The GOP is still gonna win the Senate.
They thought so last time, too, remember?
I don’t think you can say that with any confidence, by which I mean statistical confidence. One exit poll had Cooch winning 18-24 by 6%, but I think that exit poll had a sample of 143 18-24 year-olds which Cooch won by 8 votes. (64 - 56 with the rest going to Sarvis or no answer)
That sample size is much too small to accept that Cucinelli won that age group as fact, although it does indicate that it is more likely than the alternative.
And I’ll still pity America if they do.
But it did show nice competitiveness. It’s going to be interesting to see what the youth vote looks like in 2014 when we get a bigger sample. My own hypothesis is that Obama has squandered the Democratic advantage with the young.
Perhaps you missed this link earlier. There’s no evidence that your hypothesis is correct.
Well here’s the thing. Both the GOP and the Dems will claim they are going to win, especially in a close race, because a certain percentage of the electorate is SOFA KING stupid that they always want to vote for the party that will win. That’s right, they think that voting for the winning candidate/party is what matters. So both parties HAVE to announce that they WILL win the race/Senate because NOT doing so cedes the complete idiot vote to the other party, and one of the sad truths of democracy is that the votes of ignorant idiots count just as much as the votes of well-informed, intelligent citizens.
So I don’t take any party’s pronouncements that they will win at face value.
I must be brilliant then because I’ve never voted for the winning Presidential candidate, ever.
If anyone is interested. I’ve created a quick and dirty senate senate forecast model in Python. This seemed like an appropriate place to post it in case any of the thread participants felt like playing around with such a thing.
First I gathered democrat win probabilities for all 36 senate races. When available I used the probability given by the HuffPo model (example). When that wasn’t available I entered 0 or 1 as appropriate.
I don’t really know the track record of the HuffPo model. I’m using it because it was the first one I found that had enough reasonably accessible info. Later on when Nate Silver or Sam Wang post individual race probabilities I’ll probably compare them to HuffPo.
demwinprobs =[
["Alabama",0.000],
["Alaska",0.794],
["Arkansas",0.600],
["Colorado",0.673],
["Delaware",1.000],
["Georgia",0.028],
["Hawaii",1.000],
["Idaho",0.000],
["Illinois",0.010],
["Iowa",0.727],
["Kansas",0.109],
["Kentucky",0.082],
["Louisiana",0.325],
["Maine",0.000],
["Massachusetts",1.000],
["Michigan",0.981],
["Minnesota",0.999],
["Mississippi",0.018],
["Montana",0.001],
["Nebraska",0.000],
["New Hampshire",0.980],
["New Jersey",1.000],
["New Mexico",1.000],
["North Carolina",0.773],
["Oklahoma",0.000],
["Oklahoma-2",0.000],
["Oregon",0.995],
["Rhode Island",1.000],
["South Carolina",0.000],
["South Carolina-2",0.000],
["South Dakota",0.000],
["Tennessee",0.000],
["Texas",0.000],
["Virginia",0.999],
["West Virginia",0.001],
["Wyoming",0.000]
]
Then I used the function I posted in this thread since it gives the distribution of different win counts.
def polymethod(winprobs):
answer = [1]
for i in range(len(winprobs)):
temp = [0] * (len(answer) + 1)
for j in range(len(answer)):
temp[j] += answer[j] * (1 - winprobs*)
temp[j + 1] += answer[j] * (winprobs*)
answer = temp
return answer
Then I coded the “main”.
justprobs = [d[1] for d in demwinprobs]
wincount = polymethod(justprobs)
demstart = 34
print("D Seats", "Probability", sep=" ")
for i,w in enumerate(wincount):
print(i + demstart, w, sep=" ")
print()
print("Probability of D control:", sum(wincount[16:]))
print("Probability of R control:", sum(wincount[:16]))
Here is the result.
Probability of D control: 0.3750873676210311
Probability of R control: 0.624912632378969
There are several flaws with this type of model. The main one in my mind is that it treats each race as an independent event when they clearly are not.
Feel free to play around with this code if you are so inclined. Any feedback would be appreciated.
At the othe extreme, one could treat the races as completely dependent, in the sense that the probability rank orders today are the same as they’ll be on election day.
In this case, the Democrats need only win Arkansas (60%) and all more favorable states to get 50-50 control of the Senate. That simplistic model gives a 60% chance to Democrats.
But there’s a big gap between #16 Arkansas (60%) and #17 Louisiana (32.5%).
Even simpler, but much less optimistic, is to note that 11 races are likely D, 19 likely R (ignoring Kansas 11% D chance), so the Democrats need to win 5 of the 6 close races just to get to 50 seats:
[“Louisiana”, 0.325],
[“Arkansas”, 0.600],
[“Colorado”, 0.673],
[“Iowa”, 0.727],
[“North_Carolina”, 0.773],
[“Alaska”, 0.794],
What sort of political ads are running in Louisiana, Arkansas, Colorado, and Iowa?
I can comment on Colorado.
There are many pro-fracking ads. Some these could be described as anti-anti-fracking ads which paint those opposed to fracking as uninformed yahoos.
For the senate race I see ads that say Cory Gardner supported a personhood amendment and I see ads that say Mark Udall was the deciding vote on Obamacare.
The most interesting ad I see is this one telling me that Koch Industries is A-OK. It’s not really a political ad, but it appears that Koch is trying to spruce up their image a bit. I also don’t know where this ad is running besides CO.
Maybe Koch Industries is preparing for the day the conservative wing of the Supreme Court takes its “corporations are people” decisions to the next level and rules they can hold elective office.
Actually, all nine justices agree that corporations are legal persons for the purposes of the 1st amendment. The disagreement between right and left on the court isn’t personhood, an issue that was settled over 100 years ago, but over what kinds of campaign finance restrictions are appropriate.
Sam Wang of the Princeton Election Consortium currently has a 72% chance of Democratic control the senate. He also had an interesting post today defending his model.
He has a pretty good track record, but I still think it’s too early to tell since so many races are so close.
When the guys with good track records disagree, that’s when it’s fun. Although if Wang was truly going “polls only”, he’d have the GOP at better than 50%. The GOP leads in seven seats they need to pick up, although 2-3 of those are virtual ties.
I’m also scratching my head at his assertion that we’ve got a lot of good polling data. On a few races, sure, like NC and KY and LA. But in IA, AK, and NH the data is sparse.
Gary Peters must be confident. He’s publicly backing cap and trade:
We’re also getting Senators not living in their states and claiming fraudulent addresses as their residencies.
In a world where Rick Perry can be indicted for a veto, why isn’t Pat Roberts being brought up on vote fraud charges for registering a donor’s address for voting purposes?
You underestimate the scope of corporate ambitions. Corporations have been declared to be people. Corporations have been declared to hold religious beliefs. Clearly, the next step is that corporations will become our new gods. Heavenly Koch Industries, hallowed be thy logo!
New York Times (the Upshot) latest forecast gives the Republicans a 67 percent chance of winning the Senate, up from 64 percent on Monday.
I like the bet I made.