September 19, 2018

Nate Silver creates anxiety and allays it.

56 comments:

Dickin'Bimbos@Home said...

Voting Democratic = voting for corruption, lies, and witch hunts.

You will save an average of $2500.00, tho.

eric said...

I've come to the conclusion polling is fake.

It's like psychic reading, or stock market trading. You say general things so you'll always be right.

The market is going to go up today, except that this other thing might cause it to go down. You've been warned!

The Democrat is going to win, give or take 3 points, so he might lose.

I maintain polling isnt about telling us who will win or who will lose. Instead, its purpose is to influence us. Especially business and donors to campaigns.

Hey, nice Facebook you've got there. See those polls? We are about to take power. Would be a real shame if you were to lose it.

My name goes here. said...

And he gets paid for that.

Etienne said...

Nate Silver... I had to look him up. Never heard of him.

Ah... ABC, yep, no, I don't do ABC.

gahrie said...

Yeah the problem with polls the last ten years or so is underestimating support for the Democrats. /s

Joshua Barker said...

After the Dem's performance surrounding this whole Kavanaugh fiasco, I think their already anemic "blue-wave" is going to become more like a yellow trickle down the leg...

Bill, Republic of Texas said...

The walkback is starting. I've noticed over the years that the polls usually show a bluewave in the summer and early fall. The about a month before the election the polls tighten.

I think that's because the media uses polls to try to create momemtum for the Dems and dishearten Repubs.

Then just before the election the polls are more accurate but still under represent conservatives.

Then when people say the polls were wrong the media points to the last polls and say "margin of error." The polls were right!

See the 2016 election and almost every special election since then. Also Beto.

PS can an SJW explain to me how Beto is not cultural appropriation?

rehajm said...

Silver is often blinded by his own leftiness rather than going wehre the data is leading. The data is exposing a myth you subscribe to, Nate: that 'hispanics' are a homogenous, never changing voting block Ds can always count on.

PB said...

Trump has done more for black and hispanic citizwns than Obama.

Nonapod said...

To be fair, he may be correct. But to me it seems increasingly like polls are about as useful and reliable as an ashtray on a motorbike, or a chocolate teapot, or a screendoor on a submarine, or...

mccullough said...

The Red Sox have over 100 wins with 10 games left. Nate said they would win 90. The Nationals aren’t going to the playoffs.

Nate doesn’t know baseball and doesn’t know politics. He got lucky on Obama but has been very wrong on most other things.

Fake Models

AllenS said...

When you constantly poll a lot more Democrats than Republicans, it kinda screws up the numbers.

Known Unknown said...

So Nate Silver admits polling is shit?

The Crack Emcee said...

From my conversations, I get the feeling the Dems are desperate, but demoralized, so they're probably done. There's not enough passion for something, to get them all out to vote, right now. Many don't understand the implications, others, how they got here, and those "on the path" are not going to like where they "find themselves" this time. (Is it possible to, metaphysically, run out of road?)

BTW - Australian Athletes Donned Full-Body Blackface to Impersonate Williams Sisters

And that, too, isn't putting up much of a fight when it comes to how lame it is.









Known Unknown said...

Full-body blackface? Isn't that just blackbody?

Big Mike said...

Isn't that just blackbody?

Yes, and they emit radiation so watch it! (STEM joke. SJWs won’t get it so of course it’s not funny.)

Jersey Fled said...

Show of hands.

How many of you have responded to a poll in the last ten years?

I get polled maybe once or twice a year. I can't remember the last time I responded.

FIDO said...

That is the problem with 'owning' a voting block. Because of all the social pressure to conform, a growing number of dissenters are always going to be invisible, even, or should I say PARTICULARLY to polling.

Trump is rich, ostentatious, manly and he regularly bangs supermodels. He pisses more machismo than John Kerry had his whole life.

And Hispanics tend to be conservative about church, State and particularly gender roles.

So Nate is probably correct.

Michael K said...

The change from land line to cell phones has hurt polling a lot.

Also, people are less trusting of polls since 2000. Remember the exit polls had Kerry winning in 2004.

johns said...

Hispanic and black employment gains have outpaced those of whites by a wide margin under Trump. From August 2017 to August 2018, white employment rose 0.6%. Hispanic employment rose 3.5%, so a gain over five times that of whites. Black employment rose 3.0%, also five times greater. Another strong area under Trump is the reduction in part time employment for economic reasons: the reduction in these part-timers in the last twelve months is 8.4%. A subset of this category is those who "could only find part time work"; this group has declined by 7.6%.
Maybe the message is getting through to Hispanics that this is the way to help people.

Static Ping said...

UNSKEW THE POLLS!

(obligatory)

Char Char Binks said...

Nate Silver, social science denier

Darrell said...

Don't be a fuckhead and vote Democratic.

Static Ping said...

