The Truth About Polling: Yes, Romney Is Probably Tied or Winning
First: when you read a poll, you’ll always see it stated as something like “53 percent Democratic, margin of error of plus or minus 3.5 percent.” Here’s how you should read that: “The polling company believes that if the election were taken today, there is 1 chance in 20 that the actual result will be more than 56.5 percent Democrat or less than 49.5 percent Democrat.”
You see, that’s what the “margin of error” really means: by statistical methods, they believe that 19 out of 20 times — or 95 percent of the time — the real value would come out between 49.5 percent and 56.5 percent.
Second, and this is where the controversy is coming now: the quality of those results depends on the accuracy of that original model.
So, when a Quinnipiac poll comes out and says that Ohio is now 53 percent Obama, what should you do? Ready, class?
First, we restate that. I can’t find the margin of error for this poll readily (this should make you suspicious at the start), but if you can’t find one, you’ll always be close to right if you say it’s 3 percent plus or minus. So we re-read this as “they’re saying there’s 1 chance in 20 it will come out either above 56 or below 50.”
Second, we look for the polls’ “internals” — in other words, that turnout model. Now, here’s where this poll gets interesting. Hugh Hewitt interviewed Peter Brown, the head of Quinnipiac, and challenged him directly on the turnout model:
HH: I want to start with the models, which are creating quite a lot of controversy. In Florida, the model that Quinnipiac used gave Democrats a nine-point edge in turnout. In Ohio, the sample had an eight-point Democratic advantage. What’s the reasoning behind those models?
PB: Well, what is important to understand is that the way Quinnipiac and most other major polls do their sampling is we do not wait for party ID. We ask voters, or the people we interview, do they consider themselves a Democrat, a Republican, an independent or a member of a minor party. And that’s different than asking them what their party registration is. What you’re comparing it to is party registration. In other words, when someone starts as a voter, they have the opportunity of, in most states, of being a Republican, a Democrat, or a member of a minor party or unaffiliated … (emphasis added)
A little later, he says:
HH: Why would guys run a poll with nine percent more Democrats than Republicans when that percentage advantage, I mean, if you’re trying to tell people how the state is going to go, I don’t think this is particularly helpful, because you’ve oversampled Democrats, right?
PB: But we didn’t set out to oversample Democrats. We did our normal, random digit dial way of calling people. And there were, these are likely voters. They had to pass a screen. Because it’s a presidential year, it’s not a particularly heavy screen.
In other words, the way Quinnipiac does it is they assume that the thousand or so people they poll on a certain day really do represent the population, and if that comes out with a D+8 distribution, that’s the way they report it.
The problem with this is that MANY polls show a skewed breakdown of Democratic voters in swing states. If there are many polls that show this breakdown, then its starts to get statistically significant that that IS ACTUALLY the voter breakdown, and you can’t call it skewed anymore. So one poll is not that important, but many showing the same thing are. This is what were seeing..and it shows Obama winning by about 6-10 in most swing states at this exact moment in the race.
If you are telling a falsehood, repeating it over and over doesn’t make it true. I can’t see the logic in your comment.
There’s no collusion among these pollsters. They do their random samples independent of one another. If you do many independent experiments that show the same or similar results, you start to support your theory. The theory is that AT THIS VERY MOMENT, the makeup of the electorate in swing states is overwhelmingly Democrat. That theory is getting a lot of support.
That could change of course…
That there is comedy gold.
i don’t think FOX news is calling up the new york times asking them to use the same sample population for their poll. maybe, but i doubt it
The flaw that isn’t mentioned here is the fact that Brown’s polling comes down to using the polling sample known as “adults.” It is not a “likely voters” sample or even a “registered voters” sample. His sample base is not taken from voter registration rolls and he is admitting as such by saying that they ask what party affiliation someone belongs to over the phone without any forehand knowledge of the person’s political affiliation. The “adult” sample is much less accurate than a “likely voter” sample. Basically, he’s lying if says his sample base is “likely voters.”
“The theory is that AT THIS VERY MOMENT, the makeup of the electorate in swing states is overwhelmingly Democrat. That theory is getting a lot of support.”
Lots of support from who? The press?
They also believed in the Coming Ice Age *AND* Catastrophic Anthropogenic Global Warming.
And the Hitler Diaries.
And Obama.
The consensus of polls support that theory
I actually don’t think it’s so unlikely if you consider the changes in the GOP the past 4 years. With the rise of the Tea Party, there are a lot of conservatives who don’t necessarily love the GOP and thus it’s not outlandish that many who previously identified as GOP would now be self-identifying as Independents when asked in phone surveys.
This would also explain the somewhat perplexing fact that Obama is still ahead in many swing states where Romney is, nonetheless, winning Independents by a pretty large margin. A lot of people are confused by this, but it makes perfect sense if for some reason many conservatives are now self-identifying as independents when polled.
I do not understand where all this support from obama in the swing states is coming from. Less than two short years ago in such Democratic strongholds as Michigan, Wisconsin, Pennsylvania, Ohio, and NJ ALL threw out Democratic Governors, Legislatures, or both. Some for the first time in a generation are now Republican.
And NOW we are supposed to believe that the people in these states that were so ready to throw out Obama two years ago that they showed up in record numbers and made history are NOW ready to give the SAME guy four more years?
I don’t get it, what has changed for the better in that time period? Not enough to make people want more or the Great One (and I am NOT talking about Mark Levin) IMO.
This election is over. It is going to be a rout. The apoplexy on the experts faces at MSNBC and other venues is going to be epic.
I cannot wait!
I don’t believe any poll conducted by Obama’s mainstream media or their affiliates. I have started checking poll numbers at “UnSkewedPolls.com” where sampling is not heavily weighted toward any one demographic. These polls have pretty consistently shown Romney in the lead. I tend to believe that is the real picture since Obama has been an utter failure.
There was a news story yesterday out of Texas saying the Dems have started handing out free cell phones to anyone who registers as a Democrat and promises to vote for Obama. Over 250,000 had registered so far. I may be wrong, but isn’t that illegal? Last time I checked, bribery is a felony. The Dems are really getting desperate from the looks of it.
Bart, I don’t think I buy that the theory has all that much support, because the samples are so different. Think about the various values as samples from a normal random variable: You would expect the variance to be getting smaller as more and more samples are taken. But it isn’t. This suggests strongly that something is wrong.
What it suggests strongly is that pollsters – most of whom are Dems/Libs – are trying to depress GOP/conservative turnout by making it look as though Romney has no chance by skewing their polls to make it look that way by way, way oversampling Dems.
Short version…many/most of the polls being published are pro-Dem propaganda. Period. And even the few good ones aren’t very good this far out from an election.
The reason that Rasmussen doesn’t show this is because they are one of the only firms to weight their polling to account for party affiliation. Party affiliation is often a fluid thing and no other polling firm uses this methodology anymore because it causes inaccuracy. Rasmussen was 5.8 points off on average for their entire polling for the 2010 elections. That’s the most of any pollster who worked during those elections.
Bart, I don’t think the data bear that out: http://www.rasmussenreports.com/public_content/about_us/public_relations/press_room/press_releases/the_rasmussen_reports_track_record
Pretty sure Bart got that number from 538. You can read their analysis of their performance in 2010 here:
http://fivethirtyeight.blogs.nytimes.com/2010/11/04/rasmussen-polls-were-biased-and-inaccurate-quinnipiac-surveyusa-performed-strongly/
Gentlemen & Ladies,
I have. Been following the polls closely since the spring and the pro Romney
Shift has been very slow but steady.
However with the David Axelrod attack on Gallup the polls have noticably
And dramatically turned Pro Obama as has the over sampling of Democrats.
I smell a Commie Rat by the name of Dave Axelgrease behind the
Shifts. For what it is worth- could the pollsters. Be doing A CYA??
Griff
Obama won Ohio by 4 points in ’08. If someone thinks he’s going to improve on that with chronic 8 percent unemployment, $4 gasoline and turmoil in the Middle East that someone is nuts.
I admit Obama doing better in Ohio than he did in 2008 sounds unlikely on the surface, but at the same time you also have to remember that I think Romney has yet to make a good case for himself.
I also think that the GOP is WAY underestimating how badly Romney’s 47% comments hurt him among working class and poor voters who might have voted McCain in 2008.
Given this, I actually don’t think it’s so implausible that Obama would carry Ohio by more in ’12 than ’08.
Romney is a WAY worse candidate than McCain and you have to remember that people aren’t just voting against Obama, they also have to vote FOR Romney.
except that Rassmussen does do alot of swing state party affiliation polling and its not showing those huge Dem affiliation advantages …
exactly..this bart guy is a troll
Polls that show a +10D spread, though, do not necessarily mean that there is stronger enthusiasm among democrats, though. You have to remember that this +10 is entirely relative to how many people self identify as Republican.
So a +10D spread does not necessarily mean that Dems have +10 enthusiasm, it just means that for whatever reason fewer people this election cycle are claiming that they are Republicans when asked.
It is perplexing why fewer people are claiming to be GOP, but enough polls are showing this same thing that I think it would be foolish to dismiss it outright. And Rass really can’t say anything about this phenomena either way, as unlike most other polling outfits, Rass DOES weight by party identification, so their polling would be entirely unequipped to detect such changes in self-identification.
That theory is only getting a lot of support amongst others that believe after 4 years of the worst POTUS of the last 100 years will produce stronger support than in 2008. It fails the most important test of all, the BS gut check test.
The ultimate failing of the methodology used by Quinnipiac, et al., is that they completely ignore the changes in electorate preferences that occurred in 2010 in their model. As the Church Lady would say, “How convenient.”
For those that know something about the laws of central tendency, sampling, and distributions of a population, the polls are no more than election season noise.
And reporters don’t get together before a press conference to ensure that they all ask questions certain to paint Romney in the poorest light possible do they?
“The theory is that AT THIS VERY MOMENT, the makeup of the electorate in swing states is overwhelmingly Democrat. That theory is getting a lot of support.”
Absurd. Two highly unlikely assumptions would have to be made for that to be true:
1) they aren’t swing states at all but deep blue
2) that the swing states that Obama carried by very slim margins in 2008 are actually turning even deeper blue this time around.
Both are demonstrably false. Compared to his 2008 blowout, the Big Zero is even losing support in Mass, NY and CA. The polls showing blue blowouts in swing states are clearly bogus.
sure there is are we wouldn’t be getting basically lied to everyday by the media. that dog won’t hunt.
The conclusion is that the makeup of the electorate that is willing to talk to pollsters is overwhelmingly Democrat. We don’t know about the others.
It could also mean that the Obama ad campaign is working. Supposedly the gender gap in Ohio is +25 for Obama, a truly huge number. There is a reason Sandra Fluke spoke at the Convention, after all.
“[W]hile [Dick Morris] is correct that the polls are showing strong Democratic advantages, they’re doing so because that’s how people are identifying themselves to pollsters. In fact, Stan Greenberg noted last Friday, Republicans lost five points in voter identification in a month. This is not bad poll sampling. It’s reality. And while it’s true that today’s numbers might overstate what will be the case on Nov. 6, the way things are going, they just might be understating them.”
Indeed, understating is also a possibility. It is quite possible that we are heading for another Obama landslide.
http://www.thedailybeast.com/articles/2012/09/27/michael-tomasky-on-the-gop-s-self-delusion-syndrome.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+thedailybeast%2Farticles+%28The+Daily+Beast+-+Latest+Articles%29
Obama landslide? Keep dreaming your little dream, CW.
This is just silly. I don’t profess to know who is going to win, but regardless of polling data, your position just defies common sense. Why?
1) Conservatives, libertarians, and tea partiers are more energized this time around. Look at 2010. Their deep disapproval of Obama is still there. There will be more of them showing up at the polls.
2) Democrats are less energized, because hope and change didn’t come to fruition. There will be less of them showing up at the polls.
