Post-election opinion of 538

[Angel of Death]

On one hand, they did get 5 states and a CD wrong, but on the other hand, they were better off than most by hedging their bets and not all-but-guarantee a Clinton victory, which got them a lot of abuse, especially here. In poker, that would be like minimizing a loss after getting a bad hand, which is every bit as much a primary skill to have.

[100% pro-life no matter what]

Honestly, of the major sites, only RCP was better.  They did a fine job, and, if they are honest, a 28% chance should happen...28% of the time.

[JohnCA246]

Yeah they did about as good as they could do in the general election. If the polls are wrong their models are going to be off, but Silver repeatedly said that Trump was a "normal polling error" from winning. The one thing I would offer criticism on was that they seemed to significantly weigh old "reliable" polls over a slew of newer ones. So there seemed to be a lag on who was winning specific states over the election, whereas a simple aggregate seemed more current.   

[Xing]

Eh, their methodology still needs work. It's just clear now that this is true of every prognosticator.

[Make My Bank Account Great Again]

It's gone up.

Between 538, Sam Wang's Princeton model, and Huffpost, 538 is the only one that has any credibility left in my eyes. They actually wrote a piece a couple days before the election warning everybody that it only takes one polling error for Trump to win and pegged him at a 28-29% chance the night before.

538 really dropped the ball during the Republican primaries. All the polls were clearly indicating that Trump was gonna be the eventual nominee, yet they put punditry over the data and said thst his chances were much lower than the polls suggested. If I remember correctly, I believe that they put out some BS article with the headline that "Trump is leading the polls but losing the nomination" at one point.

Thankfully they corrected this mistake during the general election.

[mencken]

The least hackish of the hacks.

[Nym90]

They failed in the primaries because they didn't have a model and were just winging it.

The points they raised about Trump's negatives and the ceiling to his support were valid, but they were too quick to assume that "the establishment" would coalesce around an alternative.

[RI]

Up since the primaries, though still down from 2014-ish.

[🦀🎂🦀🎂]

Data Journalism has destroyed political journalism + politics in general in America, and 538 is the worst offender.

[GlobeSoc]

Data Journalism has destroyed political journalism + politics in general in America, and 538 is the worst offender.

[Skye]

Thry got the GOP primary totally wrong, but the general election redeemed them.

[Meclazine for Israel]

They had Trump up at a 36.8% chance and climbing with under a week to go.

Credit for that compared to other sites.

They just didn't model the Russian influence.

[PresidentSamTilden]

Up, they had this thing pegged realistically when everyone else was drinking the sweet complacency tea. Not just on atlas either...my sister told me Nate Silver was an unskewer and Huffpost model was better. In real life lol. I knew that was crazy.

I enjoy their podcast also, and they post some good climate change related stuff.