2018 Midterms

[4/5]

Just tell me Nate Silver was wrong and I'll be happy with any of the results.

Tragically right

Huh?

FiveThirtyEight was wrong in all their "Lean D" Senate races and a couple of their "Likely D" Senate projections.

Nate Silver can suck a lean D

WHY OH WHY DIDN'T HE SAY THAT DEMS WOULD TAKE BACK THE SENATE!!! HE SPOKE TEH TED CRUZ VICROTY INTO BEING.

[tragabigzanda]

FUCK ICE

[4/5]

Yeah, Silver explained what he does well there. He's not making predictions, but instead establishing odds and ranges. And sometimes the lower odds and fringe ranges win.

But isn't the net effect basically "Here, let me mathematize and then commoditize the fact that no one has any fucking clue how elections will turn out." The whole "light/classic/heavy" model is a crock.

I guess you're not a fan of math. That's fine I guess.

[washing machine]

Trying to figure out if the country is mad at Texas about Beto’s loss or not.

As upset as I am about the Beto effect costing Judge Emmett his seat, I think it’s fantastic that a gerrymandered district that has been red since 1966 went to Lizzie Fletcher this time around.

I wonder if Beto’s run was Trump-era fluke or a genuine sign of something brewing in Texas. My gut tells me
It’s a fluke, but I’m curious what the rest of the country (and RM) thinks.

[tragabigzanda]

FUCK ICE

[Green Habit]

FiveThirtyEight's map had AZ, FL and IN "Lean D" and MT "Likely D".

Looking at the high/low ranges they established, the only one that they failed to contain was IN. (Low of 46.3% of Donnelly, who looks to be only getting something around 43%.)

[run2death]

Honestly, Silver's time has past.

He use to lean heavily on predictive models from past elections, but now that we all know you can no longer do that, he's just pulling shit out of his ass.

He kinda says a bunch of nothing nowadays.

[Simple Torture]

FiveThirtyEight's map had AZ, FL and IN "Lean D" and MT "Likely D".

...based on polling. They do try to estimate, using their model, which polls have partisan leans, which have been more accurate in the past, and so on--and those parts of the process involve some judgment calls. But they are not calling races based on a gut feeling or the dominant media narrative.*** We seem to have to go over this every fucking election.

***The caveat here is that during the 2016 GOP primaries, a number of 538 writers--including Silver--dismissed the polls and numbers that were showing a Trump surge and wrote off his campaign early. Silver has written extensively how he fucked that up, and it sounds like it's one of his biggest professional regrets.

[tragabigzanda]

FUCK ICE

[tragabigzanda]

FUCK ICE

[Simple Torture]

Like 8 people posted before me, but I do stand by what I said. 538's models are just one tool that I think can be helpful in understanding a political climate, but obviously they shouldn't be the only thing you read / 100% what you put your stock in.

[tragabigzanda]

FUCK ICE

[Green Habit]

I wonder if Beto’s run was Trump-era fluke or a genuine sign of something brewing in Texas. My gut tells me
It’s a fluke, but I’m curious what the rest of the country (and RM) thinks.

Ice cold take: a little from Column A, a little from Column B. And throw in a bit of Column C in that no one really likes Ted Cruz, even if they vote for him to keep GOP policies going.

[run2death]

Yeah, Silver explained what he does well there. He's not making predictions, but instead establishing odds and ranges. And sometimes the lower odds and fringe ranges win.

But isn't the net effect basically "Here, let me mathematize and then commoditize the fact that no one has any fucking clue how elections will turn out." The whole "light/classic/heavy" model is a crock.

I guess you're not a fan of math. That's fine I guess.

That's a fucking cop-out.

The math has to based on something with merit and it's clear past models are no longer valid and exit polling is next to useless.

[run2death]

Also like to point out that I was right about 538 leaving ESPN.

[4/5]

Honestly, Silver's time has past.

He use to lean heavily on predictive models from past elections, but now that we all know you can no longer do that, he's just pulling shit out of his ass.

He kinda says a bunch of nothing nowadays.

exactly. And then others go an re-tweet his nothing to bolster their own worldview.

What worldview is he bolstering? That you can statistically model polls so that they have pretty good predictive value and that while polls are inherently imperfect because they're only asking a sample of the population instead of the entire population you can use them in conjunction with prob and stats and turns out they are usually actually pretty decent at measuring opinion at a given time?

[McParadigm]

Trying to figure out if the country is mad at Texas about Beto’s loss or not.

As upset as I am about the Beto effect costing Judge Emmett his seat, I think it’s fantastic that a gerrymandered district that has been red since 1966 went to Lizzie Fletcher this time around.

I wonder if Beto’s run was Trump-era fluke or a genuine sign of something brewing in Texas. My gut tells me
It’s a fluke, but I’m curious what the rest of the country (and RM) thinks.

It would have been nice to have one of the three possible fresh faced red state democrats forced into the party leadership ranks by voters. Current leadership is stale and oddly naive about the politics of the day (I honestly think Pelosi isn’t half the problem that Schumer is, but she’s a woman and so appears to be a better foil for Republican messaging). Beyond that, though, a lot happened in Texas that wouldn’t have without Beto’s impact, and that’s not nothing. I’ll be curious if he ends up as a VP nominee.

In the end, I suspect none of us will be taking about Texas and all of us will be taking about Florida in two years time. Florida...and the rust belt states that made Trump president and just elected Democrat governors.

[tragabigzanda]

FUCK ICE

[4/5]

Yeah, Silver explained what he does well there. He's not making predictions, but instead establishing odds and ranges. And sometimes the lower odds and fringe ranges win.

But isn't the net effect basically "Here, let me mathematize and then commoditize the fact that no one has any fucking clue how elections will turn out." The whole "light/classic/heavy" model is a crock.

I guess you're not a fan of math. That's fine I guess.

That's a fucking cop-out.

The math has to based on something with merit and it's clear past models are no longer valid and exit polling is next to useless.

I don't think the evidence of last night gives very much to support this scorching hot take. Or, for whatever evidence from last night that does support it, there is much more evidence suggesting the opposite.

[Green Habit]

Here's the deal with election models: they're going to get built by the parties no matter what, so they can figure out where to prioritize spending resources. The parties aren't going to make the entirety of their models public for competition reasons, and the little they do reveal is going to be drenched in partisanship. So we need models built outside the parties to get impartial information out to the public.

Silver and 538 are by no means infallible, as they're human like anyone else. But for any of their faults, they've done an excellent job on balance in bringing election modeling out of the shadows.