3) I can’t believe there are many McCain voters who have switched to Obama voters. It just makes no sense to me. I look at Wisconsin, for example, and wonder who the Walker voters are that will be voting for Obama. Not many.
How do we know that Conservatives and Republicans are so energized? They were in 2010, but Romney was not on the ballot then. There are lots of people who can only stomach Romney because the absolutely can’t stand Obama. Romney hasn’t done all that much to really jazz up the base, either. Obama has. If the election is won entirely on motivating and organizing base turnout, Obama could win.
yeah let’s believe Stan Greenberg a Clinton lackey..nice try
First of all, the 4 point victory in 2008 was not a landslide. Secondly, polling companies can easily skew the data by randomly calling registered voters in a heavily democratic area where democrats outnumber republicans. It’s a common push poll tactic used by politicians.
Anyone that believes Obama will have greater support than what he had in 2008 have either been living outside the country the last 4 years or are owners of “Rules For Radicals.”
Just FYI. I lie to every survey call I get. I’m not alone.
Hey punk, I was wondering when you’d coming pissing nonsense here with your stupid leftist links.
I have 100 crates of broken binoculars to sell you.
Read the article ^^^^^. In Quinnipac’s case, they apparently didn’t use a turnout model (or they lied), but most polls do. The problem that’s unique to this year, is that they’re using 2008 actual turnout as the model for 2012. Really, really bad idea. 2010 congressional elections would be a better turnout model to use.
The bloom is off teh Won. The turnout models are going to be very different this time.
The only polls that use turnout models are the auto polls (Rasmussen, Gravis, etc). Coincidently, those are the ones that differ significantly from the consensus.
Sorry, Bart, that’s not true. Necessarily, everyone has a turnout model. The question is where it comes from.
They differ from consensus, but not from reality. Rasmussen has been the most accurate polling organization in the US since 2000.
BTW, what is it with hammer & sickle aficionados and their penchant for assuming “consensus?” You know, like with global warming? It must be the totalitarian impulse to control what everyone else thinks…
No they are not using a 2008 model. You are specifically wrong on the facts.
And you know this how? I’m serious — I’m having trouble finding an explicit statement from anyone how they’re modeling turnout, except for Quinnipiac, although I’ve seen several political pundits explain what Rasmussen’s model is. (Shortly PJM should have access to Rasmussen’s poll internals and I can look more closely.)
But the reported turnout models seem over and over to match, state by state, what 2008 turnout was like. So if they’re not using 2008 turnout models, they’re using models that don’t seem to be particularly distinguishable.
Charlie: Re a 2008 turnout model, recall that a significant number of discouraged Republican/conservative base members stayed home in 2008. But now this base is more enthusiastic than the Obama voters. So, the turnout pattern can be expected to be diferent this election.
JJB, I agree that would be my intuition as well. This is one of the reasons I don’t find D+8 in Ohio very plausible. But the point here is to look at the data and see what it tells us.
Dick Morris said that. I assume he knows these things.
It’s not that simple. Find a district that is heavily stacked with registered Democrats like an urban district. Call numbers in that district of registered voters. Guess which population you will likely be oversampling? You will likely be sampling the urban poor also mainly black democrat voters. They will be registered and they will say they will vote for Obama. How many of them will turn out to actually do it this year?
That’s how you wind up with a D+8 sample of registered Democrats.
It was my first lesson in media stats class: If you know how to design your poll or study, you can get any number you want. The numbers don’t lie, but the devil is in the details.
If the Democrats are the only ones who will answer your calls, you will get a skewed poll.
Except partisan ID polls show increasing Republican share, not increasing Democrat share.
If, as you say, the partisan ID’s have become more Republican over the last 4 years, then that would suggest that more moderate Democrats are leaving their partly (for the GOP or to be Independents), leaving the residual Democrats more enriched by liberals. Therefore the % Democrat sample the pollster chooses will be stacked with Obamanite lefties.
“I’ve been in politics long enough to know that the louder one side gets complaining about the polls, the more likely it is that this is the side that, in reality, actually is losing.” Erick Erickson RedState
http://www.redstate.com/2012/09/27/doom-gloom-and-polls/
“Here’s a simple truth: the polls and polling trends are by and large accurate, but — and pay attention to the but — the polls are overestimating Democratic turnout.”
http://www.redstate.com/2012/09/27/doom-gloom-and-polls/
You forgot this part: “The reality is that Mitt Romney is behind.”
Like I said, you guys suffer from mass Dissociative Disorder. You guys are being prepped by the establishment for a soft landing. They don’t want you to give up hope because that impacts down ticket races if people stay home. It is a bit ironic after all the chest pounding over Romney’s fundraising prowess, that we now find him scrounging for donations. Savvy investors (and Romney should know this better than anyone) don’t throw their money at a lost cause.
Hey, you are the one who reeks of desperation. “Oh, hey, this is Cynical Wonder here, and uh, naw, look, it’s Obama that’s gonna win in a landslide. Yeah, listen all you decent citizens out there, uh, just don’t vote, ok? Obama is really not losing this thing, I’m serious!”
Romney is scrounging for donations?
But…we’ve seen a raft of articles of Obama asking for five bucks, asking for donations in lieu of wedding presents, and so on.
The reality is that you’re a 3rd rate leftist troll with no creativity.
Erik Erickson is looking for a nice paying MSM gig, and has been for a long time now. He’s a sad, pathetic shadow of his former self.
You’re wrong. Eric Erickson was a limelight seeking lightweight from the very beginning. I was in on the ground floor, almost a charter member at Redstate. The first thing I noticed about EE was his giddy posts promoting his desire to get a CNN gig. Then he started throwing feelers out in every direction trying to get a spot substituting for Rush. As if. Like most pol blogs run from the top down, it went into decline the day the personality cult thing reared its head. I see signs of that here with a few featured contributors going all prima donna on us.
Eric’s involvement in politics was that he won a town council election in a small hick town … you likely have as much “experience in politics” as he does …
Jeff: Yiu have described the pretender in the White House perfectly.
JJB
That doesn’t necessarily follow. Yes, it’s a possibility that the real turnout will fit those models, but there’s also a lot of examples where the turnout models are wrong. Not every pollster uses the Quinnipiac model — others are using turnout models they arrived at other ways.
In any case, that’s kind of my point: IF the turnout really is D+8 in Ohio THEN you can believe the Q. poll. If you think it’ll be R+1 (which is fitting a lot of reports of actual registration) then it will be something else.
That is what raises my suspicion , all depots are that in Ohio R Regis
Are up 400k+ D are down 200k +. So my ? Is how can the 08 model be valid. ??
Griff
Bart, please tell us just how much you know about statistics? Math?
There’s a guy named Bart over at Curry’s blog who sounds a lot like this guy. Utterly impervious to facts. Not saying they’re the same, but they sure sound similar.
What time are the pollsters calling? If they’re calling during the day, unless it’s a retired senior, a person at home during the day is more likely to be a Democrat. They need to call between 6pm to 9pm or anytime on weekends.
I don’t agree that this is the only “logical” conclusion, since you are not allowing for the possibility that, since the pollsters are sampling only the people who don’t hang up on them (like me) and the people who say what they think the pollster wants them to say (a big problem with social research of all kinds) and the people who like to screw with pollsters (like my husband) the pool of people who answer a pollster’s questions is really from the beginning not representative. In addition to all the other caveats mentioned.
Your logic also depends on all those who previously answered being excluded. I applaud your attempt to apply the scientific method, repeated experiment, to this problem, but there are too many variables unaccounted for to make its use reliable. Think about pharmaceutical testing, which proved at one point to be dangerously unreliable because the test population was not like enough to the population for whom a drug was intended. That’s the point of this article, the methodology used is too simple.
Except that Rasmussen independently polls voter identification both nationally and in the swing states. The national sample size is 15,000 voters every month. They publish that national data and for a fee will give you the state by state data. Rasmussen uses a three week moving average of those results to control for party identification.
So there is a separate measure of what the electorate looks like. Nationally in August it was 37.6% – R, 33.3% – D, 29.1% – independent/unaffiliated. The assumption of +8 Democrat advantage diverges from this sample by 11.3 points.
This is the best article I have ever read on this topic. Great explanation. I know this will make the ‘stack of stuff’ today!
Agreed, this is a great article, and original in its conception. But why is Charlie Martin’s method not being covered on Fox News Channel, supposedly the foil to MSM and its propaganda? Because FNC only offers some voice to conservatives and independent thinkers: it is “moderate” in its conception. See my article on their own version of political correctness here: http://clarespark.com/2012/09/25/thought-police-on-fox/.
Thanks to both of you! Clare, my suspicion is that there is a general tendency of newspeople to run screaming from the room when presented with math.
I share that suspicion.
Absolutely right, Charlie. They have no concept of math, markets or economics. They think the money that finds its way into their paychecks is manna from heaven rather than the result of how the market works (or doesn’t) for them. I remember sitting in a big-city newsroom with my fellow highly paid editors one day some 30 years ago … there was a lot of mirth at the expense of the grubby ad salesmen who worked out of the same palace of truth. No one seemed to glom on to the idea that the grubbies were paying the editors’ fat salaries and that once the ads dried up, so would their livelihoods. I’d been one of the grubbies at one point, so the whole discussion was a bit surreal for me. And although that was 30 years ago, you can bet your bottom dollar (or doubloon, if you’re smart enough to have begun the move out of Uncle Sam’s debauched paper) that the situation is the same now. Most news writers and editors haven’t the slightest clue about how the real world of meeting a payroll and bidding on jobs works. That’s a major reason why they fell for miracle-worker O so heavily and still support him. Unfortunately, newsies aren’t alone; academics, students, government workers, lawyers, nonprofit folks and many others share the same delusions. Hopefully not past 50% of the electorate this time, though.
Gallup does NOT use a turnout model based on party identification. As they explain on their website, they estimate the likelihood of a voter actually voting by asking him about how often he voted in the past, how concerned he is about this election, and other such questions that don’t relate to whether he considers himself a Dem or a Repub.
If we’re supposed to believe that Romney is tied or winning despite even the Gallup poll showing otherwise, then we have to believe there’s a real conspiracy out there by all the pollsters to defeat Romney. Even Rasmussen admitted on Fox News that Obama is ahead in Ohio and Florida and in most swing states.
I don’t go for conspiracy theories.
If the election were held today, Romney would lose. I’m not happy about that. Obama is and will be a disaster as President. But I would rather know about disasters in time to deal with them.
Rasmussen also said that Obama’s lead was very small which likely puts any lead he has within the margin of error.
Well, let’s see. To make your case, all you have to assume is that:
- corporate Gallup is entirely honest, and they haven’t been influenced by Axlerod’s friendly “call” to make sure that they are being “fair”
- Gallup’s EMPLOYEES are entirely honest, and haven’t been, shall we say, influenced by, shall we say, a concerted multi-year effort to get as many leftists as possible on staff.
- Gallup’s turnout assumptions reflect this election, even though the natural tendency of even totally objective pollsters would be to lean heavily on prior presidential election.
- prior presidential elections are like this one
- they get a true random sample
- all voters are equally easy or difficult to contact
- people respond honestly to poll questions
- republicans and independents are just as willing as feverish leftists to respond to polls
- Gallup’s telepphone lists haven’t been manipulated
- there is no hidden bias in their questions
- the answers to questions are never influenced by the questioner
- people don’t hedge their answers by, as just one example, being reluctant to state that they disapprove of a minority president
- respondents would come to the same conclustion in a voting booth as they do in a quick conversation on the telephone
- the almost univeral situation of polls dramatically shifting in the very last one taken before an election (i.e., the one the pollsters’ will use to brag about their accuracy) are actually a true reflection of last-minute shifts in public opinion
etc., etc., etc.
Polls are manipulation devices. Only a fool would think otherwise with all the information we have about them. Aggregating them is like believing a con game because dozens of con artists say the same thing.
