Election Probability

How are the probability calculations for the american presidential election made?

It obviously cannot be calculated theoretically and it also cannot be calculated experimentally (i think), because it's a single experiment that has never happened and that will never happen again.

I'm a Statistics undergrad and i'm reaasonably fascinated about how these probabilities are calculated.

Other urls found in this thread:

youtu.be/42ZGoUNcD7M
youtu.be/z2eS218ES6E
youtu.be/mTTbMCGv2ZY
youtu.be/X1Hs4vgmyWQ
youtu.be/3SoBUZl3Hoc
youtu.be/t-TvSa9Dcvc
youtu.be/iNz-Xs0UBZE
youtu.be/snou8qrElnM
youtu.be/yosAVMB47-Y
youtu.be/_K1-nzxzzug
youtu.be/ZYXygIcIJ6I
youtu.be/6bY_QmA9Vu4
youtu.be/qPOvePioFF0
youtu.be/78n2vsIM5CQ
youtu.be/zRogXTFtpUc
youtu.be/DkePW5PgPrM
youtu.be/o10-Aqjifa8
youtu.be/vARcHv-nZ-c
youtu.be/SPh_raqO_A0
youtu.be/u0BYqzdiuJc
youtu.be/LmWOOrwDUug
youtu.be/lDUwXFvTJfA
youtu.be/YY-CiPVo_NQ
youtu.be/tHsZxJlxHYw
youtu.be/bJecWvOTIT8
youtu.be/6z_m9twHoDE
pastebin.com/wHQsijFy
fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/
github.com/fivethirtyeight/data
fivethirtyeight.com/features/how-the-fivethirtyeight-senate-forecast-model-works/
fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/
twitter.com/AnonBabble

(OP)

Spamming these top-tier Trump videos which were made before the election so that it will be safe for YOU to share with normies. This is a reaction to the pussyleaks video and aren't meant to be a redpill as much as a way for YOU to soften Trump's image to women and introduce the real Trump. SHARE THIS AND MAGA!

SHORT
>(1m) Trump signs sign for fan
youtu.be/42ZGoUNcD7M
>(2m) Trump refuses to lower flag
youtu.be/z2eS218ES6E
>(2m) Trump fixes Wollman's Ice Rink
youtu.be/mTTbMCGv2ZY
>(2m) Actress in Trump pussy video talks about the Don
youtu.be/X1Hs4vgmyWQ
>(2m) Trump flies ill child in his jet
youtu.be/3SoBUZl3Hoc
>(3m) Trump sends money to hero bus driver
youtu.be/t-TvSa9Dcvc
>(3m) Dying Ms. Wisconsin thanks Trump
youtu.be/iNz-Xs0UBZE
>(3m) Mini Trump
youtu.be/snou8qrElnM
>(3m) Trump cameos
youtu.be/yosAVMB47-Y
>(3m) J. Jackson Praising Trump
youtu.be/_K1-nzxzzug
>(4m) Trump interview on Sept 11
youtu.be/ZYXygIcIJ6I
>(5m) Trump helps disabled girl
youtu.be/6bY_QmA9Vu4

LONGER
>(9m) Never Give Up
youtu.be/qPOvePioFF0
>(9m) Donald Trump LOVES Americans
youtu.be/78n2vsIM5CQ
>(10m) The real Donald Trump
youtu.be/zRogXTFtpUc
>(14m) Rebuild Twin Towers II
youtu.be/DkePW5PgPrM

ASSORTED
>(1m) Girl cries over Trump
youtu.be/o10-Aqjifa8
>(3/5/1m) Fallon interview
youtu.be/vARcHv-nZ-c
youtu.be/SPh_raqO_A0
youtu.be/u0BYqzdiuJc
>(3m) the video that sold me on Trump
youtu.be/LmWOOrwDUug
>(4m) Deplorables Unite
youtu.be/lDUwXFvTJfA
>(4m) The biggest recorded fuck you
youtu.be/YY-CiPVo_NQ
>(5m) Trump the Establishment
youtu.be/tHsZxJlxHYw
>(5m) Donald Trump's Christmas Book
youtu.be/bJecWvOTIT8
>(8m) Black Pigeon Speaks
youtu.be/6z_m9twHoDE

Pastebin
pastebin.com/wHQsijFy
SHARE!

> NEVER EVER GIVE UP!!!
> MAKE AMERICA GREAT AGAIN!!!
> TRUMP 2016!!!!!!

they're not calculated in an honest fashion. Many of these polls contain wording bias and an oversampling of Democrats

wow, he's back up to the 20s

There's s comprehensive methodological summary on the same page where you got that image. A first year stats major should be able to follow along.

>now under 80%
fucking kek

First of all, you need to realize that 90% of analysis are wrong because pollsters have no idea what they are doing and are making unreasonable assumptions. Got it? Ok.

