Articles and Analysis


DailyKos Fires Research 2000

Topics: Arkansas , Daily Kos , Del Ali , Pollsters , Research2000

Markos ("Kos") Moulitsas has opted to "part ways" with DailyKos polling partner Research 2000 and seek "a new polling partner to finish out this election cycle." And if that were not enough, he explains that "The decision was made, in part, on the results of Nate Silver's new pollster rankings."

Research 2000's past results aren't actually as bad as people will try and make them out to be. In fact, if there's one thing that's striking about the chart, is how closely clumped together those pollsters actually are. The difference in accuracy between the best and worst pollsters (omitting Zogby's genuinely crappy internet poll) isn't very big. As Nate told me via email as he walked me through the results:

The absolute difference in the pollster ratings is not very great. Most of the time, there is no difference at all.

And while many will focus on R2K's misses in Alabama and Arkansas, fact is they nailed several others, like the Hawaii special, the Nevada primaries, NY-23, etc. Every pollster has hits and misses, and R2K was no different. But in an industry measured in percentages, fact is they underperformed their peers.

I believe in accountability, in accuracy, and in making sure we provide the absolute best information not just to this wonderful community, but also to the outside world. As such, Daily Kos will be on polling hiatus the next several months as we evaluate our options and decide how to best proceed.

I reached Del Ali, the president of Research 2000, by telephone a few minutes ago. He said he considered the Kos blog post "very kind to us" although he takes issue with the characterization of their final Arkansas poll as a "miss:"

We're the only polling firm that did this race, so there's no comparison with other pollsters for the record. But, having said it, our final numbers were 49-45 in favor of Halter [with a] four percent margin of error. I do believe the result fell within the margin for error.

Ali also pointed to two factors that "one has too look at" in assessing his final Arkansas poll, which was fielded last week from Wednesday to Friday. First, he gave credit to the "Clinton machine" for being "very effective with the Democratic party establishment over the weekend." Second he mentioned the "controversy that's still lurking even though Halter conceded --- the forty some odd polling areas or polling precincts that people could not vote in the areas of the state that Halter could have done well in." Put that together, Ali says, and "it comes out as, you know, 'Dewey Defeats Truman.'

This news is a likely a blow to the Olney, Maryland based research firm. Ali told me in March that DailyKos and other progressive sites or organizations like the Progressive Change Campaign Committee have sponsored roughly 80 percent of the publicly released polls conducted in recent years by Research 2000. The vast majority of the Research 2000 polls we have logged at pollster have been sponsored by DailyKos.

I will have more to say on this story soon

Update - Del Ali emails with this clarification:

Kos polling and PAC's is about 80% of our political polling only. However, our political polling is less than 15% of our overall business. Overall, 85% of what we do involves consulting for businesses and non political organizations, polling for non state wide candidates around the nation and a great deal moderating focus groups around the nation. That is where we make our money.

If we lived on Kos polling and media polls we would not exist. So, where I will miss polling for Markos, the financial impact is minimal if not non existent.



Thank G-d they got rid of R2K there blog is good but their polling is awful. Markos, I suggest three letter to you IVR.



Daily KOS is an abolute joke anyway. They cannot be taken seriously.



Any way you slice it, Daily KOS cannot be taken seriously.



R2K's numbers weren't bad enough for KOS? Why does KOS even use a pollster, anyway? What use does a hate site have for opinions? They already know everything.

Do they think anyone thinks of them as a legitimate anything?

Maybe they can let the Huffington Post conduct polls for them. Or maybe Harris - they pay people to pretend that they are republican.


Glad to know you'll be tracking this story, Mark. Certainly it's the juiciest public-release polling firm news since the SurveyUSA dust-up.

One question I'd love to see explored is how well the true difficulty of a specific race is reflected in Nate's PIE measurement. Unless I'm missing something, he is accounting only for the difficulty of the type of race (President-national, President-state, Gubernatorial, Senate, House, etc.) being polled.

But the problem I see is that not all House or Senate races are created equal. As you know, a special election oddity like NY-23 is a real challenge to accurately forecast relative to a "normal" House race. But even less extreme cases may be problematic, like the old saw about the difficulty of polling in the South or accounting for potential "Bradley" effects from the candidates' ethnicities or even pure measurement difficulties like how well a Congressional district aligns with the underlying FIPS codes and telephone exchanges. I don't think it matters whether one accepts any of these as true factors in polling difficulty as long as we all agree that some races are tougher to poll than others in meaningful, irreducible ways.

Especially for Research 2000, this could be a significant effect--and, as we can now see, one with material consequences. That's because DailyKos goes out of its way--and laudably so--to poll races that are being ignored by other public-release pollsters. While we as the public benefit from the increased information, it seems like it also might be skewing the difficulty of the races that Research2000 has to poll and release for Nate's model's scrutiny.

To be clear, I'm not saying Nate's model is inaccurate in general or about Research 2000 in particular. I also think Nate's ratings are the best we've got and I appreciate his methodological transparency. But I'm worried that his conclusions are treated as received wisdom rather than simply the excellent analysis that they are.