Why do people ignore the sponsors of the publicly available polls?
The press sponsors most of the polls. The press lies, and we all know which direction their lies slant. We know they lie for Democrats even when it damages their credibility — remember Dan Rather?
The press pays the polling companies for particular results.
Not only are the polls paid for by media companies, the polling companies are often associated with professional political shops who run campaigns mostly for Democrats. Carville is an excellent example.
Dear Sinz54,
I was around in 1980, when Carter Beat Reagan.
I also remember how Mondale trashed Reagan and made him a One term President.
Of course those situations never happened, but folks who beleive as you do were convinced in September and October that it was over and Carter was going to win a second term.
If the election were held today, Obama would lose badly. Period.
If you look back just a couple of days, you’ll see Gallup was tied. The thing I don’t understand about Gallup right now is the sudden change.
Not everyboday is a politics junkie and people near and far are seeing the 47% video. Simple.
Gallup has been quite bouncy all year long. More so than Rasmussen.
I THINK that Rasmussen uses a rolling average and Gallup doesn’t. That would explain the difference.
The Pew Research group has shown that currently the response rate for most telephone surveys is 9%. I can hardly believe that sampling 9% in an essentially non-random way (some polls don’t call cell phones, a LOT of people are on no-call lists or are unlisted altogether and clearly 91% of the people they do contact refuse to cooperate) is going to give you anything even remotely verging on accuracy. To get 1000 respondents at a 9% response rate, they would have to contact more than 10,000 people. People’s reluctance to participate (which is actually quite recent and probably affects one political persuasion more than another) seriously compromises the validity of these polls. This is beyond the pollsters ability to correct, but they aren’t about to abandon their business model.
Jay Cost pointed out a couple of years ago that Gallup polls tend to be bouncy.
since Obama isn’t close to 50% in any national poll TODAY he could very easily not win an election held today … and polls are not elections … look at the 2010 polls vs the results … not even close those polls …
If it turns out to be the case that the US population votes for Obama again, this is a disaster that you cannot really deal with. It would be indicative of a long-term decline of the US that the majority would be choosing national bankruptcy because they have been swayed by spurious arguments over unimportant subjects. The debt of the US will grow so large that there will be no way out. Romney and the Republicans may not be able to get us out of that debt, but at least there is hope. With Obama there is no hope.
I hope and pray that Romney wins. However when discussing polls, in all the elections that I have followed in the US and the UK since the early 70′s, the party behind always claims that the polls are innacurrate and that their internal numbers are different. Usually the polls are correct, especially when predicting trend.
Quinnipiac used the same methodology in 2010 and were I believe amongst the most accurate of the polling groups.
My only hope is that there is still some time to go and, anecdotaly at least, polls seem to have become less accurate over the last few election cycles.
Upon reviewing Florida 2010 I found the following from Quinnipac
October 21, 2009 Crist 50% Rubio 25%
October 28, 2010 Crist 35% Rubio 42% Meeks 15%
Election 2010 Crist 29.7% Rubio 48.9% Meeks 20.2%
October 28, 2010 Sink 45% Scott 41%
Election 2010 Sink 47.7% Scott 48.9%
Oh sure, they aren’t even close a week before the election. Wishful thinking.
How was Quinnipiac so off? Did they miss turnout? That would answer the question once and for all about how miscalculating turnout misstates polls. Are the breakdowns of the 2010 Senate polls available?
Not sure where you got this data. The Quinnipiac poll had Rubio up by 11% on 10/5/10.
It is almost impossible to find;
Step 1 Google…Quinnipac Florida 2010 election polls
It is right there before your very eyes
Allan
BS. In my state during the primaries public polling was showing the guy who got the most votes as coming in at the best 3rd place. Then on election day, that guy came in 1st, but didn’t get 50.1% so he went into a runoff with the 2nd place guy. All the way up until a few days before the election the public polls were showing him to be losing by 5 or 6 points. His campaign started complaining about the polls and saying their internal polling was showing him up by double digits. Then a few days before the election a public poll came out showing him up by single digits and he won by double digits. So all the public polling was way off showing the loser winning by 5 or 6 points when he lost by 14 points and even the lone public poll who showed the winner winning was off by 9 points.
How were they off so much? My opinion is the local newspapers and local tv stations had some bias as one pollster admitted after the fact they did 90% of their polling in the county the loser lived in?
Actually though I think it worked against them at least in my household, I had forgotten about the run-off until a week before the election and I read in the newspaper about all the polls showing the candidate I wanted to win and the candidate I had already voted for losing in the polls and his campaign was complaining and used your same logic that when campaigns are criticizing the polls they’re losing. So it made me say crap we have to go vote on x day.
BTW….if by some miracle Obama was to win re-election that speaks volumes about this country and none of it good.
Sad. A leaderless army.
Sorry — a pointless comment.
There is no news. It’ all manipulation. Leftist realized decades ago that they manipulate public opinion if they can control the media.
But people caught on long ago that the media lies about the news in all sorts of ways, from refusing to report events that run counter to their bias, to flat-out lieing about events.
When they realized the public was onto that, they moved on to polls, which they present as scientific, but might as well be done with spreadsheets in Communist Headquarters.
Now they are also using Hollywood to manipulate opinion. And then there are their tricks of controlling the Universities and public schools. And don’t forget buying “scientists”, their paid shills who they shamelessly trot out to “prove” whatever they want to prove. This could be the latest con, but I doubt it. You can rest assured that they just about have religion cornered, and are working on capturing every single decision-influencing body in the world…including the Tea Party.
The people who devote their lives to ruling over you are always one step of the curve, for the simple reason that you have a live to lead, and their only life is controlling you. They have new methods that we won’t realize for years.
But Rush says only liberals are victims. You are not being a loyal conservative. You better appologize.
I swear I’m going to get them to add a “like” button and a “you’re an idiot” button.
I like that idea, since then I could label the idiots….
THANKS FOR THIS ARTICLE.
Please add “Like” and “you’re an idiot” buttons. They are very interesting over at Legal Insurrection.
And thanks again for this article.
ah Ed the difference is that liberals THINK they are victims while Conservatives are actually victimized … I thought you guys where the reality based people ?
just reality challenged I guess …
I would note that back in 1980 Gallup had Carter 43, Reagan 35.
This is not to say that the same thing is happening now, but it certainly tells us that polling is not the end all be all.
Get out and vote!
I keep thinking of the British general election of 1992. Labour held a triumphalist rally in Sheffield a week before polling which is now generally thought of as the moment when they lost: polls, up till then showing them with an apparently unassailable lead (hence the triumphalism) started narrowing from that point. But the result – a narrow Conservative victory – still came as a surprise to everyone, even in the party itself. Labour was still ahead on some exit polls. Confidence in our polling companies took a serious hit that year, and in fact led to some of them altering their methodology.
Of course, it can go the other way: because of that experience five years before, the Tories simply didn’t believe their awful polling figures during the ’97 campaign, when they suffered their worst result in decades. Which makes me somewhat wary of articles like this.
But the bottom line is absolutely true: it’s never over till it’s over.
Already did. Absentee. One more for Romney and Ryan. My money says they pull it out. I keep asking the Cynical Won to put some money up if he’s so sure Downgrade is going to pull it off. I never get a response.
I’d say the boy’s all hat and no cattle. His biggest concern is probably wondering whether Axelrod’s check for CW’s trolling will clear or bounce. Odds are good that if Downgrade loses, that check bounces like an overinflated basketball.
McCain’s campaign had more signs of implosion than Romney’s shows. They seem to be confident that they’re doing as well as they can. If there were major issues, I’d expect to see much more flailing around and staff shake-ups, but that isn’t the case. I do see a full-on concentration on Ohio which is warranted, but both campaigns are there. If Obama really had a nearly 10 pt. lead in Ohio as the polls seem to show, would he be so concentrated there? I think he’d be a little spread out trying to shift a few other close states. This alone suggests that Ohio is much closer than the media is portraying to the rest of us.
Good point. Plus, Obama was campaigning in Wisconsin recently. If his base is so energized, why the need to show up in a Democrat stronghold like that?
Another indication — the Obama campaign admitted early on they’ve “abandoned” white, working males. They’re not trying to bring in new voters; they’re trying to keep their base engaged to avoid a blow-out.
Obama is concentrating on the Women’s vote, for both persuasion and turnout.
Wait, I thought it was spelled ‘Gallop.’
Oh, don’t start that again
Great article.
Some comments.
(1) #1 above. There is something called the Law of Conditional Probablity. Just because we flip a coin and it comes up heads 10 times in a row, this does not mean that the odds of head or tails on the next flip is anything but 50%. That is really hard for most laypeople to get.
(2) The guys on both sides have the best pollsters in the business. When over $500 million is at stake, you better have the best guys. The rest? They do the public polling.
(3) You can guarantee that the numbers are worked out with every state, every county, and every demographic with different numbers based on different levels of turnout, very similar to the table presented above. There are likely very complicated equations estimating likelihoods of victory or success in every state and in every county. These have been developed by both sides, and share with the various money groups. Given the number of people who know how to do this, I’d say that the skills are probably even. However, I’d say R has a philosophical advantage since he’s used to working with metrics and numbers in his prior business, where O as a person is not. O probably “goes with his gut” a bit more than R when confronted with stats.
(4) These numbers are used to estimate the impact of message, ads, and visits in as precise a way as they can. The campaigns also are likely getting calls on an hourly basis from senate and congressional campaigns, with their own data and metrics, who konw that a visit will push one demographic up or down.
(5) The data gathered in March, April, and May has been used to develop the strategic arcs of both campaigns. Both campaigns are sticking to those arcs.
(6) Complain about R as much as you want, but O’s arc was to disqualify Romney before the debates, without talking about where O was going to take the country.
(7) The good news for R is that it did not appear to work. If I was Axelrod, I would be approaching the debates with a sense of dread and looking for something big to change the topic.
(8) The lay press is starting to question the turnout models, since they don’t all want to look like fools in this 24/7 news cycle.
(9) This means that R and O are basically tied, with the debates shifting opinion enough in one direction or the other to win it. Both campaigns know it, and just look to where they are going to know the state of the race.
(10) I think the whole Libya thing, by the way, is a fundamental that has not yet peaked in the public consciousness, and I suspect it will right around the last debate, and it could be the final straw for O.
(1) taking a poll is not flipping a coin. we have an infinite number of outcomes along a spectrum. not just heads or tails..
The point was not how many outcomes one poll has, but rather that the outcome of any one poll does not influence the outcome of another poll. The probabilities are separate.
If a coin comes up heads 10 times in a row, my assumption is that the game is fixed or the coin is biased. I’m betting the 11th time will be heads also.
The Iterated Gambler’s Fallacy:
The assumption that the next charge of buckshot will have the same pattern
and center as the last. What if both R&D know that this is the last round ?
Not a math major, I’m guessing.
GUYS!! Stop looking at all these different polls! The ONLY polling company that consistently gets it right is Rasmussen. They are pretty much spot on. Rasmussen, as of yesterday, had it 46 for O and 46 for R on a national basis. I don’t know what the Rasmussen breakdown was for the swing states, but common sense tells us a few things:
1. There’s NO WAY Obama has a 10 point lead in Ohio. Even in 2008, after 8 years of Bush and a population that absolutely turned out in droves for Obama, he didn’t take Ohio by 10 points.
2. This economy sucks and it’s the republicans who are fired up and will turn out.
This is going to be close, not a blow out as the polls suggest.
That’s actually not true. Rassmusen has only been around since 2000, and they’ve been mixed. They were way off in 2000, more accurate in 2004 and 2008, but they way overestimated results for the GOP in 2010.
So it’s pretty mixed right?
But the problem is that while they were more accurate in 2004 and 2008, their polling was also more in line with most other outfits those years. However, just like they have so far in 2012, their polling in 2010 tended to be off from other polling outfits in favor of the GOP.
