Judis & Bowers on Research 2000 polling

John Judis asks:
...when I look at the polls' internals for whom they are surveying, I find that 13 percent of the respondents are African Americans and 13 percent are Hispanics, which is over a quarter of the sample.  In the 2004 presidential exit polls, for instance, the final tally for the national vote was 11 percent African American and 8 percent Hispanic. Maybe, black turnout will be up nationally by a percentage point or two this year, but I would be surprised if the total for both blacks and Hispanics  is more than 21 percent. And since that vote tilts strongly pro-Obama, having these groups as 26 percent of your sample would seem to tilt the entire poll toward Obama.  Have I misread something?

Chris Bowers asks:
What worries me about the Daily Kos tracking poll is that 9% of it's sample is identified neither as partisans nor as independents. The current partisan breakdown of 35D, 30I, 26R only adds p to 91%. Exit polls force everyone to choose one of the three options, so with this fourth "none" option it is difficult to compare the Daily Kos tracking poll to exit poll results from recent election. Overall, the number of non-partisans, which is basically how those 9% identify themselves, feels sky high to me. In every Presidential election since 1980, partisans (Democrats plus Republicans) have been at least 73% of the electorate, but this poll posits only 61%.

Having watched how Research 2000 has polled in the past, the answer is probably that the demographics of those polled generally mirror the population; with quotas assigned to reflect the voter registration.


It's a pretty sketchy likely-voter methodology. One that would inherently favor Democrats, as Judis notes, and one that doesn't reflect the reality of the partisan make-up of those that vote, as Bowers notes, and more.


To believe the Research2000 prediction for 2008, you have to believe that latinos & blacks will vote in equal numbers to their populations as whites do, that Democrats will outnumber Republicans by 9% and Independents/Refused will be nearly 40% of the voting day population, and to assume that older people (60+) are going to be less of the voting population (even though the trend says they will be more).


It seems absurd to take it at face value. But this isn't something that Markos thought up, he's just paying for it. It's entirely the methodology of Research 2000. I'm not sure exactly how I'd re-weight the poll either; given how off it is across the board, it's probably not salvageable. Well, the MOE is 3% so I'll guess that it favors Obama by that much.

Tags: 2008 (all tags)

Comments

91 Comments

Re: Research 2000 polling

Has anyone brought this up with Markos?

As a rule, I'd take 2 points off the margin with the DailyKos poll and add 2 to the margin in Rasmussen.  That puts the margin around where most others have the race.

by Marylander 2008-09-22 11:45AM | 0 recs
Re: Research 2000 polling

Yea, we've emailed back and forth. He didn't come up with the methodology. I was under the initial assumption that it'd mirror the '04 exit polls. Research 2000 apparently sticks by their findings as what the '08 likely voter is going to be.

The problem with our thinking that Rasmussen is off is their historical record. They, along with TIPP, were the closest in 2004.

by Jerome Armstrong 2008-09-22 11:50AM | 0 recs
But does 2004 equal 2008?

I do agree that the Research 2000 poll presents a best-case scenario. That much has been consistently evident from their consistently rosy picture of Obama's performance. I prefer to follow Gallup and Rasmussen, although I doubt the race is as close as Rasmussen believes it is.

However, I disagree with the assumption that this election will be anything other than unique. This is no ordinary election, and even marginal improvements in youth and AA voter turnout can make a huge difference.

by iohs2008 2008-09-22 12:11PM | 0 recs
Re: But does 2004 equal 2008?

I agree with your turnout of youth and black voters, but I also believe that turnout among the older, 60+ voters, and evangelicals, will be at a historical high. Those groups are on the ballot just as much as the voters you point out, this election.

by Jerome Armstrong 2008-09-22 12:14PM | 0 recs
Re: But does 2004 equal 2008?

I don't see the turnout being higher than 2000 and 2004 with George W.

Simply put, my opinion is we have much more room to improve than they do. AA turnout and youth turnout was abysmal in 2004.

by iohs2008 2008-09-22 12:21PM | 0 recs
Re: But does 2004 equal 2008?

The only way evangelicals get excited is if Palin maintains her sway over them. They don't care much for ol' Johnny. I guess it'll come down to whether they believe Sarah or their own lying eyes.

As for the elderly, I think Obama has a decent shot between now and Election Day to pull a few more over. It'll probably come down to the debates.

by Spiffarino 2008-09-22 12:53PM | 0 recs
Re: Research 2000 polling

Being closest in 2004 is no mean feat, but I'm unwilling to give undue weight for historical accuracy. Remove the names of the polling outfis, and the average of the 5 tracking polls is a 4% Obama lead. Names removed, we'd be saying the Ras MOE favors McCain by the same margin that the R2K favors Obama.

