The new CNN/Time poll. What are we going to do with it? The top-line of the poll says Obama leads by 5, 49 to 44 but everything underneath says that poll has little basis in reality.
Independents
In 2008 candidate Obama won Ohio by 5 percentage point while winning Independents by 8 points. In today’s CNN/Time poll, Mitt Romney leads with Independents by an incredible 13-points but is down overall by 5-points. That is simply not plausible.
Party ID
The party ID in this poll was D +9 (Dem 37, Rep 28, Ind 29). This compares to D +8 in 2008 (Dem 39, Rep 31, Ind 30) and R +5 in 2004 (Dem 35, Rep 40, Ind 25). There is no chance the Democrat turnout advantage will exceed Obama’s 2008 best in a generation turnout which we write as D +8 based on the CNN party ID generally used. This is even though the actual 2008 party ID was really only D +5 making this D +9 that much more unrealistic.
Early voting and the likely voter screen
Early voting is creating a unique problem for polling organizations this year. Since Democrats tend to vote early, you see the Democrat candidate typically leading by wide margins in early voting. According to the CNN survey “respondents who say they have already voted, Obama holds a two-to-one lead over Romney, 60% to 30%.” So we know upfront in early voting there is an overwhelming number of Democrats.
When it comes to polls, all voters who said they already voted make it through the likely voter screen and end up in the final results. Approximately 17 percent of the respondents were early voters while 83% had yet to vote. This means a 17% segment of those polled are guaranteed to make it through the likely voter screen and we know that group supported President Obama by a 2-to-1 margin. This inherently over-samples Democrats which practically guarantees a favorable result for Democrats. The problem with early voting and polling ahead of election day is the results will skew in favor of the party with the higher early turnout, in this case the Democrats. That’s one of the ways you end up with a party ID of D +9 when there is no chance of that turnout occurring on election day.
It was nice of CNN and Time to spend the money to run a poll of all-important Ohio, but I think this poll did little more give false hope to Democrats who are likely in for a rude awakening on election night thanks to unrealistic polls like this one.
Battleground Watch 
64 Comments
Thank you for the fine rebuke of this hit piece. I tore it open as well and ran the methodology of D +9.3. Keep up the good work!
I am sitting in the smallest room of my house. I have this poll before me. Soon it will be behind me.
lol…. mental picture being developed
Flush twice, just to make sure it goes where it belongs.
The MSM will not under any circumstances show a Romney lead for fear that the game is up for their precious.
It isn’t like we weren’t warned about these kinds of polls. You said we could expect them and now they come. Once analyzed though the truth comes out as to who is where right now.
Thank you Keith for staying on top of things!
So if you look at this realistically plus Ras & Suffolk, Romney is tied or +1 at best?
that’s about right. Off to dinner. talk to you peeps later.
When i saw D+9 and Romney up with independents i had to laugh, this is becoming an OVERWHELMING JOKE! This is for real, insanity at its finest. They are just trying to skew the polls on RCP, the averages. And provide hopes to the democrats lack of early vote and absentee requests compared to 2008. Romney i heard is going to Wisconsin, he should stop by Pennsylvania, Wisconsin, Michigan, Minnesota ALL IN ONE DAY 4 rallies in the largest republican counties of each state, for just ONE day.
“trying to skew the polls on RCP”
And/or shape public expectations so that when they run the ballot boxes to reelect Obama, no one is surprised. It woiuldn’t look very believable if the majority of the polls showed Romney +2 and the results came out Obama +5. This is how they’ve been running this country for the past 4 years, smoke and mirrors — bailouts, employment numbers, stock market — al lsmoke and mirros — and so it goes on down the line, what makes you think they’d stop with an election?
The unity of the mainstream press to use their access into people’s houses to speasd propaganda is unreal this year. (don’t get me wrong, I think Fox News is skewed too (tv more than web), but never have I seen the unity up and down media soruces – tv news, national, local, web — in favor of one political candidate. Sometimes parading talking points as news. It’s been unbelievable.
Keith, good analysis… and I agree that based on the data provided (independent representation that seems implausible, and the large Dem representation in the sample) that there is no way that any faith could be placed in the obviously skewed results.
