 30 through the election, R2K basically didn't change, ending up at 51-46, still a 5 point lead, while the RCP average showed Obama's lead opening up again, to 52.1-44.5.

From Oct. 25 to Nov. 4, the RCP average went from Obama +8.0 to Obama +7.6.  Over the same period, R2K went from Obama +12 to Obama +5.  I can't point to how this was done, but I got the distinct impression that R2K was tired of being a pro-Obama outlier as the election approached and decided to tighten up its numbers.

I hate to say that, because I think it was so great that Markos paid for this tracking poll and made it so transparent for all of us.  But if you look at the R2K trendlines, you get the impression that the race was an Obama blowout for 6 weeks and then tightened at the end.  No other poll shows that and I don't think I believe it.

Sounds Zogby-like to me (without needing the hyperventilating accusations in response). Markos, why dig? Every criticism I made about the polls sample turned out accurate-- end of debate. In the public poll of polls tracking I did, I wound up not even weighting Research 2000's numbers, once they slid to the middle. They helped formulate my opinion of taking the average and giving an even allocation of the undecideds for predicting a 6.4 margin for Obama.

It is useful to turn to looking at whom among the national pollsters got it right. Mark Blumenthal has the list, but doesn't give the bragging rights, so I will rank them, according to getting closest to the 6.6 margin and the candidate amounts breaking ties:

1.   CNN/Opinion Research
1.   Ipsos/McClatchy
3.   Democracy Corps
4.   Fox News/Opinion Dynamics
5.   Pew Research
6.   ARG 
7.   Harris Interactive
8.   NBC/WSJ
9.   Rasmussen
10.  You/Gov Polimetrix
11.  Battleground (Lake)
12.  Research 2000
13.  Diego/Hotline  
14.  ABC/Washington Post
15.  Marist College
16.  CBS/NYTimes
17.  Reuters/C-SPAN/Zogby
17.  Gallup
19   Battleground (Terrance)
Opinion Research and Ipsos/McClatchy each only missed the final number by .4, so they get the top award. Democracy Corps got the margin pretty close, but was off by more with each candidate. The dog, loser, is Battleground's Terrance. What a humiliation. This is the polling group that nailed it in '04, and was the top in '96 as well. The runner-up dog is Gallup. Their "expanded voter" model was a failure, and Zogby, lol.

Tags: 2008 (all tags)



Re: The most accurate pollsters

I am really tired of this pissing contest.  Both of you need to grow up - NOW.

by baghdadjoe 2008-11-21 11:18AM | 0 recs
Re: The most accurate pollsters

"When I met you, you were an organizer with a law degree too afraid to practice law."

"Without me, you'd be a minor figure in the blogosphere. What percent of your hits come from my site? Would you even have an active discussion if it would for people I've banned?"

by Carl Nyberg 2008-11-21 11:46AM | 0 recs
Re: The most accurate pollsters

Yadda yadda yadda.

Anyway, I'm really glad to see Zogby get egg on his push-polling face.

by Jess81 2008-11-21 11:23AM | 0 recs
this ranking is nonsense

As the pollster post notes, it silly to rank them since the top 13 are within the margin of error.  Getting +5 when reality was +6.8 is perfectly fine.

That is, even if every one of these pollsters did exactly the right thing, we should end up with a ranking from 1 to 13 that would have nothing to do with their method but just random chance.  

That Armstrong doesn't acknowledge this is a hint that this is just a post to annoy people.

by John DE 2008-11-21 11:24AM | 0 recs
Re: this ranking is nonsense

Exactly right.  If reality matches the the poll within the margin of error why should we care?

Well, maybe we'd care if the margin of error was, oh, 25% or sumthin.

FP posts like this are making me wonder why I show up here at all anymore.

by Fluffy Puff Marshmallow 2008-11-21 12:16PM | 0 recs
Re: this ranking is nonsense

You beat me to it.  At this point, Obama's margin is just under 7%.  A poll of 1000 voters has a MOE of just over 3%.  And the only final polls outside the +5 to +9 range were Gallup, Zogby, and Battleground-Tarrance.

What this says is that polling is getting quite good.  Making a big deal about rankings within that +5 to +9 envelope is stupidity.

If Jerome wants to say something about the quality of R2K's polling for Daily Kos, they did lots and lots of polls in addition to the daily tracker.  There's a pile of data there if Jerome wants to come up with a solid conclusion.  Either he's simply lazy, or he's not interested in coming up with such a conclusion.

Either way, it's intellectually dishonest of him to equate R2K with Zogby on the thin basis he's provided.

My estimation of Jerome has done a nosedive this year.

by RT 2008-11-21 02:32PM | 0 recs
Re: The most accurate pollsters

Isn't the final popular vote margin going to be around 6.8%?  That would move Fox (!) to #1.

Speaking of Kos, the partisan ID stayed at +9 in their tracking.  The CNN exit poll had a 39-32 ID edge for the Democrats (39-32-29).

So was Kos that far off?

by esconded 2008-11-21 11:25AM | 0 recs
Re: The most accurate pollsters

What about the MOE?  Any consideration given to that?  Deciding who got closest or was best based on the final day of polling seems like a crap shoot to me.  I don't mean to pour gasoline on this fire, I'm just asking.  A graph showing the various pollsters vs time would be interesting and a better way to visualize some of the points you were making.

by tegrat 2008-11-21 11:30AM | 0 recs
Re: The most accurate pollsters

Give it up - Kos keeps pegging it - this site keeps making vaccuous post  - like Todd Beeton one on Hillary earlier to day.