More to the point, why is this relevant right now? Why is now, with everything else going on, and 6 weeks to go, the time to debate or pontificate on R2K's polling specifically?

by Beomoose 2008-09-22 12:15PM | 0 recs
Re: Research 2000 polling

Because people here seem to hate Daily Kos.

by Lolis 2008-09-22 12:55PM | 0 recs
Re: Research 2000 polling

And then people wonder why the Trad Media loves to run "Dems Divided" stories.

by Beomoose 2008-09-22 01:28PM | 0 recs
Re: Research 2000 polling

"People here"?  "Seem to"?!

Be honest.  Jerome disses Daily Kos for the same reason he let a bunch of PUMAs hang out here and burn his credibility:  he has some kind of irrational hatred of Obama that absolutely must be expressed again and again, and Kos doesn't share this hatred.  That's all there is to it.

by username 2008-09-22 01:40PM | 0 recs
Re: Research 2000 polling

It seems to be a lot easier to get banned here by being an Obama supporter than being a PUMA/McCain-defender/Palin-defender.  There are still a few commenters running around here who traffic in racially-charged smears of the Dem candidate.

by the mystical vortexes of sedona 2008-09-22 01:44PM | 0 recs
Re: Research 2000 polling

I find it interesting that you find time to vociferously disagree with the methodology of this poll while accepting at face falue the 10% voter ID shift in Gallup and other national tracking polls that showed up after the Republican National Convention.

This being said, likely turnout models are very shaky in every one of these polls since there is no way to tell how accurate their predictions are.  You must admit that just because somone did not vote in the last one or two elections, this is not a reason to discount their response in the poll, yet likely voter turn out models do this on a regular basis.  There is good reason to believe that this underrepresented group will turn out this year in record numbers (first and second time voters).  All of these methodologies favor McCain.

I would call it a wash at best, and certianly by your methodology of subtracting 3 points from the Dkos/R 2000 poll from Obama's tally, I'd subtract 3 point from Gallup and SUSA daily tracking from McCain's total.  Either way, these are rough estimates, with really high margins of error and large numbers of undecideds (typical for pre-debate polls) and they will shift significantly before the election.  

So perhaps not getting too worked up about the polling and working to get the Democrats elected is the best course of action, rather than complaing about methodology on a poll that in the end means little.

by Why Not 2008-09-22 12:17PM | 0 recs
Re: Research 2000 polling

from Jonathan Singer's post below:

Here are today's numbers:

Obama McCain
Diageo/Hotline 47 42
Gallup 48 44
Rasmussen Reports 48 47
Research 2000/dKos 49 43

Average: 48.00 44.00

it seems that even removing DKos/R200 polls the average is still less than 2% better for Obama.  The polling as I said below is not any better or worse overall than the rest of them (ie they are all crap).

Forget about polling and focus on getting Democrats elected.

by Why Not 2008-09-22 12:22PM | 0 recs
Re: Research 2000 polling

Also a CNN poll that has the numbers 51/47 O.  Over 50%?

by thezzyzx 2008-09-22 12:28PM | 0 recs
And Captain Downer makes his daily post

My day is complete.

Nah seriously I do think Research 2000 is a tad favorable to Obama (and Democrats), but its results aren't that far removed from the rest of the polls.  Certainly they're not as bad as Strategic Vision, which is about 5% more GOP than other polls.  If they did national polls they'd probably still find McCain up.

by Skaje 2008-09-22 11:49AM | 0 recs
Re: And Captain Downer makes his daily post

Go take your meds.

by Jerome Armstrong 2008-09-22 11:51AM | 0 recs
Wow Jerome...

claiming someone has mental illness based on an internet post.  Aren't you become a budding neo-con like Dr. Sen Frist.  And with the stigmatizing of mental illness saying that skaje's 'wrong' opinion is based on a perceived mental illness certainly is worthy of someone honored with an award named for Sen. Wellstone who was a true leader on mental health issues (did they know you have a severe dislike of mentally ill people?)

So did you really get shot down at a job with the Obama campaign at the beginning of the primary cycle.  Based on your predictive ability as well as your ability to draw in democratic partisans I would say that if you did then not hiring you was one of the best decisions the Obama campaign made, and that is only basing it on your political abilities and disregarding anything else that may or may not be relevant.

by Sanguine Giant 2008-09-22 12:48PM | 0 recs
Re: Wow Jerome...

Why all the hate for jerome. He's a senior statesmen in the blogging world and this is his sight. Sure he's not an obamabot as I would consider myself but he's a dem and he brings perspective to the situation. And as far as polls go I'd rather go with rasmussen and work my ass off to ensure success than to go with research 2000 and get complacent.

by kbal 2008-09-22 01:00PM | 0 recs
Re: Wow Jerome...

He's a senior blogger and has responsibilities too. One of them is taking criticism. Put yourself up there, then you have to be able to take flak.