While I generally do not hold Democrats in high regard and I frequently question their ability to think at anything above a GED level, there have to be some in the Obama campaign that are at least functional enough to realize that these polls are not representative of reality.
They have either taken medication in sufficient quantity to no longer care, or they must see the iceberg that they are about to hit and are looking for the life jackets that will be needed come the morning of Nov 7th.
One thing about reporting of polls is that the topline is usually just what’s reported. Very little analysis is actually done on TV. And unless punditry on a show actually care to present it (hit or miss on that), people rarely ever consider the stuff Keith and others consider. I (an Obama guy…don’t worry, there are smart ones too 😉 ) began questioning the polls, wasn’t afraid to hear stuff from different perspectives and found this page. How many people, either left or right actually bother to dig into such detail? How many partisans will listen to anything that comes from a perspective that’s not on their partisan side? I barely have the time myself, I kinda make the time. And I have to learn to turn off instant judgement and list listen. The media is kinda dumb and so as a result, the watchers and readers aren’t as knowledgeable on the issues as they could be.
well said
allthingsgeography1, I have to hand it to you. You support the other guy but you come here and make great points and unafraid to look at data objectively. I wouldn’t have been able to do that on a lefty site, I admit. Kudos. You add a lot to this site.
Thanks Porchlight. I try 🙂 I’m just happy there is a site that isn’t full of trolling nonsense commentary and garbage for once. The hardcore partisanship gets tiring after awhile, sometimes I just want to be able to talk about reality and facts (and the interpretation of those facts). Keith has created a positive atmosphere for dialogue for anyone who’s interested in communicating in such a medium. Can’t complain (except if Romney wins, then i’ll complain a lot 😉 )
We can’t ascribe everything to duplicity. We have an information revolution going on and reporters have all they can do to catch up with the blogosphere.
so basically what I gather from these polls and what the poll experts (Silver, etc) say are party ID doesn’t matter…they call, some one answers they are asked questions if they make it through the likely voter screen then they are in. If it happens to be 2% or 40% more of one party, so be it. What matters is the weighting of demographics (race, gender, etc)?
So the long and short is MORE democrats are answering the phone and answering the polls?
Strange isn’t it? Even though the process of voting involves politics, you can’t weight the political demographic to estimate who may actually show up at the polls if the election were held the day the poll is taken.
We also hear that Democrats are more likely to talk to pollsters than Republicans. So many people refuse to talk to them and who knows how many people misrepresent themselves. I’ve done that before.
The attitude of the campaigns tells you a lot. They do the best polling. Romney also likely has access to the congressional polls from Republican House members that tend to poll their district in regards to themselves and the Presidential ticket. And the Romney campaign likely commissions polls in Democratic districts in the swing states as well. We’re talking about 10,000+ in the sample for a state like Florida.
Romney’s confidence is telling as is Obama’s tiredness. It was the opposite in 2008. Obama and his campaign were energized. Fighting to swim upstream is tiring and the momentum continues to be with Romney.
I have to agree. In the 3rd debate it was very apparent who was more confident of victory (Romney) and who was in full realization that he was trailing and had to make up ground (Obama).
Certainly things could change, but everything seems to be moving in Romney’s direction.
This will make you feel better….
http://unskewedpolls.com/unskewed_projection_2012%20president_02.cfm
The only winners if early voting really is skewing the polling this badly will be the political scientists, who will spend the next four years presenting papers at the APSA and other political science meetings presenting papers about 1) how early voting distorts polling and 2) what to do to control for it. As an aside, am I the only one wondering if this is the year polling comes completely a cropper because of all the difficulties that are besetting it, from low response rates, to cell phones vs. land lines, to electorate modeling, to early voting?
A dubious poll to say the least… D+9 is a joke… credibility means nothing to this bunch… A fifth grader would know better than to publish this garbage.
Keith…can you do an early voting graph for Ohio like the excellent one on Nevada???
+ 1, on this ask
It’s not really my thing. There’s a lot I don’t like about early voting. Nevada is an exception due to some extenuating circumstances I won’t go into. The Ohio numbers are also more dubious than the Nevada #s so drawing inferences is even harder. Other people like Larry Schweikart @LSchweikart follow this closely and do a much better job than I could. I’d check him out.
