And in the US, the polls are a mere part of the media landscape, as accurate and honest as the journalists who report them. The latest CBS poll, for example, shows a 13 point advantage to the Democrats in its weighting. You believe that you believe anything, but nonetheless this bias does have an effect:
Anyone following the presidential campaign through the prism of media polls is doing themselves a serious disservice. Virtually every one uses a polling sample that is so heavily-skewed towards Democrats that it distorts the actual state of the campaign. Of course, that is a feature, not a bug of the polls. The polls are specifically designed to drive a narrative that Obama is surging and Romney is struggling. Increasingly, though, the polls are having to go to ridiculous efforts to support this meme. Friday’s CBS/New York Times poll, for example, uses a D+13 sample of registered voters. This is absurd. . . .
With a balanced partisan sample, Romney would likely post consistent leads against Obama. A week of this and Politico would run out of fuel for its daily ‘Romney is struggling’ theme. Which is why the media will never adjust its samples. This election, it isn’t so much about polling as propaganda. The polls are simply a tool being used by the media to try to depress GOP turnout and give a powerful lift to Obama’s obviously lackluster campaign.
The polls confirm that the media aren’t really biased. Rather, they are active players for the other team.
16/9 – Rasmussen Swing-State Poll: Romney 47, Obama 45.
Yes, in the states that count.
Token
17 Sep 12 at 12:15 pm
It won’t matter if this happens:
Read the whole thing.
boy on a bike
17 Sep 12 at 12:18 pm
It don’t quite understand what the pollsters are doing in the US.
The interview a sample say 1000 likely voters and tabulate the way they will vote. Why are they fucking around with poll samples and making adjustments.
What does 2004 or 2008 have to do with 2012?
JC
17 Sep 12 at 12:27 pm
I thought the same when I read the toe-sucker talk about the problems with polling data across the weekend.
Token
17 Sep 12 at 12:33 pm
They’re trying to control for sample error, by adjusting to known population values. The problem is that if you make the wrong assumptions you can lock it in, or even increase the error.
dd
17 Sep 12 at 12:35 pm
DD
Why though. They know that if they follow a certain methodology they will fall within the sample error. How does the adjustment improve on it.
And what does population value actually mean anyway?
Are they trying to figure out… finesse who shows up at the booth?
JC
17 Sep 12 at 12:42 pm
This is what the electorate looks like if Romnster gets the undecided.
JC
17 Sep 12 at 12:45 pm
It could backfire. Anyone thinking Romney was going to win will make sure they cast their vote. They risk promoting apathy on their side, but it wouldn’t surprise me if they were manipulating the data.
Twodogs
17 Sep 12 at 12:45 pm
I agree TwoDogs, its like the poll Newspoll today, it gets rid of complacency.
Anyone who wants to get rid of Obama/Gillard knows they can afford to leave to someone else.
Token
17 Sep 12 at 12:47 pm
They know that if they follow a certain methodology they will fall within the sample error. How does the adjustment improve on it.
In theory you can tighten your margins of error this way.
And what does population value actually mean anyway?
Number of people who voted Democrat last time, for example. They know that number for a fact. So they can tweak the sample using this knowledge.
Are they trying to figure out… finesse who shows up at the booth?
Yes, among other things.
dd
17 Sep 12 at 12:51 pm
More from the toe-sucker:
Token
17 Sep 12 at 12:52 pm
I don’t see how that is possibly relevant unless they are matching it to previous comparable data and trying to finesse even more.. And even then I would still ask why is that relevant to the present context when you simply ask people if they are likely voters and then whom they will vote for.
JC
17 Sep 12 at 12:56 pm
I wonder if headlines declaring that Bambi all but has it in the bag causes a higher level of indifference amongst those who would vote Democrat (and the ‘entitlement’ demographic would be looking for reasons not to make an effort) and spurs an increase in Republican voter turnout?