I'd say two other responsibilities, on a democratic blog, is to have the guts to admit you were wrong, and not go around finding fault and looking for a cloud in every silver lining.

Jerome barely has a good word to say about Obama and his candidacy. He has a right to his opinion, and unless you want your politics run like the Alaska governship, we also have a right to call him on it when he gets it wrong.

Otherwise what is point of 'progressive community' like this

by duende 2008-09-22 02:56PM | 0 recs
Re: Wow Jerome...

In Jerome's defense, it isn't easy being the fun-killer when Todd and Jonathan are constantly pointing out McCain's mistakes, Obama's new ads, good polls, voter registration advantages, etc.

One could say (oh how this pains me) that Jerome...is actually doing...good by keeping...us from getting...overconfident...

by Skaje 2008-09-22 01:06PM | 0 recs
Re: Wow Jerome...

that's one way of looking at his posts.

one only need to look at Sirota or Rosenberg to see how one can be critical of Obama within the framework of being behind his candidacy.  

In any regard, I admire your positive outlook on why Jerome has been posting a certain way.

I almost think it will pain Jerome to be proven wrong by an Obama presidency and having Markos outguess him every step of the way.

and I'm still chuckling over Jerome's prediction of a big Hillary victory in Hawaii.

by alex100 2008-09-22 01:36PM | 0 recs
Re: Wow Jerome...

Check out the page of Sanguine Giant. Looks like he's been banned for criticizing a remark above.

What next? Hey, Let's sack a librarian and burn some books

by duende 2008-09-22 03:06PM | 0 recs
Re: Wow Jerome...

even more comical is watching Jerome uprate himself and take away rating privilege to those who took offense to him telling someone to go take there meds.

by alex100 2008-09-22 09:06PM | 0 recs
Re: And Captain Downer makes his daily post

Where was that from? How can you expect anything but 'sewerage' from people when you lead by example with comments like this?

Skaje has never been anything but a genuine and generous commentator on this site. And he/she's helped to keep the place clean and honest.

by duende 2008-09-22 02:49PM | 0 recs
Re: And Captain Downer makes his daily post

"Downer"

Meds,

Get it?  Some people are so lacking in humor. Anyway, it outs the trolls that can get banned.

by Jerome Armstrong 2008-09-22 05:36PM | 0 recs
Re: And Captain Downer makes his daily post

As someone who had their rec and rating privileges once removed, I'm not hugely reassured by the troll bait

by duende 2008-09-22 06:14PM | 0 recs
Research 2000 polling

Even so, R2K, Hotline and Gallup are almost converging. Perhaps they are making the same mistakes or there is a changing dynamic?

by RandyMI 2008-09-22 11:51AM | 0 recs
Re: Research 2000 polling

Well, I take Rasmussen and Battleground as more competent with their methodology than either of those three, and they have different results.

by Jerome Armstrong 2008-09-22 11:54AM | 0 recs
Re: Research 2000 polling

Of course.

by RandyMI 2008-09-22 11:58AM | 0 recs
Re: Research 2000 polling

Yes, a "tracking poll" with results over a week old (as is the case with Battleground) is solid methodologically speaking.

by Jonathan Singer 2008-09-22 12:13PM | 0 recs
Re: Research 2000 polling

The Tarrance Group and Lake Research Partners know nothing next to you.  Anyway, they predicted Bush in 2004, but were off by 2.5%, and I'm sure they'll tighten up the days come the election.

by Jerome Armstrong 2008-09-22 12:21PM | 0 recs
Re: Research 2000 polling

No. I'm just saying it's not a daily tracking poll. It's a rolling poll, but it's not a daily tracking poll as it (a) includes numbers that are over a week old and (b) isn't daily.

by Jonathan Singer 2008-09-22 12:25PM | 0 recs
Re: Research 2000 polling

Do you think having a quarter of the sample come from over a week ago is reflective of the state of the race today?

by Jonathan Singer 2008-09-22 12:26PM | 0 recs
nate's got rasmussen at a pronounced

Republican tilt.

Competency with methodology does not equate to accuracy nor precision.

by Blog Surrogate 57 2008-09-22 12:29PM | 0 recs
Assuming turnout will have the same

demographic proportions as in 2004 creates its own problems.

I feel that it's likely we'll see higher African-American turnout across the board, but especially in the states where Obama has the most extensive field operations. Several of those states were not targeted at all by Kerry, so we may notice a big difference.

I have no idea what to expect from Latino turnout.

by desmoinesdem 2008-09-22 11:56AM | 0 recs
Re: Assuming turnout will have the same

Agreed, that's the trend, for whites to make up less of the voting population. 2004 will not be repeated on the makeup of whites, blacks, latinos, asians, others. That seems a given, but it won't be with the makeup they are predicting, as Bowers notes, regarding the Asian vote.

by Jerome Armstrong 2008-09-22 12:05PM | 0 recs
Would it make sense...

