New DailyKos/Research 2000 tracking poll

[The Vorlon]

Dems have a 10% or 11% party ID advantage nationally right now, so this is right in line.

Actually, in the 15,000 interviews conducted by Rasmussen in August 2008, the partisan ID gap had shrunk to 5.7%

The 10% or so gaps showing in the summer were in some ways artificial and a consequence of the historic and dramatic Obama/Clinton primary battle.

Interestingly, Rasmussen has noted in his data so far in September that the partisan gap has shrunk substantially again in September (possibly as the result of the Palin pick)

I do not believe that your statement about a 10% democratic advantage in partisan ID is supported by current polling.

Summary of Party Affiliation
September 1, 2008         
 Republican Democrat Other R - D Quarterly 
2008   
Aug 33.2% 38.9% 28.0% -5.7%   
Jul 31.6% 39.2% 29.2% -7.6%     
Jun 31.5% 41.0% 27.5% -9.5% -9.9%   
May 31.6% 41.7% 26.6% -10.1%     
Apr 31.4% 41.4% 27.2% -10.0%   
Mar 32.1% 41.1% 26.8% -9.1% -8.1% 
Feb 31.8% 41.5% 26.7%

[The Vorlon]

Rather than argue about a pollster's bias we could compare what they predicted in 2004 versus what actually happened...?

Methodologically, Del Ali at Y2K models/weights/stratifies his data really heavily - more so than most pollsters.  This is not bad per se, it has both ups and down.  Usually when he is close, he is really close, when he misses, he misses by a lot..

BTW - If I missed a Y2K poll from 2004, let me know, I will toss it in this chart....



If you look over their 8 polls in 2004, other than Florida they were actually pretty darn good.

Taken as a whole, on average they pulled about 2% to the Democratic side, which is historically what this firm does - they have a slight democratic bias.

R2K is not a joke firm, they know what they are doing.

Del Ali their principal pollster has a long personal association with the Democrats which optically looks bad teaming up with DailyKos.

I think if DailyKos really wanted credibility they should have hired a firm with deep blue GOP roots - such as Public Opinion Strategies, Ayers HcHenry, or maybe Clearwater Associates.

I would add about 2% to the GOP side with this firm, but that is just an opinion.  Kinda like Strategic Vision, except going the other way...

My 2 cents worth

[The Vorlon]

At least they are willing to release ALL internals of their tracking poll, unlike the major pollsters.

I actually do congratulate DailyKos for what they are doing.

My point however is that given the nature of their site, regardless of whom they select as their pollster there will be charges of bias. 

Picking a firm with a slight democratic bias, and a principle pollster with long ties to the Democratic party does not help this perception.

I still think hiring a firm with GOP roots, or perhaps a joint effort between two firms, one GOP, one Dem, may have been a better solution in terms of public perception.

[The Vorlon]

The mainstream media treats Zogby and sometimes Zogby Interactive polls the same as regular polls, so I don't know if it really matters.

Just yesterday I saw Maddow stick in a Zogby poll with Barr getting 11% in New Hampshire along with other polls.

The tragic things is that bad polls and bad journalism are made for each other. 

If you are a trash news outfit more interested in ratings and revenue that actual information you want polls that bounce around like crazy and show a different candidate in the lead every day. (ie Zogby)

Can you imagine basing a tv segment on  Rasmussen in 2004.....?

"And for the 47th consecutive day Bush is ahead just barely at the edge of the margin of error...
This marks the 87th time in the last 92 days he has been ahead by 2 or 3 points....the other days he had leads of 1 and 4 points respectively..."

Pretty boring...

Better to have California tied, Idaho in play, and a nail biter in Vermont...

[The Vorlon]

You mean ALL television journalism when you say "bad journalism," right?

I notice a lot of citing different polls.  Zogby has McCain leading despite Barr getting 11% in New Hampshire but Survey USA has Obama behind 20% in North Carolina.  Analysis follows.

Or, what is more common, citing some sketchy poll taken a couple days before to indicate a huge drop/gain after some legitimate poll shows the race close.

Actually, the fact that Survey USA now and then puts out a real stinker of a poll shows that they are a real and honest firm.  The make the calls, add them up, and put out the results, they don't fudge anything. (unlike a few firms)

SurveyUSA will likely do a few thousand polls if you add up all the house, senate, governor, statehouse, etc races between now and November.  You do that many, and the odd one blows up.

Other than the deep south where they tend to underpoll the GOP a bit (not that it matters, +20 and +30 are for all practical purtposes the same result) their record is pretty decent.

[The Vorlon]

I personally think DailyKos + Zogby would have been a perfect match