Posted on 10/14/2010 12:58:39 PM PDT by InterceptPoint
If you search back to 2008 and look at my models during the presidential election, you will see that I had variant that included a 1% "sampling bias," where I showed the Rasmussen polls as-is, and a "what-if" that assumed a +1 bias correction for McCain and a -1 bias correction for Obama (net +2 differential spread). That model showed that without the bias correction Obama won, and with the bias correction McCain won.
My take-away was that I choose to let the pundits that we deem to be "experts" spread their special sauce, and that I add nothing but pure mathematics to their punditry. I will not augment their predictions with my own predictions of their predictions. I will simply report the probabilistic implications of their punditry.
-PJ
If you're going to throw a question at me, let me throw one back at you.
What do you make of the oft-repeated statement that a 1% or 2% differential in a poll result is a "statistical tie?"
-PJ
Given that most of the pundits are still behind the curve, I expect that GOP pickups will be far, far north of 44 seats. For the sake of argument, you should move every pundit rating one spot towards the GOP (likely Dem to leans Dem, leans Dem to tossup, tossup to leans GOP, leand GOP to likely GOP, etc.) and run the numbers again to see what happens.
I agree.
Republican Pickup Projections
DontTreadOnMe2010 120
Dick Morris 73
Fred Barnes 60
Unlikely Voter 60
RCP 51
Freedoms Lighthouse 50
Sabato 47
NYT-Silver 47
Cook 50+
KHR 44
EP 47
Rothenberg 37-45
CQ Politics 36
I think that people make to much of that “when the House switches, the Senate switches” tidbit; before WWII most House switches were not accompanied by Senate switches for the same party (and one time a Dem House switched to the GOP while a GOP Senate switched to the Dems!), and the probability of the Senate switching has additional factors (such as how the elections 4 and 6 years ago turned out, and which seats are up this year) to those of the House.
Oh it’s mathematical hand waving. It’s a statistical “tie” assuming that the sample they have is a representative sample. Unfortunately for pollsters there’s no way to tell if it truly is without taking a sample of every voter. However, pollsters have been doing this for years and each one of them has ways to correct their results (using a model) for those who are hard to reach.
The sampling error is a statement that there is an unknown but true result (say 45.87498%) and the probability that the statistical result has a (usually) 95% chance of being within the margin of error. For example, say if you asked everyone in the USA if Obama sucks and exactly 45.87498% said yes, he does suck. A phone sample of the population might under report those who says he sucks because they just can’t stand the word “Obama” and hang up. So those willing to answer the question over the phone turns out to be 44.01%. Not lets say you want to have a 95% confidence that your estimate is correct within 3%, you might poll 750 people. Of those particular people, 43.67% may answer “Yes he does suck.” So your poll would report that 43.67% +/-2% of Americans say Obama sucks with 95% confidence.
Note that there is no variation in the original percentage, i.e. 45.87498% is an exact answer. It’s an exact answer but opinion can change. Maybe tomorrow, 44.67492% think he sucks. Now you have a problem. People are fickle and respond to some news and not other news. The number of people that thinks he sucks varies. I believe the trade term is responds to “news” or “events” Events have a lasting effect, news doesn’t. (I seem to remember those terms. They came from a research paper from the Federal Reserve) This, in and of itself, will produce a variation in the trend. Note that it’s not a statistical, random variation. People just change their minds.
Their model to correct their results is a closely held secret. Rasmussen claims that his automated phoning system plus his correction model is the most accurate. Based upon election results, I believe him. SurveyUSA (I believe) has probably the most interesting method, a combination of phone calls and internet survey samples. People have criticized it because a lot of people won’t take an internet survey but SurveyUSA claims to be able to correct for it. They probably do and are able to quickly track trends in opinion. Rasmussen’s polls are at the other end of the spectrum. They tend to lag behind events/news for a week or so.
“The question here is how to take a public opinion poll with a sample result and a margin of error and convert it to a probability of the leader winning.”
That requires a model. Look up how to use the Logistics Regression Function.
43.67% +/-3%
Currently, I think we get nine senate seats, but again, it might conceivably be 10.
In those districts where the Republican had a big lead, that lead is being whittled down. In those races in which the challenger drew even, or near to to even, the money is being used to beat them back down
This is not the slam dunk of our dreams. Euphoria (and confusion) will keep the GOP couch potatoes at home, as it always does. We need turn-out to win. Big turn-out. We also need a commitment to police vote fraud before and esecially after the fact. In the close races, the usual Democrat trick of prolonging the count until they know exactly how many votes they need to create to win, will be much in evidence.
Watch the race in MA with Bielat vs Frank. You will see every play in the Democrat dirty trick book.
There will also be massive intimidation. This will be a provocation, the object of which will be to provoke a violent response. What they want is to see "us against the Communist street." Make sure you get everyone you know registered, and to the polls. This ain't a done deal.
Obama and Co. are already preparing the spin for the aftermath. They’re reinforcing in DE, where they won’t lose to O’Donnell so they can tar the entire TEA Party movement.
They are organized, focused and know their enemy.
Need I post our names?
Same in WA state. That place is crooked as hell when it comes to elections. If the Murray-Rossi race is close, I'll bet a month's salary that the King Co. 'Rats will arrange things so that "just coincidentally", King Co. will be the very last county to report results, and if Rossi is leading at that time, somehow, just coincidentally, I'm sure, he won't be after King Co. reports its results. He may end up "losing" by a hundred or so votes, like in 2004, but the 'Rats will say, hey, it was "just an amazingly close election".
As the Depression deepens in '11, and '12, the Democrats will have the rudderless Republicans as "obstructionist" whipping boys for the 2012 campaign.
I see no Republican Program, No Plan, No Leader, and No Clue on our side of the aisle, for slowing, stopping, or reversing the Obama Administration's drive toward Socialism.
We win the house in 2010. It is a good thing, But BFD. Then what do we do?
You're right about that. If O'Donnell gets blown out, which seems likely given the polling, every mainstream news outlet will lead with the story, and it will be "Tea Party Repudiated By Voters". They'll use a bad O'Donnell trouncing to buttress their mantra of how conservative ideologues might be able to win primaries, but they'll die a horrible and grotesque death in the general election. The O'Donnell-Coons result will be THE story of this election in the mainstream press, even if Angle takes out Reid, or Blanche Lincoln gets beaten by 30 points in Arkansas, or if Joe Miller pulls an upset in AK, or if the 'Pubs win 100 more seats in the House. None of that will matter, because the DE result will be the only one that matters, a "victory" to burnish the glory of The Anointed One.
At a minimum I'm hoping that Darryl Issa and Co will tie up the czars and bureaucrats with enough subpoenas to slow them down until 2013.
So-o-o-o right. Wishing will not make it so.
Glenn Beck's millions are at the ready to turn out.
But we need conservatives lined up BEFORE THE POLLS OPEN Nov 2.
ACTION NOW I hope we are all making an effort to register new people to vote. Each state has a different deadline for registering. So get hopping on this.
-PJ
I'm happy with the results, at least.
-PJ
My fears exactly. My hope is that this time the GOP’ll play chess instead of checkers.
They’ve got to know what’s coming - it’s like a slow train wreck.
They can ride it to victory in 2012 or get picked up by the meat wagon.
You’re right as far as you go. Obama isn’t the lone enemy. It’s actually what he stands for that is - corporatism, the Chicago Way, anti-colonialism, revenge, racism, Marxism, etc.
As you look at the above list think money and the power to get money.
Those that would like to see the TEA Party and its ideas fail are in the shadows still.
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.