...to compare each pollster's weighted and unweighted numbers?  The various pollsters' unweighted numbers should be very similar, and I would wonder what was going on if they weren't.  If the weighted numbers diverged while the unweighted numbers didn't, this would probably be due to differences in weighting, likely voter methodology, etc.

by KTinOhio 2008-09-22 12:01PM | 0 recs
not generally released

at least kos gives the whole subgroupings. we can't tell if everyone is screwing things up wholesale, without the numbers.

by Blog Surrogate 57 2008-09-22 12:31PM | 0 recs
Is it then your contention that the demographics

of voters will look like 2004?  

I don't think R2K is off with regard to Black/Hispanic vote this year.  I live in an area with a large Hispanic population that realizes their vote is extremely important to this election.  We have a lot of Hispanic volunteers canvassing for the campaign.  We have a miniscule Black population but about 90% of them are volunteers.  There is huge excitement in both communities.  

by GFORD 2008-09-22 12:02PM | 0 recs
Re: Judis & Bowers on

Here's an idea: Just toss out Rasmussen and R2K from your sample, (they're the two outliers), and average the remaining two trackers then.

Listen, we know that you don't like the kos poll.  We heard that many times from you, and I imagine that this post will not be the end of it.  

Just use the above rationale to exclude the two polls from your tracker average, and stick with it.  That way you don't have to spend so much time trying to prove to us that the Kos/R2K poll is screwed up.  Just leave it out...and balance that by leaving Rasmussen out as well.

Voila!  Problem solved.

by OmniStipes 2008-09-22 12:04PM | 0 recs
Re: Judis & Bowers on

You forgot about Battleground tracking.

My guess is to weight Battleground and Rasmussen as is, and wait until Gallup & Hotline go to an LV model to pay attention add them.

by Jerome Armstrong 2008-09-22 12:09PM | 0 recs
If

Obama improves turnout in AA, youth, and Latino communities, even by a little bit, and we've seen he is capable of doing this (with 2 of the 3) in the primary, then the LV model from 2004 is very pessimistic. I agree with accuracy, but only to a point: there is zero benefit to using an overly pessimistic model to show a McCain lead, as that will do nothing but drive down enthusiasm and turnout. Optimistic polls can serve a purpose.

by iohs2008 2008-09-22 12:14PM | 0 recs
Re: Judis & Bowers on

Well, if that's your methodology, then you're basically skewing towards the worst case scenario for Dem voter turnout.  

Under your approach, if we're ahead, then it's dead nuts solid that we're ahead.  That approach could be psychologically useful, but may not yield the most accurate perfromance, with respect to the model matching the results.

I consider your method to be good at tracking the performance of the lower end of the band.  It's still a useful approach, but does not accurately reflect to true probabilities that exist within one standard deviation from the mean.

by OmniStipes 2008-09-22 12:17PM | 0 recs
Re: Judis & Bowers on

NO, all the pollsters move to LV by the election, some just sooner than others.

by Jerome Armstrong 2008-09-22 01:12PM | 0 recs
Re: Judis & Bowers on

Your comment does not affirm or refute any of my points, in any way, shape or form, Jerome. I'm not sure why you even posted it.

Have you ever taken a class on statistics?  You appear to be missing some of rudimentary understanding that is needed to really analyze these numbers.

by OmniStipes 2008-09-22 05:40PM | 0 recs
Problem with bashing Rasmussen

is they've had some more Obama friendly numbers in the tracking poll than Gallup at some times this summer.  

by magster 2008-09-22 12:25PM | 0 recs
Re: Problem with bashing Rasmussen

I'm not bashing Rasmussen...I'm making a statistical argument for ignoring the outliers.  

If you consider the five trackers that Jerome wants to use, just eliminate the two outliers everyday, (high and low), and average the other three.  So long as it's the same five every time, you can eliminate the highest and the lowest each time, and take the other three into your average.

This is a valid tool for dampening statistical noise.

by OmniStipes 2008-09-22 12:33PM | 0 recs
Isn't that how Nastia Liuken...

...got screwed out of the gold medal? (snark)

I guess the serious way of saying it is that if there is a cluster of pollsters saying one thing, and one saying another thing out of the cluster, throwing out the high and the low might do more of a disservice to accuracy.

by magster 2008-09-22 12:47PM | 0 recs
Re: Isn't that how Nastia Liuken...

throwing out the high and the low might do more of a disservice to accuracy.

You may want to go back and study the basics of statistical analysis.  Your statement is in direct contradiction to the generally accepted methods of the mathematical analysis of stochastic processes.

But don't let that stop you from throwing out an opinion.  Most people don't.

by OmniStipes 2008-09-22 05:45PM | 0 recs
For once, I agree with Jerome totally......