Which, obviously, would make one suspect that they might be overestimating in favor of the GOP again this year, given how, just like in 2010, they are once again consistently the outlier compared to other polls.
538 did a break down of their accuracy over past campaigns, although I’m not sure if it’s worth bothering to post given how paranoid some are about the “liberal media” on here:
http://fivethirtyeight.blogs.nytimes.com/2010/11/04/rasmussen-polls-were-biased-and-inaccurate-quinnipiac-surveyusa-performed-strongly/
What tripped up Rasmussen in 2010 was that no one had ever seen an election where the polling showed such a large Republican advantage and its impact was over estimated in a number of traditional Democrat strongholds. Republican’s traditionally out perform their self identification by 2%-3%. This over performance appeared to hold up until their advantage exceeded middle single digits. Then it tapered off and in the areas where it was in double digits they slightly underperformed their raw advantage.
2010 was unlike any election in modern history. The assumption that Republicans would perform to their unprecedented advantage as they had in prior elections with much lower numbers proved false.
However, if Obama is at +3 or below the odds are good that he loses Ohio.
There are other forms of bias. Rasmussen uses robotic calls to eliminate one of the hardest to control, that of the subtle inflections in the pollster’s voice. Take the same script, the same screens and an equally random sample and you’ll get different results if the people making the calls are heavily vested in the results. It’s human nature. A certain percentage of people will tell strangers what they seem to want to hear. This is one reason I am always a little wary of results from polls affiliated with academic institutions. The odds are the people making the calls are grad students in social sciences, who tend to be overwhelmingly leftist.
I also like to see all the data, including the exact order and wording of the questions. Adjectives and modifying clauses matter a great deal. Quinnipiac displayed a classic bit if polling bias last year in the debt ceiling debate when it asked if the ceiling should be raised so the federal government could continue to pay its bills. How much different would the results have been if they had instead asked should the ceiling be raised so the federal government can continue to spend $1 for every 58 cents of revenue?
I think there’s another advantage to robot polls, and that’s being able to “admit” that you want to vote for the guy that the MSM hates. People want approval, even if it’s from some random media pollster- and we all know who the media approves of.
#8 aharris: “McCain’s campaign has more signs of implosion than Romney shows…….” Maybe. But these two guys keep sticking large feet into even bigger mouths. Romney’s latest is questioning why you can’t roll down the windows in planes so people can get out (and the first nut case who tries that at 35,000′………). Ryan’s latest video “caper” today is claiming that rape is a form of conception (it is. but how do women feel about that?).
If this ticket loses; manages to snatch defeat from the jaws of victory; it will be in part due to the ineptitude of the candidates.
I disagree.
This election will turn on how many people in the middle really feel bad about their economic situation, and how much they trust O and the Dems to fix it.
Partisan turnout will help the Repubs this cycle as well.
If R does not win, it will be due to (1) a feeling among independent voters that they are doing ok economically or will soon do better; and/or (2) less turnout among R voters than currently predicted.
Everything else is just noise and should be treated as such. Nobody could have done better, and if they could, it would just be affecting issues on the margin.
And here we have a shining example of why I started keeping the Romney Rumors sheep list.
That was a joke that LA Times misreported.
http://pjmedia.com/tatler/2012/09/25/romney-rumor-romney-thinks-airplane-windows-should-open/
Now, repeat after me: Baaaaa!
And let’s not forget Quayle’s joke, said in a joking manner, that every reporter on the plane knew was a joke- That he was going to Latin America and didn’t speak a word of Latin. I know people who to this day believe he was serious because one, yes ONE reporter reported it as NEWS in his magazine, and it was picked up and mindlessly repeated by the rest of the media, that a senator with a law degree thought that the people in Latin America spoke Latin.
And of course, when Obama refereed to Austrian’s speaking Austian, he simply mispoke. I know not a single Democrat of my acquaintance who knows he said that- for it it was never in any of the local newspapers or broadcast in any nainstream media outlet. Therefore, it didn’t happen- though it did.
Republicans aren’t allowed to make jokes- because they are misreported. And when they don’t make jokes- their not serious and reflective- they’re humourless.
And most people also don’t know that “potatoe” was on the flash card handed to him.
MSM monkey business. It’s been around a lot longer than you think.
” Romney’s latest is questioning why you can’t roll down the windows in planes so people can get out ”
And if you’d bother to listen to the audio, instead of just endlessly repeating a leftie talking point that you picked up someplace, you would have heard the joking tone in Romney’s voice and the laughter of people around him.
See: http://nymag.com/daily/intel/2012/09/mitt-romney-joke-joking-airplane-windows.html
Get some new material.
If rape isn’t another way to conceive, the the whole question about abortions in case of rape is pretty well moot, isn’t it?
Indeed.
I’m not upset at all that these polls are skewed. I think it’s a great thing. I don’t believe they’ll actually turn off republican voters – perhaps only make them try harder. They WILL promote overconfidence in ordinary dems. AND they’ll also have two other benefits. If Romney were shown to be ahead, the dems might be going insane by now with hate. This keeps it tamped down to a manageable level. And after Romney wins, it will make it so much more delicious watching the lib heads exploding in rage and hate.
I agree and was thinking the same thing.
I have heard so much on talk radio about the polls being skewed in favor of Obama in order to suppress Republican turnout. I believe it will be just the opposite. The anti-Romney 47% will think that Obama has it in-the-bag so why make the effort to go to the polls. Historically, Republicans have a higher turnout ratio than Democrats, not counting the dead in Chicago and the felons in Minnesota (i.e. repub-voted/repub-registered > dem-voted/dem-registered).
The turnout chart in the article is key. If the D/R turnout ratio is anything close to even, Romney will win.
So the pollsters themselves are partisan and so are deliberately lying, for their candidates’ advantage? Hmmm? What should be done with lying pollsters….and with the newspapers that print the lies? Sounds like a job for the Ministry of Truth.
Nobody is lying.
Polls are data. Take them at face value.
I guarantee you neither the R or O campaigns are are taking the public polls too seriously, and have adjusted their own polling for expected turnout.
However, if O wanted to use these polls to distribute resources over the next 5-6 weeks, I’d love it.
Exactly.
Many, if not most, are lying. The notable exceptions are the campaigns’ internal polling, and only the polls that they don’t disclose to the public.
The evidence is ovewhelming and goes back decades.
You must believe that humans are totally responsible for global warming as well. After all, why would scientists lie? Everybody is good at heart and totally honest, particularly professional people like scientists and pollsters, right?
No, what we need to do is stop giving a rat’s patoot about polls.
The Republic did quite well without polling, for about half its history.
No, that’s exactly what I didn’t say. I don’t doubt Quinnipiac is fully happy with their polling method, and doesn’t think they are skewing anything. On external examination though, their method doesn’t seem to quite match up with what we’d expect. So, this gives a method where you can, by thinking for yourself, make your own estimates.
Given the scrutiny over their methods and assumptions, I wouldn’t be at all surprised to see an adjustment in their next round of polls.
So let’s see.
Are the media honest but misguided brokers as well? I mean, really, they’re trained professionals who depend on their credibility for their livlihoods. They MUST be objective. Besides, they tell us all the time that they are. Why shouldn’t we believe them.
So, despite the overwhelming evidence to the contrary, much of which you cite yourself, you persist in believing that pollsters are innocent lambs just doing what feels scientific and good.
Your naivite is charming.
And Romney is charming as well when he persists in his self-destructive assertion that obama is just a nice guy in over his head, kind of like your local insurance agent who got lucky and won a few elections.
Look, everyone wants to know what is going to happen. It’s human nature.
If it wasn’t polling, it would be looking at coffee grounds or rabbit entrails.
This is one outcome we all can affect, however–by going out to vote and turning this unserious, incapable clown of an executive out on his ass.
Now do it, and stop complaining about polls.
The Democrats are simply using polls as advertisements. I don’t really trust polls and the only thing I can say is what I’ve said throughout this whole process. I don’t see why more people would vote for Obama now when they were willing to bury him in 2010 during the midterm elections. If anything, the situation is worse now than it was in 2010, with the economy in the dumps, unemployment still painfully high, Obamacare is still very unpopular, and with the looming fiscal cliff about to destroy this nation on January 1. Why in God’s name would anybody vote for Obama now when nothing much has changed in two years? Even if you don’t agree with Romney, any clear-thinking individual would want to change just to see if it would make any difference with the economy. So, I’m probably in the minority out there, but I still think Romney is going to win big. I could certainly be wrong, but I see people just as angry and just as motivated as they were in 2010, and that’s not good for Obama.
“any clear thinking individual would want to change just to see if it would make any difference with the economy…….” If so, then those persons should vote for Gary Johnson, the Libertarian candidate.
As for Romney making a difference in the economy, let’s be realistic here. He will have to pay off those who put tens of millions into the GOP Super PACs. In the face of annual trillion dollar deficits in D.C. Romney wants to pump a couple hundred billion more into the Pentagon. How is that any different from Obama pumping a couple hundred billion more into Medicaid via ObamaCare, a.k.a. RomneyCare?
Johnson is the only major candidate who is real about the need to cut spending by making across the board cuts in ALL federal programs.
If Romney added that plank to his campaign platform, it would be a gallows;
After the economic crash, he can claim that shared sacrifice is necessary.
Only unreasonable or politically naive people will vote for Gary Johnson. And I say that as a registered libertarian. Think of the radical leftist purists who voted for Ralph Nader in 2000, in Florida. They were a small, but critical loss of votes that would have made Al Gore the President instead of George Bush. If you hate CO2 taxes, you can thank the Ralph Nader voters. Think of the “reform” party idiots who voted for Ross Perot in 1992, giving Bill Clinton a plurality win.
Like Ralph Nader and Ross Perot, Gary Johnson will not win a single Electoral vote. He has a ZERO chance of winning. Ergo, a vote for Gary Johnson is a vote for Barack Obama. Don’t be among the Libertarian purist idiots who split the “get government off my back” vote by voting for Johnson. Vote Republican, and work within to make the Republican party more libertarian. It’s not pure, I understand there’s a lot for you purists NOT to like about voting Republican. But if your vote helps Obama win, do you really believe you’re better off in claiming a purist position while being governed by a radical lefty in the White House? Eric Holder? Islamist appeasement? Takeovers and overregulation of health care, insurance, finance and energy? Tax and Borrow and Spend to expand the social welfare state? Is that what you really want?
“work within…..to make the Republican party more libertarian……” In order to do that, one has to take on the anti-abortion- of-any-type crowd; or as a sadly deceased earlier this year Colorado libertarian/common sense conservative writer & columnist called them; the “zygote zealots.” Then there’s the anti-contraception, anti-gay, anti-separation of church & state components (also known as the religious right, which has a solid control of the grassroots of the party).
If Romney loses, I’ll predict upheaval coming in the Republican Party; between the fiscal conservatives and the self-proclaimed social conservatives.
being a libertarian doesn’t make you pro-abortion. If you are a libertarian and believe life beings at conception, two NOT incompatible thoughts, then you can be libertarian and believe abortion is or should be a crime.
When the Supreme Court decided that life begins, not at conception, and not at birth, but at some magical point in between, they might as well have also decided how many angels can dance on the head of a pin. And ensured that abortion would for a very long time be a divisive issue in American politics.
The pro-abortion lobby, along with the nanny-staters, who are one and the same for the most part, have brought forth upon this a continent a nation where teenagers cannot self dose themselves with aspirin in high school,without being labeled as drug abusers, but at any age, if pregnant, they can undergo major surgery to remove a foetus from their body, WITHOUT PARENTAL PERMISSION.
And leads to laws that I believe are mutaally incompatible. A woman, at eight months, can abort her child. A man, or another woman, who klls that woman and unborn child can be charged with TWO counts of murder. The foetus is not Schrödingers cat; it either is or is not a person. It cannot be both a person and lump of tissue, dependent on who makes the decision to kill it.
