Duck Rodgers
Well-Known Member
Not Triumph's best work but still pretty funny:
https://www.youtube.com/watch?v=I46iwLIF1Fo
https://www.youtube.com/watch?v=I46iwLIF1Fo
Obama Might Lose This
- Thread starter Stoked
- Start date
Here's the problem with dumb averages: some polls are really crappy in predictable ways. There was a pennsylvania poll put out about a week and a half ago by the GOP party in Pennsylvania that showed Romney up by 5 points. If weighted equally with all other polls that makes it look close. Of course if we examine that particular polling outfits past performance (they had previously rated Romney +5 about a month prior when they were almost 10 points out of step with every other poll of the region), take into account its general lean through all past polling, and take into account its past accuracy in rating elections (fairly miserable) we don't think of it as reliable enough to actually swing polls. RCP takes that poll at face value because they use the simplest and dumbest process imaginable. In the world of RCP, the quality of your polls doesn't matter at all. What matters is the quantity of your polls.
More sophisticated mechanisms require you to make a lot of sausage with regression analysis and create models for that analysis' ultimate output. Those more sophisticated mechanisms indicate that Romney's chances in both Pennsylvania and Michigan are at or below 5%. Given that we're in that range of probability for Romney, I define that as de facto non-competitive.
Michigan specifically has had odd polling results for months because there are a number of local firms that only poll Michigan that are releasing results there. Those firms are less well known and difficult to evaluate but we've seen swings there between local polls of as much as 18 points. Quite simply, that's not statistical variance, some firms are just flat out doing a crappy job. Quant analysis accounts for that and weights the extreme polls lower. RCP treats it as a tied race.
In essence you are not disputing that the race there is essentially a tie at this point in time. I honestly have no idea who will win Virginia. Nobody does.
I would like the basis for this statement. I would also point out that the type of quant analysis I'm talking about usually is based around much more broad-based election data than the most recent election poll so correction for this type of polling bias is something that's built in to the analysis. Drew Linzer's model, for example, explicitly does not turn on this feature.
To the extent that you're repeating a version of the lines used by unskewedpolls, I will say that you should think very carefully about how Tea Party members who classify themselves as independents will effect your argument when it comes to party identification numbers in polls.
What is the basis for this statement? I believe the best poll analysis indicates Rasmussen has a house effect that's right-leaning of 1.3 points.
I don't think Rasmussen is skewing results or anything like that, but instead has substantial methodological problems insofar as they rely primarily on robocalls to landlines (most Rasmussen polls do not include cell phones) that tend to give them demographic imbalances that are harder for them to correct in-house.
I might have exaggerated slightly. It's probably closer to 1 in 9.
The ads he's running that can be seen in Pennsylvania are for the Eastern Ohio television market. Pennsylvania is not a site of any serious campaigning because Romney is dead there.
I would say that the odds are significantly higher that Obama ends up north of 300 than that Romney ends up higher than 260. The quants are putting Obama somewhere between 280 and 332 as an average depending on model and methodology. I tend to trust the lower end of that spectrum.
On a board that prides itself on intellectual honesty we expect more from you than this...
trigg3r
Well-Known Member
I don't have the time as of this second to argue the points you put forth on a point by point basis. I will try to do that tomorrow. I do however find it funny that members of both parties including yourself seem to try to arrange the polls to support your claims. RCP is now not a good site according to the Thriller because it shows Obama behind in many key states. Rasmussen isn't a good polling organization because it has Romney ahead nationally. Same with Gallup now which has a 50-46 Romney advantage amongst likely voters. I will say that you saying Obama will get to 300 ECVs is ridiculous. In other words, you have Obama winning:
Michigan(Will be interesting to see how this plays out)
Iowa(I think the President takes this even though Romney was endorsed by 4 major newspapers)
Ohio(Not sure why you count this in the win column, looks like a toss up to me)
Pennsylvania(likely Obama's)
Colorado(I actually think Romney takes Colorado, call it a guess)
Nevada(The Labor Unions will win this for the President even with the highest unemployment rate in the country)
Wisconsin(Statistical tie as of now)
Puts the President at 286 ECVs.
