Posted on 10/01/2012 10:03:17 AM PDT by smoothsailing
October 1, 2012
Scott Spiegel
Contrary to popular opinion, which is being grossly distorted by the detestable misreporting of left-leaning pollsters, Mitt Romney is not floundering in the race against President Obama.
Every time you read a story about how Romney must “reverse course” after a “disastrous month” of “Obama momentum,” realize that this narrative was entirely created by biased pollsters and a complicit media seeking to sway election turnout.
Left-leaning pollsters are actively trying to discourage conservative voters and donors, put the Republican ticket on the defensive, and suppress GOP turnout. What’s the proof?
Modern-day pollsters weight their raw sample results to match the electorate by expected turnout for key demographic groups. Almost every poll showing Obama ahead of Romney in swing states by double digits uses a weighting model that predicts Democratic turnout equal to or greater than it was in 2008. Specifically, these polls overweight demographic groups that flocked out en masse for Obama in 2008 and underweight anti-Obama groups.
Rather than assuming that unemployed college students and African Americans are more excited about voting for Obama this year than in 2008, reliable polling agencies such as Rasmussen, Purple Strategies, Gallup, and AP are using a weighting model that predicts Democratic voter turnout closer to what it was in 2004, or to a combination of 2004 and 2008. These firms have consistently found the race to be a toss-up, or for Obama to have only a slight lead.
The issue of weighting is not a statistical lacuna, a minor methodological difference that leads to slight differences among polls without affecting the overall result. In close cases, weighting is everything. It determines the results.
Imagine a state with 5.1 million registered Democrats and 4.9 million Republicans. If both parties saw 75 percent turnout in 2012, the Democratic candidate would beat the Republican 51 percent to 49 percent.
Now suppose that Democrats saw 80 percent turnout and Republicans only 70 percent, comparable to 2008. In this case the Democrat would win 54 percent to 46 percent. But if Democrats saw 70 percent turnout and Republicans saw 80 percent, closer to what happened in 2004, the Republican would win 52 percent to 48 percent.
Assuming only slight differences in numbers of registered voters across parties, as in most swing states, turnout decides results. And most pollsters are relying on an outdated turnout model that enormously benefits Democrats, despite the fact that Democratic voter enthusiasm is down and Republican enthusiasm is up.
To get a flavor of how off-the-mark most polling firms’ numbers are, consider the regularly updated Real Clear Politics average of poll averages. As of September 30, it predicts Obama beating Romney by a greater margin than he did McCain in Ohio, Florida, and North Carolina, three key swing states; and by similar margins in other swing states. Factor out the more unbiased polls, and Obama is outpacing Romney to such a degree as to make the 2008 election look like a nail-biter. In North Carolina, the average of averages predicts that Obama’s margin of victory over Romney will be almost four times greater than his win over McCain.
Taking an average of several polls doesn’t solve the problem, because most polls are making the same mistake. This isn’t a case of some polls being a little biased one way, some another way, so let’s take the average and call it even. It’s a case of almost all polls being hugely biased in the same direction and for the same reason.
Defenders of the insanely optimistic pro-Obama polls claim that even right-leaning pollsters such as Rasmussen and Fox show slight Obama leads. Fine—they show slight Obama leads, not double-digit leads. But pollsters with an agenda know that a lead of a point or two six weeks before an election is not enough to dispirit half the population.
If we’re lucky, left-leaning pollsters’ efforts will be overcome by Romney and Ryan performing well in the debates, September and October jobs reports proving the economy still disastrous, Romney parlaying his monetary advantage over Obama, states implementing new voter fraud laws, and undecided voters breaking against the incumbent as they have historically done.
If pollsters successfully influence voter turnout by suppressing Republican enthusiasm and push the election to Obama, they will have electoral blood on their hands.
It is probably weighted by the fact that more conservatives then liberals hang up on polsters (probably, no one can know for sure since there is no information in a vacuum).
Just poll likely voters period.
Obama is paying the networks massive amounts of Political Advertisement dollars, as long as they give him excellent positive news coverage.