Like I said, idiot or politically naive. Did people with similar beliefs to yours say similar things about John McCain? Yes. Did their predictions come true? No. Insanity is doing the same thing over and over and expecting a different result. Those people who refused to vote for John McCain because they thought him too soft on “Conservative” or “Libertarian” or other values, helped us get Obamacare, Elena Kagan on the Supreme Court for life, Eric Holder as our AG, Fast and Furious, the American Apology tour, kowtowing to Islamists, the Dodd-Frank bill, the shut-off of deep water drilling, stagnant economy, malaise, more food stamps, more unemployment benefits, no budget for 4 years, more redistribution, more crony capitalism, and I could go on and on. Wake up. Stop being a whiney baby, and vote for the rightist coalition, even though it’s not perfect. It better fits what you want, if you’re a libertarian. Stop worrying about abortion as the prime driver of the Republican party. Think up a good compromise for the religious right and the libertarians.
well, because defense spending has a clear positive multiplier effect on the economy while Medicare spending has a negative effect and is a sink hole …
If you want to see the accuracy of polls, EVEN supposed more accurate exit polls you have to look no further than this summers Wisconsin’s recall elections. It was supposedly SOOOOO tight that no predictions could be made. Then there were the exit polls which were STILL claimed to be too close to call when while they were saying that, Walker was up a steady 15% with 30 percent vote count, and ended up winning by I think 7-8%. You don’t think the public is being played by the MSM? LOL…well then go back to sleep!
Okay, now ask yourself: how would the polls go that far off? Forget the overt conspiracy theory.
Charlie, thanks for your article.
I’m waiting for the next oversampled Democrat poll to be something like 65% Democrat, 28% Republican and 7% Independent. That ought to be enough to give Obama a 5% lead in the poll. Margin of error? Who cares? “We don’t need no stinkin’ margin of error?”
Martin’s article is a good one, especially valuable to those with little or no grounding in statistics, but I find another question more intriguing:
What effect will the many polls that predict a runaway Obama victory have on the turnout on November 6?
If that effect is foreseeably depressive for the Romney/Ryan ticket, as many pundits have claimed, what can we do about it between now and then? Is there anything that could counteract that “you’ve already lost, so give up” jackhammering?
Keep pointing out the potential flaws in the polls?
Hah! That’s good. And true.
You Charlie Martin chucklehead you!
I am, at the next weekly meeting of Chuckleheads Anonymous, nominating you and your huge head for Lord High Chucklehead, our highest honor!
Remember. It takes a great big, gigantic chucklehead to contain all that genius!
ALL HAIL! Lord High Chucklehead Charlie Martin! Genius extraordinaire!
The Pour Le Merit pales by comparison.
Keep up the good work. As if you had a choice!
…and we nominate you prick of the week.
So instead of being reactionary and juvenile, how about addressing Mr. Martin’s assertion?
E-mail each news broadcast whenever they parrot a poll without telling you what the Dem./Rep./Ind. assumed breakdown/fudge factor/manipulation is, and how that stacks up against independent data on the real breakdown.
Also, don’t believe Charlie Martin here or all those other talking heads on TV who hasten to tell you there is no leftist conspiracy, that the pollsters are just being honestly arbitrary in their modelling. Personally, I would at least call them all incompetent, from what we have all learned in the last few days.
Go back to the first skewed polls post and look up Heinlein’s Razor.
This is a very thoughtful thread, but I would add one point.
Essentially everyone in the media keeps repeating the refrain that Romney has made this or that gaffe and that his campaign is floundering. Now Romney did make gaffes during the primaries, but with the exception of the politically harmless comment in London, none of the other alleged gaffes were gaffes in the slightest. These were the comments about Palestine, the 47 percent and Cairo.
What makes the polling so suspect is that the Dems are doing things that never won elections in the past. Booing God and Jerusalem? Jews don’t kill Americans. Arabs and Muslims, however, do. My suspicion is that Obama felt he had to go after Qaddafi, because internal polls showed a visceral hatred of him, since he had killed Americans in Berlin and at Lockerbie. I also don’t see how gay marriage and throwing safe, legal and rare out the window is supposed to win an election. Sorry, this does not compute.
So back to first point. You can say the polls are not a conspiracy. But I don’t see how you can say that about the media coverage. And so I would ask: why should we assume that political pollsters are more honorable than political journalists?
Nor did I say anything about the media.
But remember, in any statistical study, the least likely outcome is to surprise the principal investigator.
[Insert appropriate Feynman quote here.]
It’s very simple: Obama WILL NOT receive more votes in 2012 than in 2008. Quite the opposite, in fact. He has lost many elements of that high-water mark, including the white guilt vote, the “historic election” vote, a substantial number of Independents, portions of the Jewish, youth, and even black (many black ministers are telling their congregations not to vote for Obama, due to his stance on gay marriage; also, Obama’s base is generally less-energized, and confident of his win, overall) votes.
Contrast this with an absolutely galvanized Conservative/Libertarian base, heartsore with all the abuses and violations of the Obama regime. Unlike 2008, or even 2010, 2012 is the one, great opportunity to strike back, and there are going to be many MORE voting GOP this time, no matter what folks think of Romney personally.
Think 2010 on overdrive, folks. =^[.]^=
Very interesting article – even for an English major like myself who did poorly in math. Normally I try to stay away from anecdotal evidence but no one has brought up the question of WHO is more inclined to answer the phone and take the time to answer the questions truthfully. I live in a blue state and as a volunteer have called tens of thousands (and spoken to thousands) of registered voters for Republican US congressional, US senatorial, & gubernatorial candidates. Sometimes I know the party affiliation before I call – and because I am a former Dem(now registered independent) – I often am given lists of Dems and Independents. {Republicans make up only about 25% of registered voters in my state but we have elected Republican national and gubernatorial candidates.) I have found of the three groups – the Democrats seem to be the most inclined to answer the phone and seem to have the most time to “chat”. The registered Republicans tend to be more brusque and be in a hurry to get off the phone. [Perhaps because their time is more valuable? ;0)]The Republicans who do answer seem by far to be more suspicious of my intentions and more likely to say they do not take polls. [Or perhaps give a "false" response?] Also I have spoken to many union people who tell me they do not feel comfortable telling me who they are voting for – but even when I state I am calling for a Republican candidate they are very polite and curious about how I think the election is going. I suspect that if they were planning to vote Republican (and in a recent election a decent percentage did) – they would never tell me over the phone. Now granted these pollsters call enough numbers to get what they think is an appropiate sample of x number of each affiliated group. But even if they are oversampling one group – since they are not calling from voting lists – who is to say a small percentage of those people are not answering honestly? And is this enough to skew the already questionable polls further?
Good points. The point about not trusting and not taking time with pollsters certainly applies to me.
I have carefully followed all this controversy. I’ve got to say I distrust all the claims against the “oversampling”; the Quinnipiac guy told the truth: it’s not that you over sample any or other group. You just sample. When you find a higher percentage of Dems, this is a finding already, not an oversample. This first result (with the famous oversampling) should be helpful to make a second stage: a stratified sampling, with randomizing blocks and taking care to eliminate any bias. Nobody, as far as I know, makes this second phase. Then there is the turnout, which is another link of the chain. Finally, there is yet to see the effectivity of the method for predicting a result (historic results, error margins in the past, etc).
As currently presented, I distrust all the polls, they look more like witchcraft.
“When you find a higher percentage of Dems, this is a finding already, not an oversample.”
This is a finding only for a small segment of a state. It would be like polling only Philadelphia and claiming that it represents the views of all Pennsylvanians. Come to think of it, that’s probably exactly what the pollsters are doing.
Charlies article starts out good: it lays out the nuts and bolts nicely.
The random sample: choose people by some random process, 1000 gives you +/- 3%
The stratified sample: choose people in certain catagories (like mixed race Budhists and so forth) in proportion to the population, you can get +/- 3% with fewer responses IF YOUR MODEL IS RIGHT (trade off is cost against potential bias.)
The turnout model: whichever type of sampling you used you adjust because expereience has shown that voters of one party are less likely to vote then another (Deb Sam raises this in post 22).
Fine.
But then Charlie takes as an example a poll produced using, as far as we know and he knows, a random process, with no stratification, and a flat turnout model, and claims that actually it is a poll with a +8 Democrat turnout bias.
Then with no justification he says if we reduce the democrat turnout (by applying a -3, -4, -5 republican turnout model (although he does not make plain what he is doing here) the pool looks different.
Somehow, Charlie has managed to give the idea that a random sample that contains more members of one party ‘oversamples’ members of that party some traction. This is quite impressive in a slightly depressing way, how a nice turn of phrase can be used like that. But really, what’s the point?
I meant -3,-4-5, DEMOCRAT model (assuming anyone read that far).
No, you’re mistaken: I say that the poll shows a turnout of D+8. That’s not bias, that’s just what they have as turnout.
The next question is, do we believe that sample. At the Quinnipiac guy said, if they did a poll and ended up D+50, they’re probably go back and try again, because that sample would clearly appear to be somehow wrong. So you look for a way to validate it, and you look if there are any sources of systematic error in your sampling. But in the mean time, you can estimate what the values would have been with a different turnout, leading to that table.
My problem is that polling 30 to 38 democrat is not a ‘+8 democrat turnout’.
It is a poll that is 30:38 to democrat assuming equal turnout. (i.e. people who say ‘democrat’ and people who say ‘repbulican’ are equally likely to vote the way they said).
So if thats right I just can’t see how that last part of your article is at all justified. It just looks like ‘and now I will reduce the dems expected vote for no particualar reason.’
What am I missing? What’s wrong with my descripton of it as a 30:38 poll for dems? How exactly is it skewed?
Well, “D+8″ in this is shorthand for “Democrats 8 percentage points over Republicans.” That makes it easy to compare to other people who are using the same shorthand. Is it perfect? No, but then I’m avoiding a whole lot of other things that would make it more technically well-stated.
The answer to your other question is that you’re making an assumption I didn’t. I’m looking at there being 38 percent Democrats vs 30 percent Republican and saying “that doesn’t seem consistent with other data.” That doesn’t mean its purposefully skewed. It does mean that IF the reality is different than the sample they have, the results would change.
Well yes thats true. If the reality is differnet from the poll the result will be different from the poll.
Thanks for taking the time
And remember that necessarily any poll, based as these are on a sample, will be at best an approximation.
If I am understanding this right and I am lousy at statistics as soon as we get out to vote we are skewing the data sample for the past poll. We are creating the actual data set that the polls are only stumbling towards.
Therefore the only way that advanced polling can possibly be accurate (not correct because there is still a 50/50 chance the polls prediction will come true) is if there is a landslide about to occur. That 1,000 people are a hard and fast set because everyone has already made up their minds for one candidate and in the long run the turnout model would not matter. Anyway you could slice up the turnout model would be correct because the mass of voters towards one candidate is overwhelming anyway.
The answer is really turn out. Bring your neighbors and co-workers who hate Obama (they are legion – this guy is despised) – We are the sample size we have been waiting for.
– populace has changed since 1980, but perhaps not human nature. What will determine this election is at the last moment each voter asking, before marking their ballot, “Am I confident going with what seems like will be more of the same or do I take a chance on something different?”
It is more likely that folks will simply say I am voting for Obama so I can get more free stuff = Democrat Voter.
Or … I am sick and tired of paying for all those democrat voters who are on the dole all the time. = Republican Voter.
The undecided AKA the Ditzy will probably flip a coin on the way to the voting booth. = Stupid Voter.
So the question is simply … are there more of them than there are of us.
I don’t think the chart is correct. If a sample is D+8 and has Obama leading by 8 at 53 to 45, then on an even sample we should have the race tied at 49 each, not Romney leading by 8. D+8 means that there are 8 percent more D than R, so each point the D+ number goes down should leave a difference that is one point less, not two points as in the chart…right?