Aqualung
17 Sep 12 at 12:58 pm
DD
I get it I think. What they must be doing is firstly asking people if they are GOP or Dem. They then ask if they are registered/likely voters and then ask if they are going to vote and for whom. That’s the only reason I see why the party affiliation carries any importance.
But I still ask why not simply ensure say the 1000 people sample is randomly selected asking whom they will vote for as likely voters?
I think the party affiliation stuff gets them into all sorts to trouble.
JC
17 Sep 12 at 1:00 pm
Okay, I’m answering my own question I think. They ask party affiliation to offer an extra check on the final impartant question as to whom they vote for.
Okay, I get that too, but I still doesn’t get the tweaking from previous years.
JC
17 Sep 12 at 1:03 pm
I think people often deliberately lie to pollsters. Or change their mind soon after due to being asked to think about the question. A sort of push-polling effect. These sorts of polls and poll results are pretty broad-brush; just another form of electioneering used for population manipulation.
I certainly wouldn’t trust my life to any epidemiological results based on a sampling of a thousand people in a statistical universe of many millions where the sampling had been tweaked and then the results finessed in a manner that could increase the standard error.
If they were serious they would try to get more statistical power with a much larger sample size and more sophisticated stratification taking a more limited statistical universe.
Elizabeth (Lizzie) B.
17 Sep 12 at 1:39 pm
Lizzie
I think the sample they go down to has been researched and shown to offer predictive ability. We shouldn’t be too worried about that.
We should be concerned (for the US anyway) in tersm of what those doofuses are doing to tweak their methods.
I don’t believe they are lying or trying to distort the polls in some way to help the Kenyan. However I do think their tweaks are possibly screwing things up.
JC
17 Sep 12 at 1:48 pm
Say you know the following. I’m pulling numbers out of my arse here, but say that the official record is that 44 percent of people voted Democrat in the previous election. So in your poll you ask a quesiton, ‘who did you vote for lat time’ and your sample is running at 60 percent. You know that’s too high and you’ve oversampled Dem voters and need to weight them down a bit.
dd
17 Sep 12 at 2:02 pm
No, this is push-polling.
“Would you rather vote for the history making current president who caught bin Laden, or a super rich Mormon who’s trying to buy his way into the presidency?”
People do change their mind, but all you can measure is who they’d vote for today, hypothetically, if given the chance. That’s as close as we can get.
dd
17 Sep 12 at 2:06 pm
@ 2.02 Yea DD, I got that part thanks for confirming it.
JC
17 Sep 12 at 2:14 pm
Here’s my take, with an outlook that is different to what I’ve had the last few months.
Obama is probably going to win and the reason why is that the media are practically getting on their knees begging the American people to re-elect him. Notice the similarities between this election and the one we had in OZ back in 2010? We had an unpopular government in power that was presiding over an ailing economy, had screwed things up so much with boneheaded policies and yet through it all, the opposing party just wasn’t cutting through despite in normal circumstances it should have been a cakewalk. I remember reading commentators in the Fairfax (Gittins, Hartcher) and to some extent those in News Ltd. as well (Megalogenis) saying that despite how bad things had gotten, the governing party wasn’t to blame and that we shouldn’t change horses whilst midstream. Well, notice the rhetoric from people in the US is much the same? You’ve got Clinton saying that whilst Americans are definitely not better off, they wouldn’t have been no matter who was in power (despite the fact that he himself and Reagan managed to pull the economy back from the brink from equal and worse circumstances). To top it off you’ve got journos making writing stuff like this
Obama winning hasn’t got me down about it is it should be doing, for the fact that once he gets back in, things are just going to get worse for America, not better (with the fiscal cliff coming and QE3 which is just going to further kick the debt bomb can down the road) and the people are going to suffer mass buyers remorse believing that he lied to them when promising a much better term (Gillard clawing back in with a diminished majority and her famous lie about the Carbon Tax). This is going to further diminish his party and cripple his presidency and also have the effect of turning people off from the Democrats even more. No, it’s not going to hurt Republicans because to react to his bad governing the American people will just punish him even further by helping them cling to power in the Senate and Congress or even drive their numbers up even further.