The R2K daily track is a joke.

Optimists in the Hispanic political community talk about "more than 9 million" Hispanic voters this year nationally.  That would be no better than about 8 percent of the total IF the total is the same as in 2004 (121 million), and not more than that; if total voters exceed the 2004 number, then Hispanic vote share is less than that.

Regarding the partisan ID breakdown, the poll undercounts Dems slightly and Rethugs very heavily, the latter being the bigger problem.  Rethugs will be in the low/mid-30s, and while we WILL have a party ID advantage greater than 2004 or 2006 contrary to Jerome's belief based on his strange embrace of a single post-convention CBS poll, 9 points would be on the very optimistic end on what can be realistically expected.  I'd say Rasmussen probably is a good gauge of where party ID really stands as they were off by no more than a point or two the past couple cycles, and they peg the Dem advantage currently at 6-7 points.

Here's where the R2K daily track IS useful: the internals on groups with large sample sizes, the biggest being white voters.  I pay attention to the size of Obama's deficit there, mindful that Kerry lost the white vote by 17 points in a photo-finish election.  As long as Obama stays inside that 17-point margin, that's a useful metric to feel good.

by DCCyclone 2008-09-22 12:06PM | 0 recs
Re: For once, I agree with Jerome totally......

No, I've not said that '08 would be just like '04 for the partisan turnout, just that it wouldn't be a 8-11% difference. My take is that it'll fall in the upper average of the past ten years, about 3-4% in favor of Dems.

by Jerome Armstrong 2008-09-22 12:11PM | 0 recs
I doubt you've got NonPartisan there...

I think you're looking at either Southern Dems or Northern Republicans -- people who don't vote their party much.

OR you could be looking at people who are just plain disgusted with the Republicans (Obamacans, or leaners).  If you hated the Republican Party, would you identify with it to a pollster? Note, when you answer this, how many Hillary supporters said they wouldn't vote for Obama...

by Blog Surrogate 57 2008-09-22 12:35PM | 0 recs
Re: For once, I agree with Jerome totally......

As for party ID, I think there are strong reasons to expect a bigger margin of difference.  Two possible reasons - the potential for a record-breaking turnout among African Americans, and the fact that currently, the polls don't take into account the effects of the massive GOTV drive being undertaken by the Obama campaign.  

by ProfessorReo 2008-09-22 12:11PM | 0 recs
Re: For once, I agree with Jerome totally......

Also, I still have yet to see any pollster accurately account for cell-phone only households.  Auto-dialers can't legally call cell phones, so most pollsters don't do them.  They try to make up for it by overweighting the 18-29 subgroup, but I am not convinced that young cell-phone onlys vote the same as young landliners.  I think there is a good chance that young cell-phone onlys are slightly (very slightly, maybe) more likely to be liberal.  Every conservative young person I know (which, admittedly, is not very many) has a landline.

by ProgressiveDL 2008-09-22 12:38PM | 0 recs
see nate's site

most pollsters just starting cellphones.

cellphone only users are more black and more young than the populace as a whole. It's the more black that worries me, the more young looks like they've been weighting accurately for.

by Blog Surrogate 57 2008-09-22 12:41PM | 0 recs
Re: see nate's site

Nate?  Link?

by ProgressiveDL 2008-09-22 12:51PM | 0 recs
link to the baseball genius

http://www.fivethirtyeight.com/

kampai!

by Blog Surrogate 57 2008-09-22 12:53PM | 0 recs
Re: link to the baseball genius

Awesome, thanks.  I have been thinking that the cell-phone only inaccuracies would just about make up for the Bradley effect, and this seems to confirm it.  Thanks.

by ProgressiveDL 2008-09-22 01:31PM | 0 recs
Re: For once, I agree with Jerome totally......

It's also the only one that tracks favorability for all four candidates every single day. A quick glance at Joe Biden's numbers demonstrates the usefulness of that. Methinks in the quest to avoid damaging his image, he's severely damaging his image.

Besides taht, not only can we see the internals for each party, Obama is lacking in Democratic support! That's important to track as well!

by vcalzone 2008-09-22 12:24PM | 0 recs
Re: For once, I agree with Jerome totally......

Obama is lacking in Dermocratic support?  Do you really think an 86-11 lead is an underperformance?

by KTinOhio 2008-09-22 12:29PM | 0 recs
Re: For once, I agree with Jerome totally......

This year? Kinda. Look, even assuming you're right, it's still something that has to be watched carefully.

by vcalzone 2008-09-22 01:07PM | 0 recs
Re: Judis & Bo

The way I see it, it's good to have polls with different methodologies.  Research 2000 is the counter-weight to Rasmussen.  Both are useful, because Research 2000 gives us data if pro-democratic assumptions about the electorate hold up, and Rasmussen gives us data if they don't.  