Links to Schweikart analysis a couple days ago:
http://www.nationalreview.com/campaign-spot/331052/early-vote-totals-saturday-morning
D +9 (Dem 37, Rep 28, Ind 29) – adds up to 94 what happened to the other 6%
You do realize how silly this whole Party ID thing is, no? No serious polling analyst, including conservatives at RCP, focus on this.
In any case the future of polling analysis is clearly with the poll aggregators. Note that RCP and 538 agree O is ahead in the EC count.
Definition of insanity is feeding Peter over and over and expecting a different result.
By the way, did you finally understand what happened with Ras in 2000?
Yep. You didn’t see my follow up from that post and didn’t realize he started a new polling company in 2003 and admitted I didn’t have all the facts. With that said, I’m sure you will agree 😉 He obviously changed his methods since his accuracy in Presidential elections has been pretty darn good since.
With that said, I’m sure you will come on here November 7th and admitt that MAYBE these polling firms were off regarding trying to model the turnout here in ’12?
Well, yes, his accuracy did improve but given he was off by 8 points that’s not saying a lot. He wasn’t that accurate in 2010 either, off by 5 points, obviously pro R.
As for Nov 7, here are the scenarios:
1. Final poll averages show O winning and he wins. Proves that all this Party ID debate was a waste of time.
2. Poll averages change between now and Nov 6 so that final poll averages put R ahead and R wins. In that case polls are vindicated ( we can debate why the last minute change if it happens)
3. Final polls show O ahead on average but he loses. Shows problems with polls. If the difference can be attributed to bad Party ID you and Keith will be proven right.
Off by 8? Can u send me the link?
I started following him in ’04 and was off 0.5% on Kerry I believe. I guess I could send u the link.
In 08′ he tied Pew for the most accurate pollster. That was from a non-partisan rating bunch.
Out of the state races, not as accurate. He missed Ohio and Indiana (plus a couple of others), but this is were I believe even the good pollsters have trouble gaging turn out in the historic election. Even Nate Silver (with the Obama internals missed either Indiana or North Carolina..can’t remember which one).
2010 was a wave year. Did anybody get it right? Not defending him. I know Marist, Quinnipiac, PPP didn’t get it right. Off year elections, like State elections are similar but do have differences.
I know you have an innate dislike for him (and that is fine), but I feel like he has less volitilty and has been quite close the last 2. The others seem to be all over the board.
In 2000, CBS had bush by 0.2% on the last day before the election. I tend to put more weight on their polls but have not analyzed them at the level being done here. Just looking at final results. They have Obama by 2 nationally. Not that things can change though in 12 years for skwedness. Rass tends to move his results as the last day approaches. Let’s wait and see how much it moves last few days leading up to election.
“Rass tends to move his results as the last day approaches. Let’s wait and see how much it moves last few days leading up to election.” Care to back that statement up? Rasmussen was remarkably stable during 2008.
I think 2 things with that. 1) Uncommitted voters start to make their choices. 2). He adjusts his party ID as it becomes apparent turnout/enthusiasm.
You never explained what was “silly” about it.
A poll with Party ID badly out of whack with (a) past results and (b) expected turnout has no relation to reality.
I can poll 70% more Republicans and get Romney up in Vermont and Rhode Island. Doesn’t mean that’s any resemblance to the true turnout.
“You do realize how silly this whole Party ID thing is, no? No serious polling analyst, including conservatives at RCP, focus on this.”
Peter, all I have seen is generalizations about the number of Tea Party members who show up as Independents instead of Republicans. Is this pretty much it or is there some other reason we should not be suspicious of party breakdowns in the polls that show many more Democrats surveyed than Republicans (relative to what has been observed in previous elections)?
This seems dubious to me. When Independents are pushed to disclose leaning Democrat or leaning Republican, they seem to break relatively evenly (in the small number of polls I’ve seen that make such an effort).
Also, there should be evidence that midterm election results turned out many more Republican votes than the polls indicated based on party breakdown. Is this the case? If not, when did these Tea Party members switch affiliation, after the midterms?
