I did not say he didn’t have a 20% chance, but a 20% chance in an ELECTION is infinitesimally small. He can’t win legitimately with those odds. He just can’t. Elections aren’t like coin tosses. It’s that simple.
Sure it could happen, but there would be a pronounced and obvious reason, something to turn the whole thing upside down. A whole bunch of people drastically changing their minds requires something more or less extraordinary.
No, elections are like dice rolls. Ten sided dice. Rolled twice.
When Nate says 20% odds, he means his model predicts 20% odds. Anything between 10% and 40% is “Close”. 40-60% is “Tossup”. And less than 5% is “Lost, outside of a 100 year deluge”.
That said, the effects of Republican voter suppression and electronic vote tampering deserve to be examined no matter what the results are.
You don’t understand what Nate’s probability forecasts mean then. It’s that simple.
He is not, anywhere, saying that 20% odds means that Romney has no legitimate way to win the election. He is saying that there is enough uncertainty in the data his model has that in 20% of the simulations he runs, Romney wins. That’s 1 out of 5. OK? So, stop it.
A 20% chance means a 20% chance. You seem to be saying that in an election a 20% chance doesn’t mean someone actually has a 20% chance, but you haven’t offered any idea as to why this might be correct, or even what you mean by this: you just posted that Romney has a 20% chance to win, therefore he has no chance. I’m sure that’s not what you mean, but it is what you said, so surely you can understand why a little clarification might be needed.
What do you think it means for a candidate to have a 20% chance to win an election? Do you think that every single time a candidate is trailing in the polls by 2-3 points with under a week to go, he either loses or wins by cheating?
One in five is “infinitesimally small”? Who says? You’re telling me odds that are better than throwing a 4 on one die are infinitesimially small? Gosh, I’ve love to bet against you.
This year Miguel Cabrera of the Detroit Tigers came to bat 697 times and hit 44 home runs. That means Cabera, in any given trip to the plate, had about a six percent cof belting a homer - one third Romney’s chances of winning the election, according to Silver. If you saw Miguel Cabrera hit a home run, would that amaze you as being an astounding feat, one third of “infinitesimially small”? Would you assume he must have paid off the pitcher to be able to do something so unlikely?
Now, if Romney’s chances were, say, one in a thousand (which would have been Mondale’s chances at this point in 1984) then you might have something. But 20%? I’ll take those chances any day. Sure wish my lottery tickets defined “infinitesimially” that way.
On review, I see drew’s point that you probably don’t even understand what Silver’s projection means. Neither does Brainglutton, apparently, and it looks like elucidator’s in the same boat.
It’s sad, really, that almost as much misunderstanding comes from the Left as it does from the Right, with respect to his model and forecasts.
Not even just that. It could happen with the data we have now. Keep in mind that all the predictions we have are based on polling samples - which means they are not perfect. Any polling sample has a probability of error.
Consider Ohio. Weighed together, the polls are saying Obama is 2.8% ahead of Romney. But individual polls conducted in the last week range from Obama having a 5% lead to Romney having a 3% lead. We won’t know which of these predictions is most accurate until Ohioans actually vote. It could turn out that the apparent preference for Obama was an illusory one caused by random sampling.
Reasons why Nate’s model might be over-estimating Romney’s chances:
-
Among serious modelers Nate’s model tends to favor Romney.
-
This isn’t a case where the odds have switched back and forth between the 2 candidates: Obama has always had the lead, both at 538 and intrade. It’s not clear to me whether the odds reflect that. Then again, I assume the Now-cast and the forecast will converge over the next couple of days.
Reasons why Nate’s model might under-estimate Romney’s chances:
-
The turnout models underlying the polls might be mucked. Of course that effect could favor either candidate. But on net, it tends to drive the odds towards the 50-50 mark. I think – not sure about that.
-
Voter suppression might be insufficiently accounted for.
-
Electronic tampering might be insufficiently accounted for.
Been about 16 years since the last time that happened, by my count - and it wasn’t exactly a down-to-the-wire event.
I don’t see anybody accepting a Romney win yet. Typical!
I still think he’s a winner!
OK, it’s been two weeks, and I think I’m ready to accept that Romney had in his hands a real, serious, pile of nigh-unto-hopeless and NATE SILVER KNEW WHAT HE WAS TALKING ABOUT.
What a relief!
Though I’m much happier about other contests, like Claire McCaskill and Elizabeth Warren winning their races.
And even Michele Bachmann’s district making her reelection closer than anticipated, that was nice.
I’m not as freaked out as I was before the election.
Huh. I, for one, am still fucking elated.