Trump's got 42.9% approval at 538. Ya wanna know why?

[crazytown]

We’ll one reason is that Nate Silver keeps Rasmussen Report in the mix, an outfit even more fair and balanced than Fox News. 51% of Americans approve of President Trump according to Rasmussen.

RR 18-20/11 Trump 51/48% approval
Cf
18-20/11 YouGov 40/49%
17-20/11 ARG 42/55%
14-20/11 Ipsos 44/51%
14-19/11 Quinnipiac 41/54%

Yes, Nate, discounts and adjusts their results. After reweighting RR’s adjusted results areTrump 48/51%.

Ramussen is sewerage. RR’s are updated every 2-3 days, are ‘inexplicably’ counter narrative (if other polls go down, it goes up) and disproportionally dominates Registered Voters.

I’ve heard Nate justify leaving Rasmussen in with a ‘Both Sides’ argument. No good enough. If the evidence mounts that a survey is not just biased, but a croc of cooked up lies, it’s time to give then the boot.

https://projects.fivethirtyeight.com/trump-approval-ratings/

[chowder66]

1. Even if he had 80% approval, he is still rotten and anyone who supports him

is not paying attention or rotten themselves.

[MBS]

20. I feel exactly the same way.

His approval rating should be about 4%, not 40. After 2 years of this nightmare, how could ~40% of our fellow citizens still think that this vile and incompetent human being is fit to be president? It just boggles my mind. SOO depressing and frustrating.

[lamsmy]

2. Its fair

538 looks at how the polling organizers collect and calculate the data. Rasmussen's methods are not consider best practice.

However, Rasmussen was mostly right on the 2016 election results - better than the the predictions from the top flight pollsters including 538.

You may hate the results, and NS may discount their methodology, but he also understands that the Rasmussen results should not be dismissed.

[crazytown]

3. Tell me 51% of Americans approve trump.

Tell me this is writhin legitimate sampling (or even methodological) errors.

[uponit7771]

9. We need what's right more than what's fair. Nate is playing "both-sideism" by even

... including that hack of an outfit.

Rassumussen didn't have a better hit miss on the state level than other outfits which would've given Americans a better view of how the election would've gone.

and justifying keeping them in the running because they were correct 2 years ago is pretty thin.

Science shouldn't be left or right

[DemocratSinceBirth]

11. I spend some time on the predictit boards.

I have a one variable model that is highly predictive.


I simply take the results from the Rasmussen poll:

http://www.rasmussenreports.com/public_content/politics/mood_of_america/generic_congressional_ballot_nov05

and add nine points to the Democratic total.


That simple one variable model would have predicted the House popular vote this election.

[jcgoldie]

12. 538 is not even a pollster

...And Rasmussen was not "mostly right in 2016.

[DemocratSinceBirth]

13. They were nine points off on the Northam-Gillespie race

http://www.rasmussenreports.com/public_content/politics/elections/election_2017/virginia/election_2017_virginia_governor

[Denzil_DC]

4. That's Silver's methodology, frustrating as it sometimes is.

He keeps even skewed/unreliable pollsters in the mix, but weights their results accordingly.

Whether he weights them enough is another matter. No amount of weighting will counter percentages cooked up out of thin air.

[crazytown]

6. It keeps Trump's approval rating artificially high.

trumps approval is trending lower then ‘inexplicably’ RR posts 50% then 51%. It stinks.

[uponit7771]

10. Is his weighting system of RR public? tia

...

[Denzil_DC]

19. This is the most recent I can find:

https://projects.fivethirtyeight.com/pollster-ratings/

It looks like they have banned 6 pollsters. RR gets a C+ - though there have been suggestions in the past that RR's polls skew more realistic in the final stages of big races.

[CrispyQ]

5. There's no way 51% approve of Trump.

No fucking way.

[Stallion]

7. You Can Take Out Rasmussen on Pollster and It Generally Causing Approval to Drop by About 2 1/2%

[crazytown]

8. Good to know. Thank you

[Wintryjade]

14. There use to be a time if a president fell under 50% that was bad. Trump goes over 40%

and he is doing well.

[Garrett78]

15. It seems fair to say his approval rating is right around 40%.

As I keep saying, what really stands out is how steady his approval rating has been compared to his predecessors. Cult45 is real. It simply doesn't matter what he says or does. I don't think economic conditions have much bearing on his support or lack thereof. This is a cult driven by racism and sexism.

[Awsi Dooger]

16. You are adjusting in the wrong direction

Trump's effective approval rating is considerably higher than that 42.9%, not lower.

That is the mistake made around here repeatedly. The focus is always on Rasmussen, which is adjusted downward 4-5 points. Yet Nate also adjusts many other pollsters lower, and he adjusts quite a few of them higher.

The net adjustment is wrong. That was verified once again in the exit polls, which had Trump at 45%. RealClearPolitics came much closer than Nate Silver simply by raw averaging. They were at 43.6% on election eve while I believe Nate's model was at 41.9%. As I mentioned, it came in at 45%.

To be fair, Nate had mentioned many times that Trump's approval rating among likely voters is higher than among overall sampling. That partially accounts for the difference. But it does not equate for a full 3% gap between Nate's model and what the exit polling indicated.

Nate's models are still a work in progress, as we saw on election night with the House model fluctuating all over the place. I am not a big fan of adjustments, as learned the hard way in Las Vegas. That's why I use a derisive term Happy Adjuster for someone who wants to mold the world into anything he wants it to be. In this case, several friends and I from the Las Vegas wagering scene agreed that RealClearPolitics' non-adjusted approval model would probably be closer to reality than Nate's heavily adjusted version, and that's the way it played out.

Looking forward toward election 2020 it is very foolish to take Nate's number at face value toward where Trump really stands with the people who will decide his fate.

[Kingofalldems]

17. Trump just lost about 40 seats in the House.

[Rizen]

18. Base Trump's approval on the election

We took back the House with gains not seen since Watergate. Republicans were washed out. There's your approval ratings.