You're right I'm sure its due solely to the merits of HRC as a candidate!

This is yet another reminder of why they need to get rid of that garbage. They should have never even invented it in the first place, and I say that as someone who appreciates what they were and are attempting to do with it: offer different looks at slightly-different data.

If the election were held today Clinton would win it decisively on the strength of her ground game. Trump's trying to catch up, and he's doing okay in some of the Battlegrounds to assemble a plane mid-takeoff on the runway, but their "Nowcast" paints a ridiculously-untrue picture of how things stand currently. I can't even look at that and not laugh. Trump's "operation" in key states like Virginia, Colorado, and New Hampshire is so hilariously ineffective that it looks like they're organizing for a local sheriff's race in Wyoming. He would need to win one of those types of states (at least) to win the White House, and that means if he is to win he needs to make up a LOT of ground between now and Election Day.

I like the NowCast coz it's useful in its little way, but it definitely needs to be locked up on the high shelf so that children like Alex don't hurt themselves with it.

The reason I find it counterproductive is that it's the antithesis to how elections work, so it gives people the wrong idea of what the campaigns are doing and how. I can understand how people could like to look to it at various points to see "oh look at how this lead grew or changed over the span of the past x days/weeks, what a turnaround!" but that is just flat-out not how campaigns and elections actually work.

Opinions are not decided, let alone shifted, in that way, and the overall share of the electorate that is actually susceptible to changing their mind in ANY fashion is incredibly small. Oftentimes the movement that drives changes in public polling is caused by response bias (certain demographics becoming more interested in or available to pick up the phone and being willing to sit through a 20 min. survey, vs. others becoming less interested in doing it) and nothing more, or a combination of response bias and methodology disagreement/flaws between polls. It leads people to think there's these back-and-forth swings happening that aren't actually happening.

Today is different than two days ago. Hillary Clinton handily defeated Trump in the debate. Just pointing that out.

The point two days ago was that Trump had been doing well in recent days/weeks. He had a good two weeks in the polls. It was his first lead in one of their models since his post-convention bump and before the DNC convention.

I wasn't commenting on my prediction for the election, the strength of each ground game, the model itself, and etc. I was simply making the point that two days ago marked the closest Trump had ever been to Hillary in the polls and various predcition models in quite a long a time. Just to keep things in the proper scope and context here.

Edit: Fuck it. It's not worth my time.

This post was edited by ThatAlex on Sep 28 2016 10:50am

The debate could have been an inflection point, but people should be advised it could take an eternity to find out exactly what (if any) effect a debate and a debate alone can have on opinions: You have to let all surveys that were in the field before 9/26 be released, you have to not overreact to surveys that contained one/multiple pre-9/26 days as well as post-debate days, you then have to wait for surveys that were conducted entirely post-9/26, and then you compare those to that first group of pre-9/26 surveys (and neither of the other two groups). And the problems I just wrote about re: response bias and methodology quirks that have come to make public polling close to useless still apply, so you could be doing this for little to no reason anyway.

People can make the point that Trump has seen his fortunes improve since his post-DNC meltdown by using 538's other models, if they're going to incorporate 538 at all. They show his late August/early September gains, and do so in a way that's much more representative of what's actually going on in the ground than the "Nowcast." There's a larger argument that contends that almost nothing has really changed since the conventions, but if someone were to make the argument that Trump's September remedied his awful August, they're better off using the other model than the Nowcast.