The D goes down by 1, the R up by 1, no? Under the simplifying assumption that they don’t vote for Gary Johnson instead.
Sure, but when the D goes down by 1 and the R goes up by 1, that’s D+6, not D+7. To get D+7 you drop D by .5 and add .5 to R.
Think about it like I have 10 ping pong balls, six on the left and four on the right. If I move a ping pong ball from left to right, that reduces the left by one and increases the right by one.
Thanks for your patience. I agree that that’s what happens. But, as I said, the result is that the difference between D and R is reduced by 2, not 1, when 1% leave Obama and go to Romney, which means that D+8 goes to D+6. On your chart it’s marked as D+7.
I mean, when 1% is moved from Democrats to Republicans.
I think Shemp is right. The 10 ping pong balls is too macro-scale to demonstrate a 1% change in D advantage. If there are 200 ping pong balls, a 54% to 46% advantage results in 108 to 92 advantage for Obama (assuming all Ds vote for Obama and all Rs vote for Romney). If D advantage changes to D+7 from D+8, you would have the percentages at 53.5% to 46.5% (that difference is D+7), which results in a 107 to 93 advantage. Every 1% change represents 2 votes. But since each side is only decreasing or increaing 0.5%, the vote total on each side only changes by one vote. So if a D+8 advantage shows an 8 point lead for Obama, the D and R numbers being even should result in an even race.
If I am understanding the discussion here it would be like 1 ping pong ball on the left going away (staying home), but not going to the right.
I’ve only been called once this season, by a “non-partisan” polling organization. I told the lady that I’d answer the questions, but that there was no such thing as a “non-partisan” polling organization. She hung up on me.
I didn’t read all the comments above, but one consideration seems to be left out: Is there a skew in the proportion of Republicans and Democrats who are available when the pollster calls and who are willing to take the time to answer the questions? It seems to me that the populations that will be under-sampled are those that are unlikely to be home when called, or are too busy to answer questions, or who hang up as soon as the caller identifies himself as a polling nuisance. The question then becomes: Who is more likely to be unpolled: those who are going to vote Republican, or those who are going to vote Democrat? Remember: If you have a job and/or a family, you have no time for this nonsense.
Even a simpler way to look at is that Democrats are lonely people who answer their phones and need someone to talk to, even if it is a pollster. Republicans on the other hand are being productive and are out and about so fewer of them answer. Or with Republicans they are so fired up for the election and angry at the media, including pollsters, that they pick up the phone and hang up right away such as I do… Either way it looks, only by appearance, that Democrats outnumber Republicans as voters and have greater numbers for Obama. Polling is only being used to attempt to discourage conservatives…
somewhat true…actually Dems tend to be home on weekends and answer the phone so when a poll has weekends in its survey dates it tends to be off, even the accurate Rasmussen’s results are off on Mon and Tues since they include weekend surveys), Thurs and Fri are best
Romney has low poll numbers on trustworthiness. Romney repeatedly refuses to tell voters the specifics of his tax plan or his own tax return, just says “trust me.” Romney is losing to Obama by double digits in the polls. Romney can’t connect the dots. Romney is no “turnaround genius.” Romney is not qualified to be president.
Obama has many unreleased documents from college that probably show him illegally claiming foreign citizenship to gain entrance into college with student loans. This is a felony to lie on federal forms and is an impeachable offense THEREFORE Obama will keep all his college records sealed or risk impeachment… I”m just saying.
Of course the cover-up could be that he had so many C’s, D’s and F’s in college that any normal person would never have made it to Harvard or even graduated. Again assumptions made on evidence and missing information.
There is one other possibility, that his college records show so much disciplinary action from his illegal drug use that the only people left to vote for him would be 60′s potheads. After all the President is HIP!! :/
I don’t really want my president making the TV show rounds— I want him GOVERNING which our current President doesn’t do or can’t -just look at his college records. OH WAIT, I can’t, they are sealed??? Obama acts like a parent who only wants to be his 15 year olds’ best friend!
I’m definitely not worried about a successful businessman who paid over $1,200,000.00 in taxes and donated over $3,000,000.00 to charities.
Here, I will put together a series of sentences and pretend it is a logical conclusion, just like yours:
Hi, my name is Obama, and I like to eat dogs. I regularly hunt down stray dogs in D.C. and lure them into the White House using a sock puppet shaped like a cat. There are more people out there who like to eat dogs than those who like to eat cats–lots more. Cat-eaters are dumb. Cat-eaters could not find their way out of an open paper bag. Vote for me because I eat dogs.
It works for me since I am a cat. I would rather elect a dog eater than a cat eater.
And that’s different from your messiah, the divine 0, how?
A few thoughts on an interesting article. The field of statistics will eternally wrestle with the question of whether the sampling size and technique accurately reflects the population of interest. In political polling (vs reading a meter) the human element is the big unknown. There are millions of people who care deeply and are informed about the NFL referees, or the World Series race,who have not thought deeply about for whom they will vote, if they vote. They will blow off any phone sampler with any answer, today.
This population will decide who is President, because the race is razor thin.
The above reality is well known by people whose livelihood depends on the results. They are certain that their mortgage will come due in January, but they have very little certainty they will be employed. These folks spin, hard, to discourage the other side’s constituency by “proving” it is a done deal. Some people do not like to go out in the rain to vote for a loser.
Based on past elections, I predict a landslide the weekend before the voting, as last minute decisions are made. The underlining issue will be the economy, but the deciding gut issue will be choosing an old white rich guy, vs. a bumbling socialist who again promises a bright future, after failing, badly, to deliver. The average voter is broke; food and fuel are sky high, their college kid is in deep debt with loans, their job security is non existent, the national debt has destroyed their nest egg, their home equity, and somebody over there wants us dead. Put your trust in kindly Bain Capital or drop your drawers for TSA.
Reviewing the cards……. in this election, more than any in my long life time, “you bet your life” is the choice. America can not continue to pick losers anymore.
Here is what is a mystery to me.
What has changed to cause all of those who voted in 2010 and who delivered what Obama himself described as a “shellacking” to suddenly decide that they love Obamacare; want higher taxes; want more regulation; want 8%+ unemployment to continue; and to feel safer from a security perspective after Libya and all of the anti-America unrest across the Muslim world?
How is it possible that so many angry unhappy people in 2010 are now realizing that they were wrong then and that Obama is the right guy to deliver a brighter future?
Is it because those who felt this way are now so demoralized that they are going to stay home, while those who like Obamacare; pay no taxes; don’t believe regulations affect their lives; are employed and economically comfortable; and like Obama’s foreign policy approach of assuming that we won’t be affected down the road if we ignore the cancer as it spreads throughout the globe, all are so enthusiastic that they can’t wait to vote?
Maybe because mid-terms are like a referendum on “How much do I like the incumbent”, whereas a presidential election is “Do I want Romney or Oaama”, its easy to dislike Obama a bit and dislike Romney more.
Its the difference between a vote to protest about the things you dont like about a president and a vote to choose a president.
RIght, all those tea partiers were really just registering their protest but having done so are now happy.
Just like they were happy to vote Walker out of office in Wisconsin when given a second chance.
I’m thinking unenthusiastic Obama suporters rather than tea party types.
I suspect it is true that there is a contingent of low info Dem voters who only come out for general elections and not midterms.
The turnout for oh-ten will not be a proper model for this year. However, oh-eight was an exceptional year and I bet will be a poor model for this year, too.
I just assumed they called during the day. You know, when only democrats are at home.
Of course, that’s it! Finally an explanation that makes sense.
It sounds like a joke, but some variant of that is certainly possible. Another possibility — that seems intuitively plausible — is that as more people become disgusted with the legacy media, they’re more likely to hand up on pollsters.
Bingo! I work 2nd shifts, and as such I sleep from 3 AM to around 10 AM. Normally, I get poll calls between 11 AM and 1 PM.
It seems the political pollsters have taken a page from the climate change modelers.
Since a critical step in processing the data is applying one’s “likely voter model,” one can tweak the model to deliver the answers you seek.
Wait, you may be onto something, maybe there is really only one pollster and its Michael Mann playing with a new hockey stick.
“… actual result will be more than 56.5 percent Democrat or less than 49.5 percent Democrat.”
Shouldn’t that be less than 56.6 … or more than 49.5?
Excellent explication of much-muddled (usually by pols & their lamestream fellow travelers) topic!
Sorry, Sissy — it’s hard to say these things really clearly without resorting to math notation. But it’s
* ONE chance in 20 it will come out MORE THAN 56.5 or LESS THAN 49.5 — that is, OUTSIDE the margin of error
* NINETEEN chances in 20 it will be less than or equal to 56.5 and greater than or equal to 49.5 — that is, within the margin of error.
Real Clear Politics shows Obama leading in the electoral college by a BIG margin. That’s where the election will be decided. I do not rejoice, but if I had to bet the farm I’d bet it on O.
Richard
That would be a bad bet. If you remove the outliers Obama is within the margin of error and below 50% most of the swings. That’s not historically good for an incumbent.
Intrade has the mormon dropping like a rock: 23%
My thoughts exactly.
This election is about free stuff, not the economy in general or the disastrous Obama foreign policy. Free stuff has created a perverse electorate where the worse the economy gets the more the voters swing to Obama. As voters become more concerned about their economic future the more the favor free stuff because when they lose their jobs the government will be there to give them things. This creates a large potential for an economic implosion. As economic conditions deteriorates more people will get free stuff and fewer people will pay taxes. The free stuff eventually dries up as the remaining productive sector of the economy shrinks.
The polls are correct. People will vote for Obama not because they think he has done a good job in his first term or will do any better in a second. They are voting for him as an insurance policy that will give them goodies if their own economic prospects go south. There are not enough taxpayers remaining to compete against a campaign based free stuff.
Goodies are in the eye of the beholder. For instance, look at what the eight years of deficit financed Bush tax cuts for, so called, “job creators” got us. Certainly no new jobs. In fact we were losing 800,000 jobs a month when Bush left office. What did the “job creators” do with those tax cuts? I’m guessing they took a bunch of hookers on spa vacations. I don’t want my tax dollar financing “goodies” like that.
The highest Bush deficits were 1/3 of the average Obama deficit and they occurred right after 9-11. The last Bush deficit before the downturn was on the order of $170 billion dollars or a little more than 10% of the current deficit . According to “Reckless Endangerment” a book written by two New York Times reporters the crash was indeed caused by the Democratic Party controlled Fannie Mae and Feddie Mac. We all know that Chris Dodd and Barney Frank, both on the Fannie/Freddie payroll, stymied every attempt to head of the crisis. You do know that Obama was also on the payroll don’t you? He was #4 after only two years in office. So the Democrats set the conditions for the crash and the huge job losses you cite.
Bush’s term also started with the burst tech bubble, shortly followed by the 9-11 attacks. We now know that there was not one but two opportunities for the Clinton administration to take out Bin Laden. Unlike Obama, President Bush did not blame his predecessor for his problems. Unlike Bush, who as Governor of Texas, had no ability to resolve the growing AQ threat, Obama was right in the middle creating the crash which brought him into office.
These are the facts and I have no interest in debating with someone who contribute nothing of value to society. But I am going to give you something to think about. Obama’s second term will not set the stage for your socialist paradise. Like Detroit, Chicago and California, the US will go under in the next four years and much to your surprise there will be no cry for more socialism. When the Republic fails from your Cloward-Pivens driven economy what we are going get is a Pinochet. So I want you to remember this discussion when they come to round up you and your fellow Progressives and put you in some football stadium for disposal.
Another math wonder. Compare the last month of Bush’s administration as the Great Recession was beginning, with the high of his administration, and then claim Obama has created jobs by comparing the low point of the Great Recession to now.