Together, they give us both the upper and lower limits of the race.  And overall, it's good news for Obama, because, if things go as planned, he has a comfortable lead.  But, even if turn-out is lower than expected, it's essentially tied.  Meaning, McCain is fighting to just make a race come election time, while Obama is seeking to make it an easy victory.  

by ProfessorReo 2008-09-22 12:06PM | 0 recs
Re: Judis & Bo

I do agree with this. I find the notion rather unnerving, to say the least, that optimistic polls must be thrown out, but low and behold, we must embrace the pessimistic ones. As you say, they provide a bounds, and that is how I view them.

Optimistic polls serve a major purpose, in that people are overall more likely to vote for a winner than someone who is guarenteed to lose. If we generate a pessimistic attitude, it will be picked up by the MSM, and turn the tide against Obama.

by iohs2008 2008-09-22 12:18PM | 0 recs
Jerome assuming vote supression?

I wonder...

by Blog Surrogate 57 2008-09-22 12:36PM | 0 recs
Who Cares?

Leave it. There's 5 tracking polls now, at least one of which strongly disfavors Obama *coughbattlegroundcough*. You don't like the results of that one? Go use another one. Or better yet, just don't worry about it.

by vcalzone 2008-09-22 12:17PM | 0 recs
Looking at R2K's Track Record

You see the pollster pegging Republican support correctly, or within about a point, while underestimating Democratic support. It happened in last year's Kentucky Governor race (final poll had it 55/40 in favor of the D, ended up 59/41). It happened in the Nevada Governor race in 2006 (final poll had it 47/41 in favor of the GOP, ended up 48/44). It happened in the Iowa Governor race in 2006 (final poll 50/45 in favor of the D, ended up 54/44). It happened in the Wisconsin Governor race in 2006 (final poll was 50/44 in favor of D, ended up 53/45). Those are just the few that I found relatively quickly.

by Jonathan Singer 2008-09-22 12:20PM | 0 recs
Re: Looking at R2K's Track Record

So are you standing by their prediction of the national turnout?

by Jerome Armstrong 2008-09-22 12:22PM | 0 recs
Re: Looking at R2K's Track Record

I'm just pointing out facts. It may be that their presumptions are wrong this year. But when it comes to past performance, R2K has, at least in the last couple of years, hit Republican turnout correct, more or less, and underpredicted Democratic turnout.

by Jonathan Singer 2008-09-22 12:27PM | 0 recs
who had the enthusiasm in those races?

I'm guessing dems?

by Blog Surrogate 57 2008-09-22 12:37PM | 0 recs
poll

Black vote will eaily grow by 2% or even 3%. Hispanic vote has been growing 2% every four years and accelerating - also it trends D by far this year from 60% to 70%. These two factors alone create a Kerry plus 5% which is a 2% lead for O. Add another set for Ds outpolling Rs (probably by more than 4 this year and it is probably included in the Black and Hispanic sampling) and you have  a extremely difficult R roadmap.  VA is already trending O by 6 (SUSA)- and believe- that ain't changing. Soldiers are stupid and do not want to die. The military vote will be close to 50/50.  An interesting observation by someone- the R motivation is starting to fade for several reasons and this will probaly create an insurmountable problem for Rs.  As to which poll? Rasmuseen had an easy election last time= very steady- remember he was way off in 2000- so there is a house bias; however, he did pick up the +5 to +6 O gain last week- I think RCP average has a great track record and I would stick with that. Once Obama is out the MOE (a 3+ lead in the polls) he will be out of the danger zone.        

by RAULC 2008-09-22 12:23PM | 0 recs
New CNN Poll

OBAMA SURPASSED 50%

I'm literally trying to clear my head

Fifty-one percent of registered voters are backing Obama, who now holds a 5 point edge over McCain, at 46 percent. McCain and Obama were tied at 48 percent apiece in the previous CNN/Opinion Research Corporation survey. Obama's advantage, while growing, is still within the poll's sampling error of plus or minus 3 percentage points.

Where did Obama make his gains?

"In two core McCain constituencies: Men, who now narrowly favor Obama. And seniors, who have also flipped from McCain to Obama," says CNN Senior Political Analyst Bill Schneider.

When including only those people most likely to vote, the results are pretty much the same: Among likely voters, Obama has a 4 point lead, 51 percent to 47 percent.

by RandyMI 2008-09-22 12:23PM | 0 recs
35D, 30I, 26R

According to the internals of today's R2K poll, the sample included 389 Democrats, 280 Republicans, 331 Independents, and 100 Other/Refused, for a three-day total of 1,100 respondents.  If we remove the Other/Refused, we're left with 1,000 respondents, of whom 38.9% were Democrats, 28.0% were Republicans, and 33.1% were Independents.  How does this compare to what we know about the current electorate?  This, IMO, is the key question; is a 9-point Democratic edge consistent with other available data?