And then, come 2016 the American electoral map will look so bright red that it will give you seizures.
I could be wrong though and merely reacting to an inevitable post-convention bounce, but my gut feeling says otherwise.
Alex Pundit
17 Sep 12 at 2:17 pm
The one thing I am sad about though is it is going to waste a heck of a candidate in Paul Ryan. Hopefully he’ll recover from this.
Alex Pundit
17 Sep 12 at 2:21 pm
Bloody hell, the urge to proof read when you write something so long goes out the window….
Alex Pundit
17 Sep 12 at 2:29 pm
Yeah JC. Some predictive ability is not necessarily good predictive strength (statistical power) though. Thus polls in some elections turn out to be seriously wrong. These pollsters make do with as small a sample size as they can to get a reasonable level of power (the theoretical concept of statistical power – good predictive ability – is related to sample size), due to cost pressures. Sampling techniques and power and the concept of a statistical universe are at the heart of all statistical theory. I think they then try to validate their sampling by cross-checking it against the actual known population parameters for previous years (in the statistical universe they are postulating). As DD says, this can alter the Standard Error in their current sample, (the SE is theoretically derived from the sample of the theoretical sample means in a given statistical universe) and this change in error parameters can skew the results. It would not be a deliberate distortion, but the technique could still distort, even though it is an attempt to ensure sample accuracy and adjust for that, thus producing a better prediction.
That’s how I understand it simply anyway, although I haven’t really looked in detail at what they actually do. There are lots of statistical ‘pirouettes’ around to massage data, not necessarily fraudulent at all, and often very useful, but in need of explanation regarding possible pitfalls. Current working statisticians may be more enlightening on this than I can be with regard to polling.
My statistical training, probably getting rusty now, comes out of epidemiology and things are much more rigorous there. They go bananas about statistical power and proper sampling because lives may be at stake.
Elizabeth (Lizzie) B.
17 Sep 12 at 2:55 pm
i.e. favoured alternatives – I understand that DD, which is why I said ‘a sort of push-polling effect‘ not an actual push-poll. What I am referring to is when the act of the polling itself (regardless of the poll’s content) has an effect on people, thus bringing a political issue into front-of-mind for individuals who may never have previously cared to think about it or even known about it. Extensive polling itself (for example, on climate change or paid maternity leave) I believe can heighten the political atmosphere around the issue and get attention for it, just as advertising can.
Elizabeth (Lizzie) B.
17 Sep 12 at 3:07 pm
(despite the fact that he himself and Reagan managed to pull the economy back from the brink from equal and worse circumstances).
Rubbish
sdfc
17 Sep 12 at 3:27 pm
It’s all about turnout JC.
Unlike here where polling is easy, it’s all about turnout.
Gerald Ford lost his election in ’76 because GOP registered voter turnout was low.
If voting was obligatory like here Dems would win everything.
Roughly 1/3 are registered Rep, 1/3 registered Dem and 1/3 so-called independent.
The proportion who register with either party or de-register is also a mark of enthusiasm and therefore turnout.
For example hardline leftists may de-register from Dems in digust with Obama for not closing Guantanamo or not introducing single-payer but also some Catholic or moderate Dems may also go independents for his extremism.
The proportions and what percentage of independent likely voters vote Romney or Obumma comes into the calculation for pollsters.
If the pollsters use a D +15 exit polling from 2008 in the Wunderland State to correct for the proportions of likely voters now, the poll will be inaccurate or more correctly not predictive of turnout and the vote in 2012 in Wunderland.
Rasmussen is an automated phone poll and corrects for registration and the usual demographic indicators.
Its simple , cheap, reproducible and predictive in 2008 (excellent) and 2010 (reasonable)
JamesK
17 Sep 12 at 3:39 pm
Rasmussen uses only self-declared ‘likely’ voters which is higher than real turnout
JamesK
17 Sep 12 at 3:41 pm
I’m still very confident of a Romney win and that the win itself will be clear and decisive.