This site, by having a chip on its shoulder and going on and on about Hillary is just losing credibility.

The rest of the progressive bloggers have moved on.

As long as Hillary is a team player and can understand the nuance of the real world (like Iran need not be an enemy, Cuba animosity is just stupid and talking to people to find common purpose) then she can do fine as SOS. When the Clintons are not plating power politics they are actually quite capable of introspection and considering complex issues.

I would ask that this site once again becomes a source of insite and not bitchy gossip emanating from embarassed ego's

by dbeall 2008-11-21 11:44AM | 0 recs
they're still counting

Per Green Papers, the spread is currently 6.86% and rising: O- 52.75%, M- 45.89%, or 8,823,446 votes.  Any pollster that ended up at 7 nailed it.

The total votes counted per Green Papers is 128,610,721, almost 1.5 million higher than the dkos scoreboard total.

by mboehm 2008-11-21 11:47AM | 0 recs
Re: The most accurate pollsters

Wow, it seems the only thing that can tempt Jerome to post here is the opportunity to gloat.

by Cincinnatus 2008-11-21 11:48AM | 0 recs
Give it up

Dude, retire as a prognosticator. Obama is going to get hammered, Barbie is going to be the biggest hit since Abe Lincoln, now this.

by MikeB 2008-11-21 12:04PM | 0 recs
Re: Give it up

Heh.  He may still be able to post to the front page, but he's irrelevant now as he yells at people to get off his lawn.

by username 2008-11-21 06:35PM | 0 recs
Re: The most accurate pollsters


by martellpdx 2008-11-21 12:26PM | 0 recs
Re: The most accurate pollsters
Armstrong's other predictions:
-Obama can never win! (He's black you know)
-Sarah Palin is the next Reagan
by martellpdx 2008-11-21 12:28PM | 0 recs
Re: The most accurate pollsters

Jerome you said:

his is the award that goes to the pollster that, given the final result, is consistently off until the final days before the election, when they poll to make it more in line with the other poll projections. So congratulations, Markos, on your R2K success.

How did R2K "poll to make it more in line with the other poll projections" without cooking the books? That is a serious accusation, especially since they didn't change the internal breakdowns. They would have had to flatly lie to do what you accuse. As Kos said, let Nate Silver decide if there appeared to be any funny business in the R2K tracking poll.

by Jawis 2008-11-21 01:35PM | 0 recs
Re: The most accurate pollsters

Am I the only one that finds it amusing that people feel the need to come on here to tell Jerome how much they dislike him. Way to drive up his hit guys.

If you're that eager to yell at people you dislike why not go over to red state and tell them how much you hate Republicans.

by world dictator 2008-11-21 01:46PM | 0 recs
Re: The most accurate pollsters

You're the member there.  Are you trying to drive hits on the RedState page?

by hootie4170 2008-11-21 01:50PM | 0 recs
Re: The most accurate pollsters

Jerome - This is sad and an indication of how this site has fallen. By the way, do you still think there weren't any voters in WV who were racist and that Texas Darlin and Alegre were just the greatest?  

By the way, what's the name of that new blog where Canadian Gal and others went? It was Moose something, right?

by politicsmatters 2008-11-21 02:32PM | 0 recs
Re: The most accurate pollsters

Oh, I found it: diaryId=669

by politicsmatters 2008-11-21 02:33PM | 0 recs
Re: The most accurate pollsters


by politicsmatters 2008-11-21 02:35PM | 0 recs
Re: The most accurate pollsters

What it really sounds like is that you can't admit you were wrong about this race throughout the campaign.  Just admit you were wrong and move on.  That would be much better for you.  The cover up is worse than the crime you know.

by Marylander 2008-11-21 02:37PM | 0 recs
Re: The most accurate pollsters

How 'bout a "Bill Kristol" award for the blogger/ pundit who was most consistently wrong about everything during the campaign?

Not sure why you feel a need to pick a fight with Kos when you were wrong about so many things during the campaign. I don't really have a dog in this fight but didn't Markos pretty much say "If Obama runs, he wins" before the guy had even declared his candidacy? Even if you're more concerned with polls than punditry, didn't you keep telling us the Battleground poll was the one we should trust most re: turnout model etc?

by animated 2008-11-21 04:34PM | 0 recs
Re: The most accurate pollsters

Doesn't Mr. Armstrong deserve an award for the following statement?: "Obama is definitely the weakest nominee that Democrats have put up this decade."

Mr. Armstrong, you are so unprofessional and so wrong on many levels. Please find a new profession.

by tomanderson13 2008-11-21 05:52PM | 0 recs
Sad. Pathetic. Intellectually dishonest. Liar.

The poll was completely transparent--there was no changing of the voter mix in the last week. The poll tightened because the tracking numbers tightened--nothing in the methodology was changed in the final week to affect the result.

Jerome, you have been consistently wrong from the beginning of this campaign cycle until the end--you have now completely finished the transition from "crashing the gate" to "suckling at the consultant class tit".

Guess what, the gate was crashed and your only role was pathetically trying to prop it up. And you are now irrelevant.

by Brannon 2008-11-21 08:53PM | 0 recs
Re: The most accurate pollsters

Hey, why don't you and Kos get out the rulers and be done with it?   Seriously!

by cooptimo 2008-11-22 07:02AM | 0 recs
26 comments, 1 in support of Jerome.

Anyone else think it'd be hilarious if everyone who commented lost rec/rate for "piling on"?

by nathanp 2008-11-22 07:37AM | 0 recs


Advertise Blogads