Quote: (11-02-2012 07:01 PM)porscheguy Wrote:
And I'm going to tell you to go back and look at Nate Silver's predictions in '06, '08, and '10, and then come back here and continue to disagree. I find it ironic that no conservative took issue with him in '10 because he accurately predicted numerous Republican victories. But now you all consider him to be a part of some great liberal conspiracy because he's making predictions you don't like. And what happens next Tuesday and Wednesday as the results come in and once again prove the accuracy of his predictions. What will you say then?
You can discredit the guy any way you want. But look at his past track record.
Well, I'm not personally invested in the polls - after all, they really don't matter that much, it all comes down to Tuesday.
But I will say this about Silver: he underestimated the Republican landslide in 2010 by +3% points.
"In 2010, the 538 model fared well - but no better than the poll averages at RCP. And that was only after Nate was much slower to pick up on the coming GOP wave than Scott Rasmussen, who called it a lot earlier in the cycle."
No model can or ever will be perfect, and if financial moguls with billions to spend make mistakes, what makes Silver so impervious to errors?
Quote: (11-02-2012 07:39 PM)gringochileno Wrote:
Quote: (11-02-2012 06:51 AM)ElJefe Wrote:
@ Gringo. Agree more or less with your assessment.
Check out this article, though: "A page of history is worth a volume of logic"
I also read this yesterday: 538 is basically a Monte Carlo simulation
I would be betting on an Obama win simply because InTrades has him at +65%, but those two articles articulate the reason why I'm still uncertain of an Obama victory.
What this shows is that most of Romney's chances at this point rely on the assumption that the state-level polls are systematically biased against him by at least 2.5-3.5 points on average.
Exactly! For instance the Quinnipac/NYT poll had Party ID for Democrats up by +5-8 over Republicans. Given that Obama might lose the white vote by as much +20 points (that number fluctuates), if Gallup research is correct in assessing the voter make-up will be identical to 2008, then a 7-point Obama win turns around into a 5-point Romney win.
So all of this hinges on whether the State polls are weighting the electorate realistically. I'm not so sure.
Thanks for offering the bet. The money isn't so important as my dignity, so I'm not going to bet on a Romney win, simply say I really don't think it's in the bag for BO.
Quote: (11-03-2012 03:05 AM)Athlone McGinnis Wrote:
Even Silver himself is refusing to make that claim.
Quote: (11-03-2012 03:05 AM)Athlone McGinnis Wrote:
I have yet to see a more credible and proven alternative model put forward by critics.
Yeah, but just about everyone else on the Left is, as if that's all anyone needed to know. Even you referred to it and assumed that basically settles the discussion. That's what I'm less sure about.
It all hinges on the accuracy of the state polls, and conservatives are hoping this time around they're off the mark (it's happened before), and Democrats are hoping they're right once again.
I haven't look at all the polls, but I've seen Gallup research believes the demographic make-up will be the same, and Obama will lose whites by a greater margin than in 2008 (where he lost by 12, but won nationally by 7). That means the break-even point, if Gallup is right, would be 7 divided by 2 divided by .75 + 12, ie. approx 16.8.
So MR needs to win the white vote by at least 17 to eke out a win. Is that so unlikely? Maybe... probably? We're sure going to find out.