JamesK
17 Sep 12 at 3:42 pm
Interesting piece by Chris Cilliza (lefty but decent guy): Will Obama win in November? Wide gap between preference and prediction.
RTWT
JamesK
17 Sep 12 at 3:51 pm
Hold the line. Romney will win. Independents will fall 3/4 to the Republicans. And we will have the House and the Senate.
dover_beach
17 Sep 12 at 3:52 pm
James
So Rassie goes with self declared and doesn’t use the party affiliation stuff and the rejigging that’s done by others? If so that’s sounds far better and cleaner. I do have an issue with their phone polling though.
JC
17 Sep 12 at 3:56 pm
Now that’s a place with a woman problem.
JC
17 Sep 12 at 4:09 pm
They all correct for sex and age JC, including Rassie.
In other words if the respindent says he./she is not a likely voter the y are excluded.
Post results re-jigging is done on formal voter party registration proportions, I think rather than the last elections exit polling voter party registration statistics.
Gender and age population proportions re-jigging to conform is also done.
But nothing else.
He keeps it very simple.
The three weaknesses that I can think of are it excludes cell phones, voter party registration proportion is not party turnout proportion and even ‘likely’ is an overestimate of turnout.
JamesK
17 Sep 12 at 4:12 pm
The methodology for choosing Democrat-Republican sample percentages is simply based on what people reported to exit pollsters at the last election. That methodology has reasonable predictive power as long as the turnout is taken into account, which might be an issue for Obama this time around – depending on how many dead people, Disney characters and the incarcerated vote for him.
Jack Lacton
17 Sep 12 at 4:46 pm
James K, thanks for your detailed information about what the pollsters are, or think they are, doing.
Very helpful in understanding the processes they are using. As you say, very different to the simple sampling processes and categories used here.
Elizabeth (Lizzie) B.
17 Sep 12 at 5:02 pm
Thanks Lizzie.
It’s far from easy but polling of generic adults or polling registered voters is not worth looking at.
The two nationals of likely voters are Rasmussen (3 day rolling average and automated) and Gallup (7 day rolling average by a pollster over the phone landline and cell).
Both have been reliable but the Obumma Justice Dep’t is threatening Gallup with an expensive lawsuit so now one is too sure of Gallup reliability as a result.
So far they seem plausible with a small consistent Obumma lead.
What’s so amazing is that, at least seemingly, the
Dems gets away with this kind of thuggery.
The state polls are a whole lot more difficult to assess but Rasmussen uses single day method for states intermittently using his usual methodology which are pretty good.
The recent Ohio Marist poll showing an Obumma 7 point spread over Romney is farcical because they are using 2008 exit polling voter party id to correct and a recent Florida poll which did not correct for voter registration at all but had a high proportion of Dems in their sample.
The main purpose of many polls reporting is probably to demoralise the Republicans which in turn will lower their turnout in November.
Pollsters become more honest 2 weeks and less out when their reputations will be made or broken depending on the actual result.
JamesK
17 Sep 12 at 5:24 pm
Business is business, James. The market wins in the end.
I wonder if poor polling really does demoralise and create low turnout. It could also impel action. At the same time, good polling could produce laziness in turning out.
I guess in the States turning out the voters is an artform in itself.
Elizabeth (Lizzie) B.
17 Sep 12 at 5:35 pm
Yeah both those arguments are made in the Chris Cillizza Washington Post article I hyperlinked to earlier.
JamesK
17 Sep 12 at 5:49 pm
sdfc says “rubbish” and then says no more….
Alex Pundit
17 Sep 12 at 6:18 pm
I’ll give you a hint.
http://research.stlouisfed.org/fred2/series/CMDEBT
I’m going out so you’ll have plenty of time to try and explain why it doesn’t matter. Good luck.
sdfc
17 Sep 12 at 6:34 pm
Thanks for that, US polling systems are a bit different to Australian and I wasn’t sure how they were doing that myself.
dd
17 Sep 12 at 7:42 pm
@sdfc
Again, you say nothing. You get nothing as a response.
