Yes, apparently the person with the most “senior moments” gets the nomination.
Then why does the Libertarian Party remain such a feeble joke cycle after cycle?
538 shows real electoral gains and the Vegas odds are about 2:1 in favor of Obama.
Looks like Nate may be underestimating the race and Romneey is spiraling into the ground.
Crane
Look at it now. 8.4%.
Give today’s fiasco a few days to percolate, and then check the polls.
Fox News has Obama leading Romney by 5 points… and this is before Labiagate*.
Even more interesting is this:
*Sorry, couldn’t resist. I know the “Labia” thing is a joke, but still, it’s damned funny.
I thought you guys considered Fox News polls inaccurate (or, at least, that’s what the highly revered Nate Silver says)? Anyway, at the moment, I’m not too worried since-- aside from the Fox poll-- the polls have consistently shown Romney with sizeable lead among independents. Even the most recent polls-- outside of the Fox News poll-- show Romney with a pretty good lead among independents (14 points here)(11 points here). Assuming Romney stays the favorites for Independents, it boils down to turnout. And if it boils down to turnout, I think Romney will win. You simply can’t win an election when you’re behind with Independents by 10+ percentage points unless either the other guy’s base or those independents who lean towards him stay home.
You sound pretty confident OMG. Would you wager 3:1 odds in favor of Romney? That is, would you give him a 75% chance of victory? If so, you could make oodles at intrade, which places his odds at just under 37%. Personally, I think Romney’s odds are closer to 28%, which is much better than the odds of rolling doubles with a pair of dice. FWIW, that’s my baseline for “Low odds”. So I wouldn’t count Romney out. Anyway, will you take my wholly symbolic 3:1 quatloo bet? Or will you (reasonably) demur?
As an aside, upthread OMG opined that we should examine the House and Senate if we want to know what happens after the election rather than the Presidential point spread. I agree.
According to BOTH your cites, Obama would win the election.
The problem with concentration on “independents” is precisely the same problem as saying “Obama will win because he dominates the black vote” or “Romney will wn because he’s winning by a mile in Texas.” Sure, sure, we know each candidate has particular demographics they’re winning in, but that’s not how the election works. The election works by giving all adults a chance to vote, and then tallying up the EVs. And the polls say Obama is winning in the way that the election is actually decided.
Wanna give me even odds?
I know this discussion was way back at post 92, but maybe this will help. The last picture is an example of overfitting. It correctly “predicts” every single data point fed to it, but the function itself is a mess and doesn’t follow the trend of the data itself. The third picture, on the other hand, fails to predict numerous data points, but is very close to true in most cases, and given the function shape is more likely to be closer than picture 4 for future data.
ETA: Remember that a simple fitting problem is shown, political prediction is much less simple. So they can’t just choose “the obviously right one” like you can probably pick out here.
And? I’d be willing to bet that’s because they’re assuming a higher Democratic turnout relative to a low Republican/Independent turnout a la 2008 (In 2008, Democrats had a +7 advantage). I’d bet good money you won’t see what happen in 2008 happen this year (i.e., far more Democrats coming out to vote than Republicans or Independents) as you won’t see a depressed Republican turnout like you did in 2008. There’s no way to lose Independents by double digits and still win the election unless your base simply shows up in far greater numbers than does your opposition’s base.
That’s interesting, considering for the last two years all I’ve heard is how “independents will decide the election”. I guess that only matters when you want it to matter.
(BTW> The Texas and Black voter quips are complete red herrings, which I’m sure you know.)
Maybe, maybe not. But I strongly believe the polls are underestimating GOP/Independent turnout this election cycle, much the same way this board does (i.e., they consider 2010 an anomaly and believe that turnout will be closer to 2008, which itself was an anomaly).
No, because there’d be no way to test them.
Actually up until the convention all the polls showed a MUCH larger ‘enthusiasm gap’ for Democrats than in most past election. In other words, the gap between registered voter polls and likely voter polls was much bigger than usual.
Unfortunately for Romney, even taking into account the missing enthusiasm, Obama has been leading Romney all along.
Polls since the convention seem to suggest that enthusiasm among Obama supporters is on the rise, which explains his runaway in the polls.
There’s actually a whole detailed post explaining why you’re wrong on 538 today.
I’m not expert, but I don’t think your example applies at all. It’s simply about finding a function that fits a trend.
The issue in this thread is about more than one variable, and a cause and effect relationship. It involves finding which variables correlate the closest to the outcome variable, and how close they correlate, and inferring a causal relationship from that. It has nothing to do with finding a line or curve that fits data. It’s finding how different data fit with other data.
It’s done using multivariate analysis.
You had to dig pretty deep in those polls to find some good news for Romney. I couldn’t find how much of those polled called themselves “independent”, but considering that the “independents” (in the CNN poll at least) significantly favored the Republican Party over the Democratic Party (when asked about the parties’ favorability), I’m not sure how actually “independent” they were. It’s unlikely that these independents are actually the same independents as in 2008 (Obama won 52% of independents) or 2004 (Kerry won 49%) as seen here. Obama’s way ahead in self-described “moderates”.
But what really matters is the electoral map. If Romney can’t win Ohio (which would likely mean he didn’t win states like WI or IA), then he’s very likely cooked. If he can’t win Florida, then he’s cooked almost for certain. Romney is losing in both of those states (FL, which is almost a tossup, is closer than OH). Obama’s campaign has far more field offices in those states. It doesn’t really matter who independents in Texas or California, for example, vote for.
What you’re doing is describing the exact same thing twice and claiming they’re different. And overfit is a problem in both cases.
Jragon, I don’t know if you read the whole history of discussion on this but you’re probably never going to make any progress explaining it.
A number of polls have come in the past week showing that Romney’s biggest advantage over Obama has evaporated in the past 2 months - as to whether Romney or Obama is better at handling the economy and jobs. What was once a solid Romney lead has evaporated at best.
No, I don’t think it’s the same thing.
Maybe overfit is still a problem - I don’t know enough about it to say so - but I don’t think finding a trend line for one set of variables is the same thing as finding how two different sets of variables fit with each other, or which variables fit the best. In the first case, you can pick whatever points you want, as a function.
Is it the same when using each to predict the future? Is that your point?
It’s just a straight-up bet. Election Day is the test.
No, my point is they **are **the same thing. What Jragon tried to do is give you a simplified case that is easier to understand, but exactly the same thing. I tried to do the same a week or two ago without any success.
They aren’t the same thing at all. The problem with them may be the same - I haven’t decided for sure about that - but comparing the correlation of multiple variables isn’t the same thing as simply trying to find a hypothetical curve that fits one set best. That’s just trying to see a trend, it’s not finding a cause and effect relationship between two trends.