I reran the numbers with and without the Other/Refused group, and the figures were nearly identical; in fact, Obama's lead actually increases a bit when Other/Refused is taken out of the sample.

The current R2K averages shoe a 49-43 Obama lead.  Obama leads 86-11 among Democrats, 45-42 among Independents,and 46-42 among Other/Refused, and trails 88-6 among Republicans.  Note that the Independent and Other/Refused figures are nearly identical; would it make sense to combine the two groups for purposes of analysis?

by KTinOhio 2008-09-22 12:24PM | 0 recs
out here in California...

there is no formal category for "independent" - it is formally "decline to state".  Decline to State registration in California is currently about 20% (and rising) of the total electorate.  This may be part of the missing 9%.

by mboehm 2008-09-22 01:05PM | 0 recs
So who are the independents, anyway?

Surely 30% of voters aren't registered independents.  A state-by-state breakdown would be helpful here.

by KTinOhio 2008-09-22 01:13PM | 0 recs
538 Pollster Rankings

All this makes it so much more amazing that R200K remains one of the most accurate pollsters.

http://www.fivethirtyeight.com/search/la bel/pollster%20ratings

Think how accurate they could be if they actually did things like Fox, Gallup, ARG, CBS, IA, PPP, LA Times...

Oh never mind, all those other shops are less accurate, scratch that.  

by nextgen 2008-09-22 12:29PM | 0 recs
apples meet oranges

Jerome's comparing 2004 results. which seems dunderheaded, as we aren't in the last week of the race.

by Blog Surrogate 57 2008-09-22 12:42PM | 0 recs
not entirely imposibble

the problem with Judis is he's taking exit polls as a factual basis. No one knows what the real numbers were in 2004. They are in fact, no better then the latest r2k or Rasmussen poll.

But I digress. The U.S. is 13% black and over 15% latino now. If we can't break 21% shame on the Democratic party and our ground game.

In fact, while r2k is perhaps overreaching in their minority sample size, those numbers are not entirely bogus or far fetched. Latinos are more politically energized then ever and blacks have a chance to vote for an African American for the first time in their lives.

by alex100 2008-09-22 12:37PM | 0 recs
Re: not entirely imposibble

Not to defend Jerome, but it possible that the AA and Hispanic populations may skew younger, which would reduce the percentage of possible voters.

by the mystical vortexes of sedona 2008-09-22 01:37PM | 0 recs
Re: not entirely imposibble

that's a completely valid point and I completely agree with you.

I'm not saying blacks and Latinos will make up 26% of the voting population as r2k breaks down.

However, it is faulty logic to peg these numbers against exit polls from 2004. We'd just be compounding the problems that are inherent in exit polls.

by alex100 2008-09-22 02:17PM | 0 recs
Re: not entirely imposibble

Latino skews younger and roughly 40% are foreign born. Of those foreign born many are not citizens, and thus ineligible to vote.

by souvarine 2008-09-22 02:09PM | 0 recs
Re: not entirely imposibble

of course. I've said this many times myself already.

according to Pew, 75% of Latinos are native born and 25% are naturalized citizens.

by alex100 2008-09-22 02:42PM | 0 recs
of under 9% of eligible voters

Sure, Pew has Latinos at just shy of 9% of eligible voters, of that 9% of eligible voters 75% are native and 25% are naturalized. Which roughly works out, if Latinos are just under 15% of population, and 40% of those are foreign born, then around 9% of the electorate is eligible voters.

by souvarine 2008-09-22 05:00PM | 0 recs
Re: Research 2000 polling

Well...
Here in Denver I have been heavily involved with the Obama campaign both with registration and GOTV efforts since June in heavy Latino areas.
We go in twos with one of us bi-lingual. The knowledge of voter eligible immigrants is higher than I expected.
The desire to vote for a candidate with darker melanin is high. The Latino turn out in Denver and Weld Counties will be of historic proportions.
This will result in not only a CO victory for Obama but a defeat for Musgrave...
Just this week, Obama open another office in Dem heavy North Denver...It is geared toward a GOTV effort. Those of you all ready involved are aware of the organization in your area. For those of you with doubts? Check it out. I can guarantee you,; you have not seen or been a part of anything like it.

It is not too late to put your voice and/or your feet on the same line as your mind.
http://my.barackobama.com/page/content/a ctioncenter

"Move your mind, your ass will follow" ...George Clinton

by nogo postal 2008-09-22 12:41PM | 0 recs
I agree with Jerome

God, that sounds weird to say.

But the Hispanic numbers are frankly, fucking impossible.  

