Good points on the 2012 Nate Silver model. I remember four years ago we were all discounting it here. However Nate's model is scary accurate because he was correct in 49 out of 50 states in 2008 (missing one state by only 0.1%). Then he was 50 out of 50 in 2012.
In 2012, Nate had Romney at 14.3% just prior to the first debate and then, due to the strong debate performance, Romney improve to 38.9% but it was steadily downhill from there. By Election Day, Romney's chances were pegged at 9.1% (see graph below from 2012). The live link is here if you want to take a closer look at this and other graphs from that year.
This year, I believe Nate is "holding back" on us, trying desperately to preserve the "horse race" - and keep his audience in suspense. But in the end, he'll get it right so that his reputation is intact. He does seem to have a pretty reliable predictive model here. However, it probably won't be until much closer to Election Day that we see him move the states required to push Trump over the top, even though by then, it should be fairly obvious to the rest of us that Trump has this won.
Looking at 2012, he added 13.5% to Obama in the final few days. We will likely see a repeat this year only Trump will be the beneficiary.
Thanks for the post.
I would add that large moves at the end are also appropriate—there is a chance that something can be pulled out of a hat and change momentum, but the less time remaining the more immediate and total the change has to be.
Silver may have his finger slightly on the scale early on, but I doubt that he has it on at all at the end. I think he really is a stats guy first and a liberal second.
I’ve actually had the page you linked to open in another tab for several days. I’d be really curious to see the equivalent for 2008.
I watched Silver fairly closely in this election cycle when a Never-Trumper early on told me he was better than sliced bread.
He's best at getting it correct making his final projections that are within 24 hours of an election. About all of us can do the same and be just as successful when the tea leaves are very easy to read. When Silver goes off the rails and in la la land when he has to prognosticate early spouting the usual leftist talking points feeding his followers of Dems what they want to hear.
Little Silver said Trump had a better chance of playing in the NBA finals than winning the Republican nomination. Well, Trump should play in the NBA too LoL. Silver thought or still thinks getting endorsements in this election cycle was the way to election nirvana. Silver was talking Yeb! v. Hillary - counting who's getting the "big or bigger" endorsements from some politico or from some Hollywood brainless, which was up front on his website's home page.
However, it probably won't be until much closer to Election Day that we see him move the states required to push Trump over the top, even though by then, it should be fairly obvious to the rest of us that Trump has this won.
Got to keep'em on the 538 plantation as long as possible.
Silver's flaws in his "methodology" is he can't or did not read or understand what the Republican electorate was seeing comparing candidate vs. candidate. Intangibles and qualities: leadership, smarts, likeness, ability, wisdom, record, truth, gamesmanship, energy, achievement, willpower, communication, job history, and winner...et cetera. It was obvious which presidential candidate had the edge early on. Trump is a new proto-presidential candidate who has exposed the flaws in relying on skewed and flawed polls that are essentially done by Democrats who run the liberal colleges and poli-sci departments, and a liberal media whose purpose is to defeat the Republican candidate as they are hysterically afraid.