I took a poll recently, but hung up when it became clear it was a push pull for a third party candidate.

Yancey Ward said...

A commenter above has it about right- given the extreme tilt of the US media to the left, the polling paid for and reported is designed to push people in the questioning itself, and designed through the sampling and modeling to push the results towards a position where it looks good for Democrats and bad for Republicans. It has been like this my entire life, but is getting increasingly worse the last 20 years.

Silver does well when Democrats win, and very badly when they don't. This should tell you something important about Silver, polling, and polling analysis, at least if your IQ isn't under 95.

Leland said...

Quinnipac now showing Cruz with a 9pt margin. Democrats outside Texas, please send more money to the Beto campaign; he needs more of it to burn. Beto's the next Obama (or RFK if you prefer), so send that money in today!

MadisonMan said...

BTW - Australian Athletes Donned Full-Body Blackface to Impersonate Williams Sisters

How is an Australian supposed to make fun of Williams?

Charlie Currie said...

Blogger Nonapod said...
"To be fair, he may be correct. But to me it seems increasingly like polls are about as useful and reliable as an ashtray on a motorbike, or a chocolate teapot, or a screendoor on a submarine, or..."

...turn signals on a BMW...

viator said...

"Even as cable news networks debate reports of the existence of a recording of President Donald Trump using a racial slur, a new poll from Rasmussen Reports says that the president's approval rating among African-Americans is at 36 percent, nearly double his support at this time last year." August 16, USA Today

Notice how USA Today frames the lede, thus fulfilling it's propaganda function.
But Trump's support among African American voters has doubled with hardly a peep from the press.

Big Mike said...

@Charlie Currie, your experience with BMW drivers mirrors mine.

wbfjrr2 said...

The problem with the Nate Silver and RCP numbers is that they are averages of apples, oranges, and watermelons. The pools these disparate polls use are widely divergent. Many poll “adults”, others registered voters, and a few poll “likely voters”.

The first of these , “adults”, is totally useless, since relatively few adults actually vote. Registered voters are a slightly better pool, but still overweight a group that still has a high number of non voters. Likely voters are the best pool, but it’s still a crapshoot over how many will actually vote.

Taking an average of all these inconsistent groupings is just lipstick on a pig. Or garbage in, garbage out, as we used to say in the IT world in my day.

Anecdotally, I notice most of my conservative friends boast that they hang up or deliberately give misleading answers if they get called. I would do the same but have never been called, although I’m an adult, registered, likely voter. Have been for over 50 years.

Lastly I suspect that random calling by definition mathematically over-represents large population blue states.

Fernandistein said...

"Vote like a Mexican!"

The Crack Emcee said...

MadisonMan said...

"How is an Australian supposed to make fun of Williams?"

I think it's funny, some don't know, after what "whites" (in other contexts) would claim is "all this time" - it's been 50 years since Jim Crow, right? But they still run to racist shit because they're out of ideas. Or never had any other ones, maybe.

I don't know how this kind of shit keeps happening. I mean, isn't the word out on this one? If not, doesn't it speak to how important our behavior is, when the rest of the world is watching and emulating us?

I mean, sure, I could bag on the Williams' without being racist, obviously, but why would I want to? They're American icons, carrying more than the general public - even the American public - gives them credit for: They willingly carry their race on their shoulders. So, whether anyone else respects that or not, I'm from South Central and we do, so I can't see them - or the worldwide abuse they're subjected to - as a laughing matter.

If I'm laughing at anything, it's with them about being on top, in this environment.

Mike Sylwester said...

Men will vote eagerly in order to express their support for Diane Feinstein and Christine Blasey Ford.

Men like the way that Brett Kavanaugh is being treated.

tim maguire said...

Etienne said...Nate Silver... I had to look him up. Never heard of him.

Nate Silver is a lefty, with all that means for his politics and political ideas. But on the "science" of polling, he's one of the best there is. You can say that's a low bar and I won't argue with you, but ignore his analysis at your peril.

Yancey Ward said...

wbfjrr2 said...

"Likely voters are the best pool, but it’s still a crapshoot over how many will actually vote."

Even "likely voters" is a guess. This is usually determined by asking the pollee if they have voted in the past election cycles. Since being a non-voter is considered to be a bad thing, the pollee has an incentive to lie in response to that question.

In my opinion, the future of polling is the USC/LA Times polling method of 2016- identify a pool early in the cycle, and poll it over and over. You still have the problem with pool construction, but at least have a chance to identify real trends over the cycle.

Kevin said...

If the poll showed a red wave, would they publish it?

We start with our priors and look for data that supports it.

wild chicken said...

But, what do the Data Analytics say??

How many of you have responded to a poll in the last ten years?

I get calls all the time, but usually from candidates or pacs. I always respond see see what I can find out.