Alex Pundit
17 Sep 12 at 8:38 pm
sdfc explained in one link
http://research.stlouisfed.org/fred2/series/TDSP
Jack Lacton
17 Sep 12 at 9:16 pm
Alex
You’re the one who made the statement that Reagan and Clinton pulled the economy back from worse circumstances. I have provided evidence that the statement is incorrect. In short neither of those earlier presidents were facing a deflationary environment.
That you are unable to counter is your problem not mine.
sdfc
17 Sep 12 at 9:17 pm
What’s your point about private debt SDFC.
Isn’t expanding private debt an indication that people are able to get credit to finance improved lifestyles with incomes indexed with inflation?
(I’m not ignoring the housing and now education debt bubbles)
Token
17 Sep 12 at 9:30 pm
Yea, the 80/81 recession was really a boom, SDFC. Boom times.
JC
17 Sep 12 at 9:31 pm
Hey, I’m sure some of you guys knew this, but there was apparently slavery in Brooklyn NY in 1898.
If you think of calling bull***t on this, I’m standing by my source – the rep for Brooklyn NY in the US Congress.
How could anyone who was the Democrat’s leading member of the Subcommittee on Cybersecurity, Infrastructure Protection, and Security Technologies be wrong? The Democrats wouldn’t give the position to a person who knows less than Ted Stevens, would they?
Token
17 Sep 12 at 9:41 pm
Debt purchased by banks is what backs the money supply Token. Falling debt outstanding is a signature of a deflationary environment.
JC
The Fed created the 80/81 recession through a deliberate policy of disinflation. Vastly different circumstances to the recent recession.
sdfc
17 Sep 12 at 9:41 pm
SDFC, doesn’t expanding credit indicate confidence by consumers and probably the fact that they are no longer being crowded out of the debt market?
Token
17 Sep 12 at 9:42 pm
So? High inflation recession / low inflation recession, is that your point?
Token
17 Sep 12 at 9:43 pm
Consumers have not been “crowded out” of the credit market. That is not how the monetary system works.
Government debt makes no call on bank capital.
The problem with a low inflation recession and severee financial crisis is that it can deteriorate into a deflationary depression.
sdfc
17 Sep 12 at 9:51 pm
The fed also created the 08 recession, SDFC. Take a look at the yield curve. It was negative from 06.
JC
17 Sep 12 at 9:58 pm
The Fed allowed a credit boom to go for too long by being too slow to raise rates between 2004 and 2006.
In the early 80s there was a deliberate and well publicised policy of disinflation.
sdfc
17 Sep 12 at 10:05 pm
Disinflation is bad?
.
17 Sep 12 at 10:11 pm
That depends Dot. In the early 80s it was a positive though very painful policy.
sdfc
17 Sep 12 at 10:45 pm
Rasmussen Daily Presidential Swing State Tracking Poll
Romney 47%
Obama 45%
JamesK
17 Sep 12 at 11:33 pm
Race42012 Electoral Map: Chicken Little Edition
Obama: 241 (+20) Romney: 206 (-)
Tossup: 91 (-20)
Without Tossups:
Obama: 326 (+29) Romney: 212 (-29)
JamesK
17 Sep 12 at 11:38 pm
Link
JamesK
17 Sep 12 at 11:39 pm
Jack,
I looked at his link, you don’t need to post it again. He merely posted a graph and didn’t exactly explain what he was saying. Just because it included a large dip, what exactly was he trying to say? That is the question. Just because a chart shows a ‘dip’ in something doesn’t prove anything. It can be interpreted many ways, one of which I believe he is trying to say below.
