Thank you for the link to the Mail article of the Norpoth projection. In 2015UKGE there was an entire fleet of modelers predicting the outcome, but I am not aware of any modelers predicting 2016POTUS, except possibly for Silver. Are you aware of any modelers predicting 2016POTUS and if so, who are they?
There's the "13 Keys" model, and
Abramowitz's "Time for change" model
Others will take as their inputs the polls in June.
Useful as ever, Rod. Thank you very much.
Frankly, the "Keys" are too subjective, and liable to ex post facto rationalization. FWIW, they narrowly favour Clinton, maybe...
Abramowitz has faffed around with his model so much, I'm not such what it's supposed to show any more. Narrowly favours Trump, I believe...
I think you know more about this than I do, but if you recalibrate your model every time a new result comes in, eventually that makes it worse, not better: sacrificing an crude view of the future for a precise view of the past. This is one reason why I think PB should abandon predicting vote share and size of lead and stick to just predicting which one will win.
Maybe. As long as you recognize your model is degrading over time, your Monte Carlos will just cover a wider and wider range of results.
If its prediction is significantly better than a coin-toss, it may still be worthwhile...
The way to stop Trump is for the elite / establishment to ignore him for five minutes. Every time they express how horrified they are by him his ratings go up.
The way to stop Trump is for either Cruz or Rubio to withdraw, otherwise they are splitting the anti-Trump vote. Right now they are racking up a few delegates as they are awarded proportionally, but as of 3/15 all primaries are winner take all and Trump will win in a walk.
Do you think one of them will drop out after next Tuesday?
If Cruz doesn't win Texas, he may well. If Rubio doesn't win Florida, he may well. After Super Tuesday (last time I looked Trump was leading in 12 states, but that may have changed in the last few days) it might be too late to stop Trump.
I'm struggling between Trump and Rubio. My concern about Trump is his temperament. But neither of them could be as bad as what we have now.
From a GOP point of view, Cruz should drop out before Rubio because there's no way Cruz can compete on the west coast and north-east where about 50% of US voters live.
If Cruz wins Texas and Rubio loses Florida, as seems likely, Rubio will of course drop out before Cruz. The latest California poll from January actually has Cruz leading Trump 25% to 23% with Rubio on 13% http://www.field.com/fieldpollonline/subscribers/Rls2521.pdf
Thank you for the link to the Mail article of the Norpoth projection. In 2015UKGE there was an entire fleet of modelers predicting the outcome, but I am not aware of any modelers predicting 2016POTUS, except possibly for Silver. Are you aware of any modelers predicting 2016POTUS and if so, who are they?
There's the "13 Keys" model, and
Abramowitz's "Time for change" model
Others will take as their inputs the polls in June.
Useful as ever, Rod. Thank you very much.
Frankly, the "Keys" are too subjective, and liable to ex post facto rationalization. FWIW, they narrowly favour Clinton, maybe...
Abramowitz has faffed around with his model so much, I'm not such what it's supposed to show any more. Narrowly favours Trump, I believe...
I think you know more about this than I do, but if you recalibrate your model every time a new result comes in, eventually that makes it worse, not better: sacrificing an crude view of the future for a precise view of the past. This is one reason why I think PB should abandon predicting vote share and size of lead and stick to just predicting which one will win.
Maybe. As long as you recognize your model is degrading over time, your Monte Carlos will just cover a wider and wider range of results.
If its prediction is significantly better than a coin-toss, it may still be worthwhile...
It's not supposed to. They attempt to get equal numbers from the major parties (and some from the others), not to get a representative sample of wherever it happens to be held.
It's a mistake often made here, when people say "Wow, all those Tories in Liverpool" or "cor, that's a big Labour contingent in Sussex".
The EU has all kinds of problems, but it's the world leader in fudge, compromise and can-kicking. This kind of prediction has been made about various issues about a hojillion times. It's right up there with '48 hours to save the NHS'.
Comments
If its prediction is significantly better than a coin-toss, it may still be worthwhile...
http://www.field.com/fieldpollonline/subscribers/Rls2521.pdf
https://www.economy.com/dismal/topics/election-model
https://www.youtube.com/watch?v=-n3sUWR4FV4
It's a mistake often made here, when people say "Wow, all those Tories in Liverpool" or "cor, that's a big Labour contingent in Sussex".