VOTERS ... are not misled
Only if the pollster can require answers. Otherwise there is the likelihood of self-selection of the people polled. For example, I don't answer my phone if I don't recognize the caller ID. I don't have time in my life to explain why I don't want to upgrade my cable, install new windows and donate for homeless squid. My friends know my cell phone number and if the landline call is important the caller can leave a message.
Now that's not a problem statistically if the non-answerers are spread evenly among the political spectrum. On the other hand, if Republicans are all in the basement polishing their guns while Democrats are just begging for attention and answer every phone call, there needs to be a way to remove that skew. That's where the assumptions of D-voters vs. R-voters come in.
The idea of trying to match a prior election "turnout" as a model is only slightly bizarre ~ but it makes little sense to totally destroy the random selection characteristics of the poll when you don't need to.
Let's say you are using data generated by 9% of those polled ~ the situation Pew says is normal these day. If Obama gets just more than half and Romney just less than half, they are splitting up 9%. That could be 4.51% for Obama and 4.49% for Romney ~ and that's what it is ~ statistically speaking ~ not the 51% and 49% you think.
At those levels it is conceivable Virgil Goode might actually be winning with that other 91% who didn't respond!
cannot overcome a statistical deficiency like that.
Exactly. All they can do is assume (pretend) there is no bias in the sample that provides them no information. Clearly that is a bogus assumption.
And then they crow about how accruate they are measured against the vote totals.
Well, there is a 98% chance that the popular vote will be 49% to 51% +- 5%. So it’s not surprising their final polls come out around this point.
Most of the likely voters you can get ahold of are probably on their free Obamaphones.
This would be like a quality control system where you pull every 10th piece to check it for some characteristic ~ let's say HARDNESS.
So, there are the items going by you, and you reach out to select the 10th piece and it jumps out of your way. They continue doing that.
You sit there entertaining yourself for the 2 hours allocated for the test, and you end up with samples taken from every 100th, not every 10th piece.
Your supervisor comes around and asks you where the other 90 pieces are and you say "Uh, they jumped out the way ~ I couldn't grab them".
Besides your boss figuring you for a total idiot, the fact is the 90 pieces missing from the record had a characteristic everybody in the place might like to know about ~ JUMPING SAMPLE PIECES.
You could make all sorts of stories up about why you didn't collect all the samples you need, or collect them in the right order, over the duration of the period of performance (imagine sample selection as a river flowing by and you dip a cup in to take a sample)and do this over a set time).
In fact, what the pollsters have done is design a sampling system that doesn't work, and now they have some results they tell you are "pret'near good'enuff fer'anybody" and they bring out a guy with a doctorate in statistics to tell you about how great this is, and you are a lucky person to be able to benefit from their wisdom and all that great data.
Now, back to that river. Let's say it's our job to detect pollution in the stream. Every hour we dip in a bucket, put a drop under a microscope and count coliform bacteria. We do this months on end and never once do we see even a single bacteria. That river is pure.
A competitor for your job samples the river every half hour and he finds that on the hour there's a pulse of pollution where the water turns brown due to the presence of trillions of coliform bacteria.
So, he gets your job. He's doing great and then you get the really great idea of simply watching the river all day long and sampling only when it's brown.
You get your job back!
That's what these professional pollsters mean by "art and experience". Watch for it.
Yes, every time we do a senatorial or presidential election we prove one more time that the primary effect of a single member district is to bifurcate the electorate fairly evenly into two parties!
"art and experience"
In the end all the so called science is meaningless and you are left with this. Which is proprietary and therefore secret and therefore not available to be scrutinized by an independent evaluator.
The link below is a time machine going back to October 2004 when the maggot infected mediots and their knee pad pollsters were declaring Kerry would be president:
They were wrong then and now!
The media depend on them because they've simply never noticed they're no good!
I don't think President Kerry has gotten over it to this day! :)
“I don’t think President Kerry has gotten over it to this day! :) “
May president Kerry live a long and miserable life trying to figure out how he lost what he hadn’t won inspite of his pollsters.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.