I honestly don’t know what to think. With the benefit of hindsight we can verify whether or not the party breakdown of the actual election tended to match the most accurate polls. I think that generally speaking, this has been the case, and that is why the analysts you mention do not “focus” on this issue. There was no issue in the past because we never had an alleged hidden supply of partisans disguised as Independents before (Tea Party members). I am not convinced these faux-Independents exist in any greater number on the Republican side than on the Democratic side. So for now I will remain suspicious of polls that show more Democrats sampled than actually voted in 2008, a generational event.
Michael Barone had this to say about midterms:
“Starting in 1996…the Democratic percentage of the vote for president and for the House of Representatives have differed by no more than 1 percent.
“In addition, the percentages for the two parties in the popular vote for the House in the last three off-year elections have been almost exactly the same as the percentages for the parties in the vote for president two years later…
“In 1998, the popular vote for the House was 49 percent to 48 percent Republican. In 2000, the popular vote for president was 48 percent to 48 percent Democratic.
“In 2002, the popular vote for the House was 51 percent to 46 percent Republican. In 2004, the popular vote for president was 51 percent to 48 percent Republican.
“In 2006, the popular vote for the House was 53 percent to 45 percent Democratic. In 2008, the popular vote for president was 53 percent to 46 percent Democratic.”
Why shouldn’t we expect a result similar to the popular vote for the House in 2010? Those people haven’t gone anywhere. They’re just as eager to send a message this time around.
Reality is that this is terrible for Obama. D+9 and you can’t clear 50%? He needs to have historic turnout in Ohio for this to work for him. As National Review showed, 2008 was really D+5 and 2010 was D+1 while 2004 was R+4. Let’s also not forget that Ralph Reed is involved in evangelical turnout this time like he wasn’t in 2008 and that can make some difference in Ohio. Despite the D+9, Ohioans give Romney a FOUR point edge on the economy, the critical issue this time out. And his white voter support is behind most other states, meaning white Democrats are oversampled.
I believe the early voting in Ohio is trending well for the GOP, ahead of 2008 with a narrower gap between the Democrats and them.
When you dig further, you see a huge gap in between “definitely vote” and “probably vote” that favors Romney by 13 points – in a D+9 sample. You also dig and see the Romney voters are much more engaged, which indicates people will turn as the time to vote actually hits.
Also, looking at early voting, it appears they misrepresented the population areas by their weighting. And it appears that unweighted would have favored Democrats even more at D+9.5, putting Republicans at an historic low turnout for Ohio at under 30%.
This is pretty much a garbage poll.
On early voting in Ohio, Adrian Gray tweets a continuing good trend for us. More like 2004 than 2008:
OHIO: Bellwether counties 2012: Republican turnout beating Democrat turnout as % of 2008 turnout (106% to 74%).
OHIO: My bellwether counties (Hamilton, Lake, Montgomery and Stark) went combined 51-49 for Bush in 2004 and 52-46 for Obama in 2008.
OHIO: In bellwether counties, 7% more D’s than R’s voted early this point in 2008. Today, 1% more R’s have voted. A 8.4% net GOP gain.
I wish they’d give us the RV count so we can see how tight their LV screen is.
Today’s IBD/TIPP poll has a LV screen you can drive a truck full of dead Chicago voters through:
922 LV / 993 RV = projected turnout of 92.8%
This is just impossibly high turnout, which we haven’t seen since 1964 (and before motor-voter laws made it so that so many uninterested people became RVs). I think this wildly loose likely voter screening is why we’re getting such goofball DRI breakdowns. Remember, this is Axelrod’s big complaint against Gallup – their LV screen is too tight. Well, it’s not, if you assume anything other than historically (wildly) high turnout this year. Gallup’s LV screen is still too loose when compared to recent LV/RV percentage turnout – but it shows a big Romney lead.
They did. 783 RV versus 742 LV. 92.46%. It’s in the PDF.
Ah, thanks JeffR, I looked but didn’t find it.
So yet another insanely optimistic model of turnout.
