Great prognosticator- Nate Silver &538- 75% Dems take House

quote:

---

Great prognosticator- Nate Silver &538- 75% Dems take House

---

This is all but guaranteed now

If I'm a Republican candidate, I'm telling my campaign communications folks to create a montage of the craziest hate-filled rants that Democrats have uttered over the past couple of years. Emphasize the ones where they insult, degrade, and slander Trump voters and supporters.

Buy some air-time.

That should just about do it.

quote:

---

How is this guy still around?

---

He's still living off his 50/50 projection in 2012.

Legit statisticians and data scientists think Silver is a hack.

It still amazes me how upset the Poli Board gets about a statistical analysis. It's based on raw data and mathematical trends from previous elections.
If you read any of his pre-election stuff (which i doubt) he said straight up repeatedly that the national polls were pretty solid but that the state polls were potentially less reliable. His national numbers were damn near prophetic. Some key state polls were wrong.

I think a Dem majority in House is the more likely outcome. There is a lot of history to support this.

Obama and Clinton both saw significant losses in midterms but still managed reelection by healthy margins.

I see the same scenario here.

Dems eek out a majority in the House, misread the results, dramatically overplay their hand and get crushed by Trump in 2020.

quote:

---

It's based on raw data and mathematical trends

---

how is that “raw data” processed? Tell us of your understanding of statistical analysis.

From his twitter:

p.s. The chances our model gives Democrats—between 70 and 75 percent depending on what version you look at—are pretty much exactly the odds that Hillary Clinton had of winning in 2016! So hopefully everyone's learned their lesson and won't mistake that for a sure thing. ??????

That's the point. He isn't going around creating data points. He is taking the data points available, applying an algorithm and then synthesizing the output into a data model. Hillary and Trump were polling within the margin of error. This was a possible outcome.
It's the same reason you can't accurately estimate cost on a project down to the dollar. With all of these projections you build an algorithm based on reasonable assumptions and scientific estimation techniques (typically relying on a combination of analogous situations and parametrics).