Sdfc,
I thought of the numerous rebuttals to what you said at the time but the fact is that you were too chicken to lay it out at the time and used the excuse of being occupied to say anything (had you had a strong argument, you would have made it at the time regardless instead of just posting a graph).
The fact that household debt is on the decrease in this period has a lot more to do with the fact that homeowners are being released from their mortgages due to foreclosures and are being stuck with the only debt being the difference between the mortgage the owed and the value of their property. This isn’t necessarily a bad thing (in fact, should probably happen more. I know that sounds cold but having less people take on more debt than they can handle can only be good). But this isn’t due to the actions of a big spending president who is convincing the Fed to pump more dollars in to the economy. Also, why then despite the countless pumping of cash in to the system in an attempt to cause inflation not worked?
Saying that private debt has decreased and that is bad is typical Keynesian claptrap.
When Obama took office, the unemployment rate was at 7.6%, average inflation before he got in was at 3.8% and the rate for 30-year fixed mortgage was 6 percent. Unemployment today (that’s the official rate) is currently at 8.1%
When Reagan got in, unemployment was at 7.5% and rising, eventually going up to 10.8%. Inflation was at 13.5% (something tells me you like inflation though) and the rate for a new home mortgage was at 13.7%. The double dip that occurred was painful (caused by the actions of the Fed) but necessary and was something that should have happened this time around. Unemployment by the end of Reagan’s first term went down to 7.3%.
I will concede one point to you though, Clinton probably didn’t inherit as bad a situation as Obama (thank you Bush 41), but Reagan definitely did. That is beyond dispute. The situation now is not worse than the 80′s. What it is true though, is that it’s taking the longest to recover. The blame there though, lies with Obama and Bernanke.
Alex Pundit
18 Sep 12 at 12:23 am
Jack, my apologies, I thought you posted the same link that sdfc did. I think you actually proved my point.
Alex Pundit
18 Sep 12 at 12:39 am
Oh really? You talk like someone who has never had a mortgage before. So when the RBA here decides to lift interest rates, the banks do it as well, just for fun, eh?
Alex Pundit
18 Sep 12 at 12:47 am
And when the RBA has the rates high, it doesn’t make it harder to apply for a mortgage because people can’t service debts at higher rates from the banks?
Again, talking like an undergrad.
Alex Pundit
18 Sep 12 at 12:48 am
Alex
Let’s start with your reference to undergraduate economics first. Uni macro, at least when I was there, was basically a New Keynesian framework where money is an exogenous variable. What I am discussing here is the workings of a monetary economy with a sophisticated banking sector where the money supply is endogenous to the system. In other words money creation is dominated by the financing decisions of banks. Hence a decline in private sector debt outstanding is a signature of a deflationary environment. This was not the case in either the 81/82 recession or the 90/91 recession.
It’s strange that you think that bad debts are somehow indicative of a healthy financial system. Just what do you think this does to financial sector balance sheets? Not to mention the credit scores of the households and businesses reneging on their debt.
The “countless pumping of cash into the system” is a function of the financial crisis as a result of all those bad debts you think don’t matter. The impairment in the transmission mechanism is the result of the crisis which is why monetary policy has been relatively ineffective in generating income growth.
The inflation rate during the early 80s recession is pertinent but not I suspect for the reasons you think it is. The 81/82 recession was a deliberate policy of disinflation by the Fed. It seems to have escaped your notice that the Fed has had the cash rate at near zero for close to four years in an effort to keep the economy out of deflation.
There is no comparison between this episode and the early 80s, for a valid comparison you have to go back to the 30s, the last time the economy was facing anything anywhere near comparable circumstances.
What relevance is the RBA to the US?
As for being “chicken” that’s funny.
sdfc
18 Sep 12 at 7:04 pm
[...] from rightwing eminence grise, Ray Evans, linking evolution and climate science. And here’s Catallaxy pushing poll trutherism Categories: Oz Politics, World Events Tags: Comments (12) Leave a [...]
John Quiggin » Republican conspiracy theory update
8 Oct 12 at 9:16 pm