And this fits in with what I’ve said. Firstly, Romney voters are more enthusiastic and likely to turn out. Obama also has a lot of soft support. This poll showed Romney up by 16 points with men and Obama up 16 points with women, so it expects higher female turnout than male turnout to account for the 5 point lead. Part of this is likely because white males did stay home in large numbers during 2008 in Ohio as did evangelical voters. This is still predicting a 2008 style electorate and, frankly, I do believe the Romney up by 16 with men, but I’d guess the gap with women is narrower and it is made up of too many Democratic women as well.
There are lots of problems with this poll.
Are there any polls out there that that you’ve found skew to republicans?
Here’s a Michigan poll showing it tied 47-47. May be an outlier, but the most interesting part of it is that they had to make 35,000 calls to get 1100 respondents! http://www.myfoxdetroit.com/story/19905504/michigan-poll-obama-and-romney-in-dead-heat
Been reading for a few weeks, but first time to comment. Keith, love the commentary and website. Just my two cents, I agree with most the MSM is trying to save Obama and sway opinion, but I think there is something even deeper to their method. I think they see the internals an know O is in trouble, but don’t want to report the truth in fear that the inevitable will lead to lower election night ratings
Plenty of truth in that Brad. the networks would be PISSED if their news teams reported things that would turn down viewership regardless of their politics.
It has to be perceived as close for people (the left at least, those of us on the right will be celebrating with pudding and Value-rite) to tune in 11/6/12 and if the “preference cascade” is beginning to happen they can’t let the sheeple know
Which internals are you talking about? They wouldn’t get a campaign’s internals..especially if they were bad. I believe those are highly confidential and the most expensive (a million per month per DATECH guy). I believe the polls and the internals that ABC/CNN/CBS/NBC get what they pay for which are cheap polls and run with the data they can get ( or the people that pick up the phone). There are internals there, but I don’t believe a campaign in general gives that data out.
And the internals I am refereing to are the ones they are seeing from the polls they have commissioned, not the campaign polls, but the one paid for by the media outlets
Actually, if BO loses, I believe MSNBC will get their highest ratings of the year 😉
I agree they will, but it has to be “unexpected”
That’s a good theory but this MSM bias has been going on all year/summer. That’s why you saw such a monumental shift in opinion towards Romney after the first debate. For the first time, unfettered from the summer slant of the major networks (and where people were getting their info from through the summer), pelople saw a man that was actually capable, smart, intelligent and eager to be president. It was the perfect storm too because Obama took his task too lightly and did not prepare for the first one. Obviously he prepared for the second and third debates.
So this has been going on all summer. Which argues that there’s possibly a bigger plan at work. This may have been one reason why Microsoft dropped their affiliation with NBC. NBC wanted to take the station in a more liberal direction. Just speculation but the timing is interesting. Yes, the liberal crescendo at stations like MSNBC and NBC Nightly News and so forth has gotten rediculous in the last few weeks. I mean it’s hard to tell MSNBC from a partisan democratic website.
One could argue that all the networks were doing was giving viewers what they wanted when Obama was ahead. But their tatics didn’t change when Romney made the comeback. Instead they got worse. This also argues for a partisan ploy.
The Time poll is a very complex one to reweight given the mix of likely voters and those who have voted. We focused on tables 9 (likely voters) and table 10 (already voted). The LV in table 9 broke down as 206D/174R/176I/556total. Table 10 voted was 67D/32R/39I/138total or when totaled for party weights of .393D/.297R/.31I when the don’t know/won’t tell responses were factored in. So this was a D+9.6% model which is high historically but also had a +5% Obama lead.
For the internals the combined democrats broke 91.6%/5.5% for Obama, the republicans broke 84.0%/9.2% for Romney and the independents broke 36.3%Obama/49.8% Romney. Data coming from table 9 [Dems go 186d/14r repubs go 145r/18d and independents go 57d/96r in real numbers]. Table 10 was dems 64d/1r, repubs 28r/1d and independents 21d/11r.