14.8% of the U.S. population is Latino, but around 25% of that is probably undocumented immigrants.  This whittles us down to 11.1%.  Then you have to factor in Latinos are more likely to also be legal immigrants than non-Latino whites or blacks, and a disproportionate number of Latinos being young, you'd be lucky if Latinos make up 10% of the eligible electorate.  

13% is just damn foolish, unless you think white Americans are going to stay home.  

by telephasic 2008-09-22 12:51PM | 0 recs
Re: I agree with Jerome

Latinos made up 15.1% as of 2007.

Not all undocumented or illegals are counted by the census bureau. But yeah, it will hurt us.

With that said, a factor that will hold the Latino vote under their national average is age. It is a younger demographic.

Anything under 10% would be a failure IMO.

by alex100 2008-09-22 02:06PM | 0 recs
I've always said

that the best use of daily tracking polls is to gauge the changes in voter sentiment, nationally. And in that regard, my R2K polling has consistently picked up the same movement everyone else has. And in fact, no one else out there is providing daily splits on the approval/disapproval numbers of the candidates. That allowed us to talk about Palin's collapse before anyone else had.

That's the reason I got my own daily tracker -- so I wouldn't have to beg other pollsters for that information.

If you want to adjust the numbers 2-3 points in the other direction to "compensate", be free. That's why I made sure all data was made public, unlike everyone else that still hides the composition of their samples. And at the end of the day, other than Rasmussen, the other three tracking polls are within two points of each other.

But I find it funny that you complain about R2K being too pro-Obama, but then approvingly point to CNN's latest poll that has Obama up to 51 percent, higher than anything R2K has shown. AND, you have no clue what that CNN sample looks like. You might be equally horrified if you saw how they broke it down. Or how Rasmussen breaks it down. You're assuming they've got better samples, but you're doing so on faith.

I don't do faith very well, prefering past performance. And on that front, R2K kicked CNN's ass in the primaries. And sure, while past performance is no indicator of future results (see Zogby), it does lead me to give R2K the benefit of the doubt.

by kos 2008-09-22 01:19PM | 0 recs
Word

BTW, you're welcome to comment on our silliness at Motley Moose.  We aren't as trolliscious, but I'm sure we can come up with something contemporaneously controversial to clamor over...

:~)

by chrisblask 2008-09-22 03:59PM | 0 recs
Re: I've always said

We do agree that national polls are worth only their trend. And you are right, I might be horrified of CNN's internals (that'd be the follow-up post), but I can't see them, and who really cares about that red herring.

That doesn't do anything to the lesson the criticism here, at all.

I'm not complaining about it being "too Obama" but of it being illogical in its turnout model (same as Judis & Bowers).

Research 2000 has no national polling history to lean on, so your faith in the face of their wacky assumptions is all the more questionable, and your extreme defensiveness over the criticism (without at all refuting it) raises the question:  Are you carrying water for them because you are paying for the poll, or because you think they are right?

by Jerome Armstrong 2008-09-22 05:48PM | 0 recs
Re: I've always said

Hawaii burning up your balls?

I would have loved to have seen your internals when you made that proclamation.

you were what? at least 53% point off!

I know I will be banned but please do everyone a favor. Leave the site to the others writers.

You don't have the flare for the critical component. Not like Todd or anyone else out there.

by alex100 2008-09-22 09:37PM | 0 recs
Kos

Good Analysis kos, I agree with your analysis. Nate is far more credible than Jerome as far numbers are concerned and he lists R2K as high on the credibility list.

by YourConcernsAreNoted 2008-09-22 01:27PM | 0 recs
Re: Judis & Bowers ...

Check it,

I think you guys are really underestimating the turnout for AA this year.  I'm guessing that as in the rest of America, AA are in the minority when it comes to blogging, but I bet not many of the MSM or white people in general listen to Urban radio.  People like Tom Joiner, Michael Baisden, Steve Harvey, to name a few, who are nationally syndicated urban radio personalities.
I'm telling you, ya'll should check these shows out one day.  All they talk about daily seems to be this election, and Barack Obama.  They are on point when it comes to registration, voter suppression, and GOTV.  

Not only that, there are AA church groups, fellowships, greek organizations, NAACP who are all doing all that they can to GOTV, and help elect Barack Obama.  Trust me when I tell you that this election cycle, you won't have to worry about the turnout of AA voters.  It's the voter suppression tactics that will be attempted that will need to be watched out for.  And even on that subject, Urban Radio host daily talk to their listeners about voter registration and voter purges.  

There are a lot of Urban Radio stations, they may not individually have the numbers that Rush or Hannity have, but as a whole, they reach a lot of young people, AA, other "urban voters".  

Don't underestimate AA voters this year, you will be in for a surprise.  This won't be another "why didn't AA turnout" type of election.  This is our time, this is our moment.

by lamh3176 2008-09-22 04:25PM | 0 recs

Diaries

Advertise Blogads