But pointing out that unemployment is still higher than it ever was for Bush and would be higher still if so many people hadn’t just dropped entirely out of the workforce is probably racist.
Kool-Aid, Kool-Aid, yumm yumm yumm.
When I drinks it down I drinks a ton.
One fairly simple way to check for bias would be to ask registration, and compare that to actual registration. One of the reasons these swing states are swing states is they have pretty close to equal numbers of repub and dem registered voters. So any poll there with dem +8 would be a little suspect in my book. But this can be verified by checking actual voter registration rolls. I think most swing states do have more dem registered voters, maybe +4. But then you have to say how many of these registered voters will turn out, and once you factor in historical values for that the actual voter turnout suually comes about about equal between repubs and dems, with independents, where most polls, even the skewed ones, show romney with a small lead, decide things.
In 2008 there was an unusually large dem turnout, and any poll using 2008 turnout numbers is probably suspect. In 2010 you had a historically high repub turnout, and using that turnout model could also be suspect. I suspect the best turnout model would be somewhere in between those 2, which I think is about dem +3, which I beleive is what happened in 2004, and which if used in this articles poll would give romney a small lean nationally, but might allow obama to win some key swing states.
I have noticed the 2 polls that do not have dem skewed turnouts, gallup and rasmussen, show a dead heat nationally, and a 1-2 obama lead in most swing states. To me, that says Obama is slightly ahead, but not enough to matter with 2 months still left. I think either side, or the media, is making an error at this stage predicting a landslide for anybody at this stage. Now if this possible obama lead is still hlding up a month from now, I might start to worry.
Look, the same methodology has been in place for a long time. No one is saying they can’t change in the next forty days but to deny the 47% comments, coupled with the Obama campaign’s ability to define Romney as out of touch with the average working stiff, has not had an effect, is to deny reality. Obama has the edge and it will be very difficult to alter the narrative without something bold on the part of Mitt and let’s face it, Mitt hasn’t a bold bone in his body. He always plays it safe. He won the primaries by finding his opponents soft spots and exploiting them. While Obama certainly has weaknesses, it’s not enough to close the deal. Romney has effectively been defined as a rich white guy who’s only concerned with the plight of other rich white guys. He has not done anything to alter that perception. He needs to come up with actual detailed policies and not generic platitudes. His positions are short on details and full of contradictions. He wants to lower taxes on the middle class but he wants the 47% who don’t pay taxes to pay more? Which is it? It can’t be both. He wants to lower taxes on the wealthy but he warns them their taxes won’t go down because he’s going to “close loopholes”. Which ones? Unless he can make a case that he’s a viable solution for middle class angst, he’s a goner.
” He wants to lower taxes on the wealthy but he warns them their taxes won’t go down because he’s going to “close loopholes”. ”
Are you old enough to remember 1986 ? Reagan supported tax reform that cost me that year $100,000 in additional taxes. I had to borrow money to pay my taxes but I supported the bill. It closed loopholes and lowered rates but that didn’t help me, a high earning professional. I had a midyear corporate year end. That allowed me to plan my income that year and minimize taxes. The new law banned mid-year corporate year ends. So, I had to pay 18 months of taxes out of 12 months income. I suspected that the next time Democrats were in office, they would raise rates and add loop holes. They did. I was able to pay lower taxes under Clinton although the rates were higher. At least Clinton was able to learn from the GOP landslide in 1994. I only wish I had gotten fully invested in the stock market after that but I was a student working on another degree after early retirement because of back surgery. If Obama is re-elected, I will be wiped out by hyper-inflation but I’m 75 so it will be your problem. Good luck.
All Romney said is that the 47% of people who don’t pay income tax aren’t receptive to talk of income tax cuts.
President Downgrade’s comments about people who live in small towns clinging bitterly to guns, religion, and racism were much, much worse.
My guess?
Right-leaning moderates and conservatives have lumped pollsters into the same group as MSM reporters.
In significant numbers, they’re refusing to play a rigged game, and today’s poll numbers reflect that.
In order words: D+8 samples are common ’cause of self-selection.
Some pollsters are weighting based on their presumption of the turnout model. Others are “taking the raw data”. If there is a systemic shift across multiple pollsters finding more “self identified” dems, ask yourself why?
Some pollsters are criticized for not including cell phones in their calls. The argument goes that a larger and larger portion of the population doesn’t have a land line, that those with only a cell phone tend to be younger (and thus lean dem) and doing so skews polls away from the dems. There is certainly some validity to the idea. Of course with Obama’s economic policies producing about 1 in 4 un or underemployment with those same younger crowd, you have to ask yourself how much the assumption they lean dem holds.
But what about a cross poll tendency to find more dems than GOP. How can that happen? Something similar actually happened in 2004. The initial exit polls showed Kerry winning in a landslide. The only poll that really matters (aka the election voting) begged to differ. They found that GOP supporters tended not to answer the exit pollsters questions, producing a similar biased poll that followed an accepted methodology and produced spectacularly different results.
While this may come as a shock to the NYT editors, a large percentage of the American electorate has reservations on how much they can rely on the main stream media to present a balanced view of the world and not a view that consciously or unconsciously leans to the left. I’m not sure it has gotten to the point that if the MSM reported the sun rose in the east, that a fair number would look out the window before accepting the assertion.
Could it be that like 2004, a large enough portion of the electorate is declining to participate in the MSM farce and avoiding the pollsters samples?
So what is the political junky who knows about the problems with the poll but still needs a daily fix do? One answer is to re-weight the polls. Before you start crying about just substituting one fantasy that Obama is doing as well or better in dem turnout than 2008 with another, re-weighting to any turnout model will help bring the story into clearer picture. Because all the polls are using different turnout models (explicitly or implicitly), they are automatically at odds with one another. The polls saying the turnout is going to be D+8 and D+5 may both lean to the left, but they aren’t in sync. Re-weighting both to the same D+8 or D+5 at least gives you a certain amount of alignment.
If you are really ambitious, you can re-weight to a spread of potential results from D+20 to R+20. At a minimum this will teach you that turnout models make a huge difference in results.
What is the best turnout model? For the last several elections, the previous off year house elections give a pretty close approximation for the following presidential election. For 2010, that would be even (D+0 or R+0).
We shall see what we shall see in November. But remember, there is also something called “battle space preparation”. Specifically, “we was robbed”, just look at the pre-election polls can be a first step towards delegitimizing a Romney election.
I hate you.
I REALLY HATE YOU!
Why didn’t I think of that name?
I stand humbled before you, Master Yetanotherjohn.
Don’t get cocky.
YaJ: “Just because I’m paranoid, doesn’t mean nobody’s after me.” Suppose Obama’s internal numbers show him losing by a very small margin. One way to reverse the result would be mass riots, which could be engendered by convincing the would-be rioters that Obama goes into election day with a commanding lead.
In the end, the only thing that matters is the vote. Keep remembering that Obama was elected by only 30% of registered voters. There was a 58% turnout and of the people who actually voted, 52% voted for Obama. Multiply 0.58 and 0.52 and you get 0.30. Just 30% of registered voters saddled us with Obama. Forget about the polls and just vote. Any poll that isn’t taken from the same quantity of Republicans and democrats is obviously skewed.
most of these polls are showing Romney winning Independents 60% – 40%
based on that Obama needs turnout in Nov to be +10% nationally to squeak out a win … thats over 2 points higher than the 2008 party split …
If you believe that you are beyond debating with, you are an idiot …
OBambi’s not worried about that!
Soros is taking care of THAT!
If Ostupidhead and Biteme ever tell the truth, I wonder if their noses will fall off?
Teh Won’s Mom and Pop were both conmen. The Obamic apple didn’t fall far from the tree. When a con is running, you ALWAYS double down! All else gets you arrested. The only way to save the con is to reinforce it.
Given that, (and the MSM complicity, as they go to that happy hunting ground reserved for dying species) how would they even begin to STOP lying? Has anyone given that much thought?
If you are standing in Times Square, naked, pretending to be perfectly normal, how do you finally end it? You can’t. Someone needs to take you away for your own good. So you stop harming yourself and others.
The Left and their pathetic champiom have blown their load. Lies and deceit are all they have left.
Before we get too cocky, Hitler and his buds were in much the same place. A brave war hero blinked. All the rest was carnage.
We’re not playing any more. This is deadly serious. Deadly serious.
There’s a reason it’s called “deadly” serious. Pray you never find out why.
Except, I don’t know a Republican who is willing to do a poll, because they don’t trust the pollsters. So, anytime you do a poll and get all Democrats, it’s because they are the only ones who’ll talk to you.
Hey, who are you?
I got here first!
Regardless, I agree with you.
Election Night:
Anchorman: “Well, Bob, based on what you’ve seen at Democratic Party headquarters tonight, what’s your election results prediction?
Reporter: (*Shifting uncomfortably*) “My prediction? Ummm, err, PAIN.”
From what I understand, the supposed partisan movement towards the Dems isn’t showing up at all in contested House races in these swing states. I understand the philosophical reluctance of pollsters to want to hard-weight their results using a plausible turnout model, and instead let the data speak for itself. Unfortunately, it’s obvious that their supposed random sample is nothing of the sort — perhaps Rs just refuse to speak to pollsters at the same rate Ds do.
When we re-weight these things, we are imposing priors on what we believe to be bad data, which ultimately seems like a reasonable thing to do. At least, it’s more reasonable than assuming Ohio is D+11.
Not to mention the fact that I’d wager the average R is more pissed at pollsters (and media) than the average D. I NEVER answer polls.
This alone would screw their calculations.
None of this concerns me in the least. Reagan, at this time in the 1980 process, was declared a loser by America’s finest, college-trained statisticians. George W. in 2004, likewise deemed DOA by “scientific” prognostications that carried through right up to the “exit polls.” Remember? I sure do. I remember the glorious sight of several hundred tattooed and bestudded liberal arts majors at Macalester College in St. Paul, pouring out onto Snelling Avenue in an absolute RAGE that those polls had not been fulfilled. This happens year in and year out with the leftist media. It is such an overused tactic that it has become entirely pathetic (or at least it should be, despite our sniveling RINO “experts” in the alternate media).
And they wonder why everyone is tuning out the alphabet networks and America’s “leading” newspapers. It’s like they never left their grad-school seminars on the superiority of Neo-Marxian theory. They simply cannot think, rationally & substantively, for themselves.
As always when their guy is in the toilet, the pollsters call it close in his favor. Anybody who’s been around long enough to have seen Nixon, Reagan, and the Bushies take the Whitehouse understands that the MSM is smoking rope, believing their own bullcrap.
I am a very experienced professional pollster (our clients are mostly media organizations but my company’s polls are NOT biased and I have data on thousands of polls to prove it) . I can tell you that raw samples are always adjusted and weighted to account for the fact that different demographic groups have different likelihoods of responding to polls, but the weighting is generally done for gender, age, and race and only sometimes done for party.
It doesn’t matter whether Quinnipiac chose to get a 9% D-R advantage in Florida and Ohio, or whether they just got a skewed sample and didn’t correct it enough. Those numbers defy common sense because they imply Obama and the Democrats will do better than in 2008, in defiance of everything that has happened the last 4 years and all previous political experience.
The MSM has very strong incentives to paint Obama as doing better than he is, and some of them do it unconsciously and others consciously. Look at the candidates themselves and how they are allocating resources and you will get a better picture; that picture says that Romney is more comfortable with the situationt than Obama is. (Note I did am not saying either that Romney is likely to win, nor that he thinks that he is is likely to win; rather, I am saying that his campaign is basically proceeding according to plan, while Obama’s campaign is flailing about.)
Whoever wins Ohio and Florida wins the election. If they each take one of those two states, anything can happen, and we will be in for a bumpy ride.