Romney winning:
Florida
North Carolina
New Hampshire
You have Virgina as a toss up which either catapults the President to 299 ECVs. I don't see it Kicky. I think you are seeing the Romney momentum continue. Maybe not as fast. Here's a real question, why does Romney currently even have a 1 in 8 chance of winning Wisconsin as you put it? It's pretty easy to see the President's job approval rating is slipping along with his likability factor. He is currently acting like he is the challenger and Romney is the President by focusing on the little things. I am calling it right now for Romney barring a major gaffe. If I am wrong, I will own up to it and say I was wrong.
Anyway, you made me late.... YOU SOB!!!!!!!!!!!!!
The 4 worst words a single man ever hears.
RCP has Mitt's approval rating soaring ever since the first debate.
D
Deleted member 365
Guest
False.
I actually had never heard of RCP previous to your posting of their website.... If that's not a red flag I don't know what is. I think kicky has pretty much dismantled any reason why you or I should trust in RCP and their pathetic records.
You're right, it hasn't been a good polling organization. And it hasn't been for a while now (not just because of Romney). On Nov 1, 2008, they had McCain ahead of Obama in the polls.... How'd that work out for ya?
So why are you trusting in them this time around? What's changed?
D
Deleted member 365
Guest
Food for thought...
In FL, Rass's final poll had McCain winning by 1 point. Obama won by 2.8 points.
In Indiana, theyhad McCain winning by 3 points. Obama won by 1.1 points.
In Iowa, they had Obama winning by 8 points. He won by 9.5.
In Michigan, they had Obama winning by 10 points. He won by 16.4.
In Missouri, they had it a tie. McCain won by 0.7.
In Montana, they had McCain by 4. He won by 2.2.
In Nevada, they had Obama +4. He won by 12.5.
In NH, they had Obama +7. He won by 9.6.
In NM, they had Obama +10. He won by 15.1.
In NC, they had McCain +1. Obama won by 0.3 points.
In OH, they had it a tie. Obama won by 4.6.
In PA,they had Obama +6, Obama won by 10.3.
In VA, they had Obama +4. He won by 6.3.
In WI, they had Obama +7. He won by 13.9.
Rasmussen was wrong in every single battleground state in 2008. This is why I don't believe they are a good polling organization.
Make a case that they really indeed are a good polling organization and that we should trust in their opinion despite the overwhelming evidence I just put forth. I dare you.
In FL, Rass's final poll had McCain winning by 1 point. Obama won by 2.8 points.
In Indiana, theyhad McCain winning by 3 points. Obama won by 1.1 points.
In Iowa, they had Obama winning by 8 points. He won by 9.5.
In Michigan, they had Obama winning by 10 points. He won by 16.4.
In Missouri, they had it a tie. McCain won by 0.7.
In Montana, they had McCain by 4. He won by 2.2.
In Nevada, they had Obama +4. He won by 12.5.
In NH, they had Obama +7. He won by 9.6.
In NM, they had Obama +10. He won by 15.1.
In NC, they had McCain +1. Obama won by 0.3 points.
In OH, they had it a tie. Obama won by 4.6.
In PA,they had Obama +6, Obama won by 10.3.
In VA, they had Obama +4. He won by 6.3.
In WI, they had Obama +7. He won by 13.9.
Rasmussen was wrong in every single battleground state in 2008. This is why I don't believe they are a good polling organization.
Make a case that they really indeed are a good polling organization and that we should trust in their opinion despite the overwhelming evidence I just put forth. I dare you.
In Rasmussen's defense, i'm sure several of those are within the margin of error. Besides, only three of them are outcome determinative.
Empirical results are part of the reason I trust the quants. Silver was 49 for 50 in presidential state predictions in 2008 and 35 for 35 for the Senate elections.
I suspect there's a good chance he'll miss on at least one of Colorado or Virginia when all is said and done because one or both will probably be a coin toss on election day.
Not particularly. Like I said, I suspect the real number will be slightly lower but my point was merely that 300 ECVs for Obama is more likely than 260+ for Romney. Because of issues relating to swing states rising and falling together my point is merely that while I might assign the risk that Romney gets more than 260 at something like 35% I would put the risk that Obama gets to 300 at more like 40%.
I want Obama to win the EC but lose the popular vote. The flip flopping from both positions (where the left was screaming about the popular vote being correct and the right screaming that 'this is the way the founding father's intended it to be') in 2000 would be epic to watch.