If Halter had won last night we wouldn't be having this discussion right now haha. The war shall continue between MarKos and Blanche.



Probably a good idea. The Kos/Research 2000 polls have been terrible in all four big state-wide races of 2009 and 2010.

FINAL VIRGINIA KOS POLL: McDonnell +10% (actual- McDonnell +17.4%)
FINAL NEW JERSEY KOS POLL: Christie +1% (actual- Christie +4.1%)
FINAL MASSACHUSETTS KOS POLL: Brown and Coakley tied (actual- Brown +4.8%)
FINAL ARKANSAS RUN-OFF KOS POLL: Halter +4% (actual- Lincoln +4.0%)

Now the question is- Will Kos want to hitch themselves to a pollster that does not think it is still October 2008 or will they insist that the first midterms of the Obama Presidency will have the same turnout models of 2006 and 2008.



Fortunately people like Moulitsas give me some hope. The progressive forum on CSPAN THe other night, cheered me up a lot. I like how cynical Moulitsas is, and I think he is very prophetic. Like me, he knows that conservatives fail to ever see the big picture, and they fail to really be logical. THey are very clever though. I would give conservatives a lot of credit for twisting the facts, like the "death Panels" and their politicians either go along with it, or they refuse to go along with it, but when asked to condemn hateful remarks from people like Limbaugh or Beck they are scared.




I was pretty disappointed when Nate released his pollster ratings that he didn't (that I could find) provide an error estimate for his PIE measure.

This is crucially important information, particularly if one is going to use these values to make major decisions like firing polling firms.

Perhaps if you asked him, he might be more likely to respond? There's no way he pay any attention to an email from me...



Zirinsky - he also estimates the difficulty for the year the poll was taken, and the precise race if the polling is "robust" (if there are enough polls from some of the most prolific pollsters). So he drills down to "Senate in 2008" and such. There's more in Silver's methodology post - you should read many of the "dummy variables" like they are absorbing the PIE from the difficulty inherent in that category.



Well, I think Lincoln's chances of winning are slim, but I think Arkanas won't go the way of Alabama and Texas and become an automatic Republican Stronghold. I would sort of like to say, Arkansas to the south is like what NH is to the northeast. It can trend either way, depending on the climate, but can be unpredictable.

Arkansas has moved more Republican and NH moved a bit more into the D column, yet, Dems in Arkansas are very loyal, just like Republicans in NH are very loyal.

NH is still the only New England state that hasn't been sold on liberalism, and AR, even though Huckabee was popular, and Obama lost by 20 points, is different than the other southern states.



I am not in any way going to fault the progressive groups trying to nominate Halter. I think the pundits are calling it a waste, but I disagree. I think he got his ideas across, and made an impact on people in his state and nearby places. A lot of ideas don't get implemented right away, but it motivated the 1/3 of Arkansas that is somewhat likeminded as Halter. Lincoln also reached out to the people of her state.

If this doesn't end up good for Lincoln, the effort was not in vane. I never ever say that spending money on strong progressive candidates is a waste of money.



@ Farleft-

You are correct that Arkansas has traditionally been more Democratic and slightly more liberal than most Southern states, but that distinction is getting less and less, especially when it comes to the way Arkansas votes in Presidential elections:

1996: Clinton (D)- 53.7%
2000: Bush (R)- 51.3%
2004: Bush (R)- 54.3%
2008: McCain (R)- 58.7%

Like most of the states around it (Oklahoma, Missouri, Louisiana, Tennessee), Arkansas is trending more Republican and will continue to do so for some time.

As for Blanche, she will get a bump in the polls for the next couple of weeks against Boozman, but when November rolls around, she will be lucky to get 45% or more of the vote. She remains very unpopular in Arkansas as a whole.



@ Farlefty-

Halter is not as "progressive" as you assume. Most of his positions are very middle of the road. He was simply hitched onto by unions, the netroots and all the other leftist groups because they so badly wanted to take out Blanche. If Halter was some hardcore left-wing "progressive", he never would have come close to being elected Lieutenant Governor of the state of Arkansas.



I agree Jms tiger that the progressive groups picked the wrong guy to be their hero. I was not so sure that Move On should be involving themselves in a state like AR. I actually suggested that they find a better Democratic candidate in IL. They didn't listen. I think they have a much better shot at funding the progressive candidate in Colorado. It is a bigger Union state than Arkanas.

I think things worked out better for Conway in KY. He didn't identify him as a progressive or get labeled by the media that way. I think he'll campaign as a centrist.

I try to be pragmatic and "know when to hold them and know when to fold em", and like the Democrats to find candidates who can win in their districts and states. If they can find a Democrat who could win in Montana,Texas, Kentucky or any other conservative state, I don't mind if they vote with the Democrats only 70 percent of the time. If that is what it takes to keep the balance of power and get re-elected I'm satisfied.



Perhaps it's time to upgrade to Research2010. ;)


Post a comment

Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.