Interesting bit also in that Marist poll that upon first read gave me pause and then after the pause some reassurance.
Trump versus Clinton hypothetical head to head (yeah too early to mean too much but still) was Trump +7.
The pause. A head to head showing Trump on top? And by as much as 7? Really?
The thought after the pause.
Romney won Indiana by 9. Checking that 538 app Indiana is not just very White, it is relatively very non-college educated White. (National is 37% college educated White and 33% non-college educated White; Indiana is 42 and 41% respectively.) This should be his wheelhouse, where, if anywhere, he does better than Romney did. And at least at this point (with all appropriate caveats regarding current head to heads), he is not.
I agree. The methodology might be sound, but the confidence with which they make the projections should have an asterisk if one single poll will reverse the projection by 64 points.
Further, the “plus” model takes endorsements into account. Well, the IN governor endorsed Cruz, but does Cruz get a full “point” for that endorsement when the governor did all but fellate Trump while endorsing Cruz?
Again, all hail Nate Silver, and disagree with him at your own risk.
I’ve definitely lost a lot of respect for Silver this election season. As for the endorsement, I actually think endorsements generally don’t mean very much anymore.
Is Pence even all that popular in Indiana? If he is, then maybe it means something, but otherwise, I doubt it.
The confidence of the prediction is the probability given. 65% to 69% confidence one way or the other is a pretty low amount of confidence, as is appropriate for having only a fairly small number of polls to go on and their being pretty disparate in their results. Flipping from one to the other when a larger poll from a house with a solid reputation comes seems like what one should expect to me.
The way you write that it sounds like you confuse that shift in confidence with a shift in prediction of the margin. Polls plus average margin is now Trump +3.2 and was Cruz +0.5 a few days back. They are consistently calling it a close race.
Funny enough I’ve gone the other way this season Ibn. I came in bemoaning that it may be end times for polling meta-analysis based on structural difficulties with polling as a process (who answers unknown numbers now?) … but overall their aggregate analysis has done fairly well with only a few major misses (like MI on the Democratic side). Not a review of 538 and their opaque special sauce but Sam Wang did a review of how well aggregate polling did on his PEC blog and found that at least on the Democratic side it has done surprisingly well.
You misunderstand. 65% probability one way, or 69% probability the other … both are expressing a fairly low degree of confidence as to who will win. Each are saying that there is about a one out three chance that this prediction is wrong. They are saying I have more confidence than a flip of a coin one way but not by much, to more confidence than the flip of a coin but not by much of the other outcome. One big poll with a solid result by a reputable house should alter the confidence that much when previous estimates were based on weak data.
The other thing about 538 is that they acknowledge that this is a weird election, and that they therefore don’t know which of their models is better. So they’re offering both, and we’ll see which one turns out to actually be better, and we’ll thereby learn how to be better at predicting the next weird election when it comes along. That’s science.
On the Democratic polling side results are fairly wide, 12 points, going from Clinton +3 to Clinton +15. Yes, all Clinton positive so high degree of confidence in who will win. (92% on polls plus)
The range though on the GOP side? Cruz +16 to Trump +15, a 31 point spread in the polls. Yes most are clustered with Trump up in single digits but any reason fundamentally that the GOP polling should be so much more variable in results than the Democratic side is?
Only thing I can note is that the biggest outlier in each case are IPFW polls, giving a the big Cruz lead and giving the biggest Clinton one. The Cruz one is just a much bigger outlier.
So Hillary in 2016 is somewhat more appealing to non-college educated whites in Indiana than Obama was vs. Romney in 2012? Why would we expect Trump 2016 vs. Hillary to do better than Romney 2012 vs. Obama among non-college educated whites anywhere, and why wouldn’t we expect Hillary 2016 vs. Trump to do better than Obama 2012 vs. Romney among non-college educated whites anywhere?
Trump’s greatest strength is with rural non-college educated Whites, both in general and relative to his GOP peers. His general election candidacy is a doubling down bet on that demographic, on getting such margins and turnout there that he can pull off a win.
Most simply, if he does worse than Romney in Indiana, then he does worse than Romney everywhere.
And that proves me wrong how?
Trump needs about 240 delegates. Between now and California there are these primaries.
May 3, 2016
Indiana
Number of Delegates: 57 Total Delegates
May 10, 2016
Nebraska
Number of Delegates: 36 Total Delegates
West Virginia
Number of Delegates: 34 Total Delegates
May 17, 2016
Oregon
Number of Delegates: 28 Total Delegates
May 24, 2016
Washington
Number of Delegates: 44 Total Delegates
Making 199 delegates available Between now and California making it impossible for Trump to clinch before California.
It is really already clear that he is going to the nominee and that there is no realistic chance of stopping him. His winning tomorrow will be enough to also stop that storyline in the media. With that storyline over the circling of the GOP wagons around Trump will begin … reluctantly, begrudgingly, with some sitting on their hands, maybe, but begin it will. No contested convention, no more overt divisiveness. Sitting on their hands yes, using their hands to choke each others throats, done.
I agree that Trump will likely run the table. Let’s say he gets 1350 pledged delegates. Is it possible that some of those delegates may violate their pledge to deny him a first ballot win? What is the penalty for violating a pledge? Jail? Discounting of their votes?
Rule 16(a)(2) says that their vote will be recorded as for their pledged candidate regardless:
There’s Conspiracy Theory #2,305 going around about how the pledged delegates for Trump (who are against Trump) could just not show up for the first ballot, denying him a majority. I suspect that their vote would be recorded as for their pledged candidate anyway. Certainly denying Trump a majority that way would be burning the place down.