You hit it on the nose, and it’s interesting that in the last 48 hours or so I detect a small but detectable decrease in the MSM chatter about Obama’s poll numbers. Something is up. Watch where they are campaigning to get a true sense of where the race is at. Whoever wins OH and FL will win. This is traditional R turf, despite last cycle. I think the fundamentals for O are tough, personally, in both states.
Here’s the intriguing question. Do you think Axelrod’s suit against Gallup and the perceived threat of investigation for not weighting minorities properly has anything to do with the polling this current election cycle?
When one poll has O up 2% in NC, and nobody is campaigning hard there, you know that the results are not realistic.
I’m really curious about PA. Assume D turnout is down in Philly due to the voter ID law. Also assume that R turnout is up Western PA due to disgust over coal and fracking regs. The Susqehanna poll had is 48-46 O a few days ago? Real, or fake? Will R take a shot after the debates?
It’s really very simple
The People who do 99 percent of the Polls
Want Obama to win and to suppress the Vote for Romney
even if they have to start making numbers up out of thin air
It Happened with Carter / Reagan
It is just history repeating itself
When the Landslide for Romney happens
The Liberal News Media talking heads will be spinning and showing shock and passing out on the floor
It will be Glorious
What time of day do the pollsters call?
If the calls were made during work hours, only “takers” would be asked, the “producers” would be working.
What happen to those voters who don’t have land lines, and the voters who don’t answer the phones when the numbers displayed are “unknown”? How do pollsters account for those?
In 2008, I occasionally picked up polling calls. I’ve dropped my landline and haven’t accepted “unknown” calls since 2009.
Kerry was supposed to have won the election based on exit-polls. His Intratrade was above 50 until the last week of October.
Advice: vote early, vote often; find some votes in abandoned warehouses; set up polling places in cemeteries; those who are in Chicago, vote in Wisconsin, they don’t check ID’s; do the right thing: throw away your husband’s Republican vote as a Democrat journolist advised; make sure your husband vote the way you do, so at 60 years old, you could keep your “reproductive rights” and your “free condoms”…
I wax orgasmic just thinking about it. :-O
I get polled several times a week. It must be because we have a land line. I’m sorta fed up with the media, with the crappy polling, and the total lack of actual science in any “study” that seems to make it to the news today.
So I just make stuff up when they call. I have been polled 20 times and each time I am a different race and voting for a different candidate. I was even female when they called 3 times a day for days and would only talk to my wife. I recognized the caller Id, and then used a really lousy fake female voice. They took the poll.
So how are they factoring for this?
Somthing is up in PA.
62% of the population strongly support the current voter ID law as is. 66% of independents support it in the most current poll today. Most democrats are against it.
If O was going so strong in PA, wouldn’t more people be against voter ID?
the one thing about the polls…
they usually feature 80%+ of the respondents saying they are going to vote.
turnout in 2008 was 60%.
I would rather have romney’s 46%, with their intensity, versus the lukewarm support that the pollsters are counting in voters, to get obama to 50%.
Gentlemen, gentlemen (and/or ladies). We are becoming bogged down in mindless minutae.
The authors most important point was correct. It all depends on how many Democrats show up to vote on November 7th.
I encourage them all to show to vote that day. IF they can tear themselves away from whatever a “snookie” is and her continuing adventures on the telly.
They are not if they think you are voting Democrat. Not only are they NOT factoring it in they are probably counting you two or three times in the grand tradition of the Democratic Party.
Democrats are always over sampled because it is they who are most likely to be home watching TV or just lazing about the pad having a beer or a toke ….. Democrat voters are more likely to be unemployed etc.
On upside, those democrat voters are more likely to stay home and watch TV or play video games on election day … that they may not even realize is voting day.
Something concerns me about this debate about polls.
Quite often when a candidate is losing according to a released poll, they will release at least part of their own internal polling to show that they aren’t losing after all. Why hasn’t the Romney camp released any internal polls or even mentioned them? We know they have their own polling figures and we have to assume they aren’t part of a conspiracy to show Obama in the lead.
By the way people, stop panicking. Most polls show this to be a close race with Romney behind, but within the margin of error. Let’s wait until after the debates.
64. Northern Light
Something concerns me about this debate about polls.
Quite often when a candidate is losing according to a released poll, they will release at least part of their own internal polling to show that they aren’t losing after all. Why hasn’t the Romney camp released any internal polls or even mentioned them? We know they have their own polling figures and we have to assume they aren’t part of a conspiracy to show Obama in the lead.
Maybe because every time a Romney supporter hears of another BS poll from the left leaning media, it enrages them and makes them just that more determined to crawl bare foot, over broken glass, if necessary, to get down and vote. On the other side, the Dem voters become convinced that it’s all in the bag; so they can roll over and go back to sleep.
1st Roman citizen: “Say, how are the Christians doing in their quest for influence?
2nd Roman citizen: “I’m not sure, but there’s going to be a debate in the Colosseum next week.”
1st Roman: “Oh? Between who?”
2nd Roman: “The Christians and the Lions . . .”
“There are three kinds of lies: lies, damned lies, and statistics.” (Attributed to Twain.)
There are 3 big problems with stats:
1. pre-test probablity
2. sample bias
3. Human beings. These tend not to be rational or truthful.
My guess is that the most significant event effecting the election id the economy on 11/6. And no one can predict that.
I wouldn’t be surprised in Obama released the fuel reserve or the last of the stimulus money a week before the election.
This is as misleading an explanation of the sampling process as the final results themselves are.
Sampling methology stands or falls on whether the sample is selected in a random number process from a population.
Let’s take the given example of a 1000 people from a 100 million population. These 1000 are, first of all,not people, but phone numbers. And, therfore, that 100 million must refer to that many phone numbers. Do they? Nope.
Then, take the very real problem of landlines and mobile numbers. A good list of ALL landline numbers can be found in any city directory. But, mobile numbers? Is there a reliable directory of mobile phones? Also, condiser two further issues. People, esp young, educated, working people are likely to have more than one cell (just as there will be more than one landline phone in a household). Also, they replace their cell very often.
For a population of 100 million phone numbers requires a good way to merge both landline and mobile phones into a single list, from the 1000 sample numbers are selected. Can that be satisfactorily?
Let that rest for now. The writer says that the pollster takes the sample districution of party allegiance and corrects that to fit the turnout model.
Yet, the pollster quoted said he does no such thing. If the sample shows 37% Democratic, then, that it is.
Yet, the sample may throw up only a 32% Democratic. Why not, since it is all random? It is here that the pollster “Corrects” the raw data to the turnout model.
And that is exactly why we all say the polling is skewed. The turnout model they arbitrarily use is the exit-polling data from 2008. Not from the 2010 election. And they also ignore the “enthusiasm gap” their own polling throws up.This is how raw data is massaged.
First, the raw data cannot “represent” the population of 100 million, since you don’t have all those numbers. It can be got, but it will bankrupt you.
So, the “correction” becomes necessary. If the 1000 represented the 100 million, no correction would be necessary.
Any pollster who neglects the following is fooling themself:
MAJOR PARTY ID (exit polls):
R/ D/ spread
Nov. 2000/ 35%/ 39%/ D + 4
Nov. 2004/ 37%/ 37%/ even
Nov. 2008/ 32%/ 39%/ D +7
11-01-2010/ 36%/ 34.7%/ D –1.3
09-21-2012/ 37%/ 33%/ D –4.0
source: Jay Cost
I saw this over at hillbuzz:
Torture the numbers long enough and they’ll confess to anything.
The polls are as distorted as the media outlets proclaiming them. They used to ignore the polls that didn’t look good for them and broadcast those that did, now they’ve taken it a step further, manufacturing what they want to broadcast. Much of their Romney coverage falls into this category. Edited films like at Mother Jones and Andrea Mitchell’s on MSNBC. Manufactured ‘gaffes’ instead of a terrorist attack on our Libyan embassy.
If you buy these polls after all the proven misinformation in the press, you deserve this President.
If Mr. Obama buries Mr. Romney in a landslide, which I see as likely, will there be a follow-up column explaining why this writer got it so wrong?
Disclaimer: I am a US Army veteran (E-5, Honorable). Your patriotism may vary.
Absolutely.
If there is a near landslide or strong victory for O like the current polls appear to predict, in this economy (the worst since the Great Depression), I would be the first to say “wow, we all were wrong, what were we thinking.”
Buries Romney in a landslide? Look at the facts on the ground. Why isn’t O going to AZ, or MO, much less NC? That gives you a better sense of where the internal polls really are. If this was going to be a landslide, you would see a lot more of that.
I think both sides are waiting until about a week to 10 days from now (letting the first debate settle in) to see where the race is really going, and to determine where to fight.
Right now it is likely a draw, with perhaps O up by a point or two in some swing states, down by a point or two in others.
As someone who thinks Obama is an unserious, unskilled person way in over his head who should be tossed on his butt out of office, I hope and pray that the guys running his campaign are thinking like you and manage their campaign accordingly.
Yeah, sure you are. And I’m a ballerina.
So the question which you didn’t explain is, why do the polls skew for Obama rather than the other way around?
Told ya.
http://firstread.nbcnews.com/_news/2012/09/28/14140650-romney-were-going-to-win-pennsylvania
Maybe it’s irrational exuberance, but I think they are going to make a mild play after the first debate. If they make a play at PA, it means that they are stronger than the media is letting on.
Hmmmm.
A 14 year old could fool a pollster into believing he/she is a Democrat who plans to vote for Obama. Pollsters just want to finish the sampling data and go home from work on time. Just like the rest of us. Polling is only popular because media outlets, too lazy to pursue real news, can always be counted on to waste endless amounts of air time yucking it up about the latest poll. So, all your polling outfits want to be the latest poll. This practice is pathetic. The only poll that will count is the one WE do in November. Get out there and vote, no matter what the pundits are saying. Their opinions and a couple of bucks will just about buy a cup of coffee. Obama can and will be defeated! Communism will be dealt a major blow! ABO2012
3 Democrats Charge the American Press with CRIMINALLY NEGLIGENT BEHAVIOR
How does the turnout account for people who lie in their responses? ie the Wilder effect?
I got bored trying to read all the comments, but there is one point that I have seen made before and makes a LOT of sense. In general, Reps are reluctant to answer polls at all because they value their privacy. Out of the 25 people in my family only 6 are Dem and while all 6 would answer a poll, it is very likely that out of the remaining 19 Reps, only 3 would answer!
I cannot believe that people are stupid enough to believe that Obama has a chance to win this election. He has done everything he could to destroy this country with enormous debt and is just waiting to get past the election to put the finishing touches on bankrupting our financial systems and causing a meltdown of the dollar to 3rd world country status. If People cannot see what this man has done to our economy and our laws by issuing presidential orders for everything to bring us to the brink of a disaster worse than the 1929 crash and placing communist, socialist, Marxist and revolutionary groups into positions of power to disrupt our government and all its departments into destroying business opportunities for everyone in America then they deserve what will come next.
Ya know, all this reminds me of the OJ Simpson trial…Remember the shock wave at the vertdict?? No one & I mean no one could believe what happened…Greta Von whatsherface started her career, other people were ruined, books written…This election will have the same impact..I don’t believe any poll..& I don’t answer any poll..
Ha Ha! Complain all you want about polls. Romney is losing because he is a terrible candidate (how bad must a candidate be to be losing to Obama this year!?). Romney can’t keep his stories straight — he has no opinions — no one likes him and no one should.
Pollsters have been dead-on for elections since Reagan-Carter (and even there it was clear why they were wrong by so much — a very late debate changed the balance, and there was international politics at the last second).
Romney better hope for a bump in the debates — that or a world crisis that he manages to handle better than Obama this time.
Carter made empty threats against Iran that he was not prepared to backup. Obama isn’t that crappy yet. The feeling of rage against Carter was palpable. Most Americans think your hero Obama is an idiot, but they don’t hate him yet.
One sign you should worry about. No little read and black Fascist ‘Obey’ bumper stickers this time around. The former era of intimidation is over.