Next we considered the effect of early voitng on the polls. For this poll of the 694 adjusted voters we included, 138 were early voters or 19.9%. From a historical perspective, OH early voting has been 10.6% in 2004, 29.7% in 2008 and 25.8% in 2010 (with voter laws changed in 2005 and seemingly reflected in the data). We went with the 19.9% early voting weight as analyzed here, though one could reasonably argue that 25%-30% is more valid and our analysis underestimates early voting and thus skews favorably to Romney.
Click to access EarlyVotingReport.pdf
On balance, models show Romney holding the D/R+0% model through the R+5% model and Obama holding the D+1% through D+5% model. Our models also show Romney leading in a 2004 and 2010 model with this poll indicating an Obama advantage in the 2008, 04/08 average and 08/10 average models. We continue to see a slight Romney lead in the state, but do note recent momentum both in the state and nationally has turned to Obama in our models.
http://insidepolling.com/battleground_states/ohio
The Republican field director put out a press release today stating that among Ohio early in person voters and absentee ballot voters, Dems lead Repubs 34.4% to 28% with independents making up 37.6%. With 800,000 votes in, that means 52,000 more Dems than Repubs have voted. Because exit polls show Ohio Dems are more likely to vote GOP than Ohio Repubs are likely to vote Dem, Obama’s lead among all Dems and Repubs to vote so far is 35,000 – 50,000 with early voting 2/3 over. His final margin among early voting Dems and Repubs will be 50,000 – 75,000. The polls indicate Romney will beat Obama by 100,000 among Ohio independents. Furthermore, McCain had 140,000 more GOP voters on election day than Obama had Dems and Romney is likely to top this margin. That all points to a Romney win of 170,000 to 250,000 votes, which would be about 3 to 4%. This is entirely consistent with the demographically credible polls (Rasmussen, Gravis, Suffolk) showing it even with 5-6% undecided and the undecided breaking strongly toward the challenger at the end.
For Obama to win, he needs a level of Dem brand loyalty never seen before, getting him fully 75,000 ahead among Dems and Repubs wh vote early. He also needs to beat almost every poll’s projection and run within 5% of Romney among independents, so he only drops down to even with Romney. He then needs more Dems to show up on election day than GOP, which has never happened in Ohio. If you need all three of those things to happen together, it’s time to turn out the lights, cause the party’s over.
Where can I find that info, because that’s the best news I have heard in nearly 4 years!!
Keith, thanks for the site. I’ve been reading, but my first post. I wanted to share another point about Ohio polling. My 75 yr old aunt lives in the Cincinnati area. Has a land line phone. She’s been getting PUMMELED with poll calls to her house. I should also point out she’s a registered Republican. Her senior group decided IN AUGUST to start telling the polls they were voting for Obama. I was there just before Labor Day and watched (listened) to her follow through on it. The reason? To paint a picture that looks better for O than it really is .. and so my point is .. with the already heavily skewed D+9 numbers that previous posters have covered so well, realize that O numbers are coming out better still than they really look.
that’s funny
Obama latest 48 hr swing tour is like a desperate Olympic Mile Runner going into his finish kick on the 3rd lap …
I’m no statistician, and I know this post is kind of old now, but it seems to me they oversampled–by a HUGE margin–white, female Democrats in this. In the first place, whites make up 86% of this poll sample. That’s high for Ohio, isn’t it? In addition, the breakdown of Dems is 35% male, 65% female. There is NO WAY there are twice as many female Democrats as male Democrats in the general population of any state. Also, Republicans are broken down 50/50 between males and females which also doesn’t sound right to me. I think there are more male Republicans easily. If I was a conspiracy type of guy, I’d think the pollster went out looking for white, liberal females and found them.
“In my gut” I believe that rain is not really wet.
All the hot air about Ohio polls being tilted unfairly toward Dem’s is just that, hot air.
Click to access Absentee_and_Early_Voting_in_Ohio_and_Virginia.pdf
Like with global warming, evolution and the science of rape (solely a GOP study), you people keep talking to each other and not to someone who actually basis their opinion on ALL the facts.
Ohio is already lost. You better win VA, FL, NV, WI, CO.
Stop smoking your own bad weed.
One Trackback
[…] turnout which we write as D +8 based on the CNN party ID generally used. This is even though the actual 2008 party ID was really only D +5 making this D +9 that much more […]