Also, I don't see why cell phones are any problem, if that's the number the voters gave when registering. Any polster worth a shit buys the voter rolls from the states. I've made plenty of political and sales calls to cell phones and they answer as often as anyone else.

Landline numbers usu send you straight to voice mail anymore.

BADuBois said...

In others words, we might be right, or we might be wrong.

That'll be one million dollars, please.

johns said...

Polls also apply weights to various aspects such as R and D, because the samples are not large enough to identify all the sub groups that they want to report. The weights can be determined by the percentages of R and D etc that turned out in the last election, but that isn't very foolproof. Looking at the RealClearPolitics polls, Trump has 47% approval at Rasmussen and 38% at Quinnipiac. A lot of that difference is how they weight the responses.

President-Mom-Jeans said...

Why is the homeless man's Jesse Jackson (Crack Emcee) ranting about that sore loser Serena Williams on a post by the Unindicted Career Rape Accessorie's (Althouse) blog post aboput a usually wrong baseball and political prognosticator? (Silver)

Achilles said...

rehajm said...
Silver is often blinded by his own leftiness rather than going wehre the data is leading. The data is exposing a myth you subscribe to, Nate: that 'hispanics' are a homogenous, never changing voting block Ds can always count on.

He isn't blinded by his leftness.

He is blinded because he is a tool owned by ABC and by extension Disney.

He is told what the results of his "analysis" will be.

n.n said...

American?

Oso Negro said...

Sometimes you gwine to be rich, an’ other times you gwine to be po’

iowan2 said...

Polls are not fake. They perform exactly like the people that commission the poll want. Drive media narrative. That's it. The exception is the "internals" commissioned by politicians. Those tell the pols what the people are thinking. That's the reason there has been a reversal on the Kavanaugh smear. The "Internals" tell them, they created a shit storm and its only hitting the Democrat Party. Republicans are pitch perfect in their talking points. By Friday afternoon they will be on to the Presidents hording of the ice cream at supper.

rehajm said...

By Friday afternoon they will be on to the Presidents hording of the ice cream at supper

Anybody remember Woodward's book? Seems like ages ago...

Drago said...

Rehajm: "Anybody remember Woodward's book?"

Woodward?

Wasn't Michael Wolffs book supposed to end Trump before that?

Jim at said...

How many of you have responded to a poll in the last ten years?

Ten years? Try never.
Not even once.

eddie willers said...

I've noticed over the years that the polls usually show a bluewave in the summer and early fall. The about a month before the election the polls tighten.

Everybody knew Jimmy Carter's reelection went down in flames with the smouldering helicopters in the Iranian desert.

But the media had him "ahead" until about two weeks before the election when they had to get real. Yell "October Surprise" and let it go at that.

cubanbob said...

I give Nate Silver the benefit of the doubt and assume his polling is correct. However that means nothing. What does matter is turnout and who turns out. Crack I tend to agree with your comment about Democrats up thread. Regarding Australia, do they still have Coon Carnivals? If they are going to do the blackface thing, might as well do it right (sarc tag on).

JAORE said...

your experience with BMW drivers mirrors mine.

What's the difference between a porcupine and a BMW?

Big Mike said...

@JAORE, the pricks are on the inside.

Roy Lofquist said...

Yancey Ward mentioned the USC/Dornsife poll:

https://cesrusc.org/election/

stever said...

The NYT models rely on an "enthusiasm" turn out to create the blue wave. A model which ignores recent elections and reflects a great deal of wishful thinking.

Bricap said...

mccullough said...

The Red Sox have over 100 wins with 10 games left. Nate said they would win 90. The Nationals aren’t going to the playoffs.


I would look at the overall body of work for how his baseball models have performed. A cursory attempt to google them hasn't had any luck. And I would compare that body of work to what other models have predicted. I don't know the answer as to who has the best prediction model for MLB, regardless.

The Red Sox have overperformed this year. David Price reemerged. JD is raking, and damn close to a Triple Crown. Betts and Bogaerts have remained relatively healthy compared to last year. Benintendi did not have a sophomore regression. Eduardo Rodriguez has exceeded anything he's done before. Porcello has had a big bounceback season. A ton of upside surprises for this team this year.

I don't blame him for getting the Red Sox prediction wrong. I'm sure the simulation has a few iterations that include this possibility, but this is a perfect storm of everything going right for a team. If you think he was way off on this one, you should see how he predicted the A's to perform this year. I would be surprised if anybody had a statistical model that offered this kind of expectation for them.

hstad said...


Blogger tim maguire said...Nate Silver is a lefty, with all that means for his politics and political ideas. But on the "science" of polling, he's one of the best there is. You can say that's a low bar and I won't argue with you, but ignore his analysis at your peril. 9/19/18, 11:39 AM? Yeah, well Silver missed on Trump. Only Professor Alan Lichtman predicted a Trump win.