So you take each poll in each state. You're obviously not polling every single person who would have voted there's some margin of error. What's that margin? Well, what are the odds that the people you randomly polled happen to really like Trump? Do some bayesian math, and you get a standard deviation around your results.

Calculate the probability distribution for a win in each state, sum them up, weight by electoral votes and you've got a relatively bad guess.

nervous man sweating buckets

Clinton. Due to rigging she will win key counties in key states.

Vegas fag here, I'm starting to think we'll stay bule. I can't voice my opinion because both at my work and uni it's the same buzzwords tossed around about Trump. Unless the silent majority is real here I don't see how Trump can win Nevada

>Hillary is sub 90%

GET READY TO MAGA BOYS

Literally at the bottom of 538's forecasts they have a link to how they do it.

fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/

Why is Florida blue? Isn't he doing very well there in early voting? I've heard he's doing much better than the republicans usually do

I can't see Ohio going for Hillary either

Step 1) Decide who you want to win
Step 2) Mangle the statistics until your preferred candidate has a huge lead

Ive been screencapping posts like this. Im going to post a collage on post-election day. Just like I did for R-Money. Just like I did on Sup Forums For McPlane

He just said at the rally they are leading 4 points in florida

Its blue because Nate accepts +12D sampled polls the same as he accepts a +4D poll

And if Trump wins?

Trump just said at his rally

>WINNING FLORIDA, NORTH CAROLINA, IOWA, OHIO

ABSOLUTE MADMAN!!

I've started my screencaps too. Look at my first one. The 9th is going to be glorious.

Trump is doing good in FL but he's winning because Hillary can't seem to break over 160k votes over him that are needed to even break even.
He is, that's not a lie.

Odds at predictit are traded with money. Money trading is the only sensible way to determine odds. People usually do not want to lose money, so they try to objectively analyze odds.

REMINDER THAT BREXIT ONLY HAD A 3.4% CHANCE OF WINNING

Top KEK. Consider yourself collaged

Trump Landslide.

Put in the collage now, even though you are going to delete it on the 8th.

55-45 Florida for Trump.

A lot of big money was put on Remain and smaller bets on Brexit. Betting odds aren't indicative.
Now really, what happens if he wins? Would you die?

Some cuck takes a lot of money from a particular group and shits numbers out his ass.

Bout it.

Lol. Most accurate summary I've ever heard.

>A lot of big money was put on Remain and smaller bets on Brexit. Betting odds aren't indicative.

Doesn't work for predictit. There is a 1000 dollar limit per person per trading question. You cannot manipulate it except by creating many fake accounts just to manipulate the odds. Not sure why anyone would do that, as predictit isn't reported much on.

His message has to be:
>if you vote, we win.
>your vote is the most valuable thing in this country, and with it you can change the world.
>double check to make sure they don't steal it from you and let's all take back the white house!

>What if he wins

Mostly nothing. Worst case scenario is he's a slighty offensive version of W Bush. No matter the outcome of the election, my life will be fine.

Good to hear, reckon it's likely he'll win PA? Seems pretty much like a must win

You're assuming what anyone not ctr on this board will tell you is not true, which is that nate formaline is not a complete hack.

I haven't seen him explain his """adjustments""" so I presume they're just his wittle feelies

What if it actually happens?

Any chance Trumps odds go down again? I've been waiting to place another bet, but idk if there will be a better time than now.

Good to know.
PA will be hard to flip but definitely possible. Would be easier if Trump manages to flip NV, NH and ME though.

yet they go to vegas and blow their savings all the time
Little hint to everyone too that was brought up in brexit and somehow isnt now, the bookies payout rates are also based on how many bets were made for any particular candidate
if everyone is betting on hillary, whether they support her or not, the payout will be lower regardless of actual odds so they still make money.

Most likely not.

His adjustments are literally wishful thinking.

Damn nigga you retarded

>Texas
>red

>500 dollars is alot of money

Damn nigga, you're poor.

Feels

Screenshot this.

Nope i capped your posts, will post when trump wins then youll feel irritated

Ill post them around this time make surr youre on here

>mn red
xd

Challenge accepted nigger

I have this capped for a potential Trump victory

Oh, he has no chance, but do you still get money when he loses?

>Indiana only around 50
>Texas only around 50
Sure fucking thing Shillary.

>when trump wins
Top kek m8
He'd have to take Florida, Ohio, North Carolina AND Pennsylvania to win. He doesn't stand a chance, especially in Pennsylvania.

The assumptions used matter A LOT, and there's no way to get better assumptions about a future event. I mean you could continue to ASSUME that the future event will play out similarly, or maybe even identically to past events of that nature. But shit, I mean you're even making raw assumptions about SAMPLE SIZE and shit. I mean, the first Stats class you ever have states that you start every analysis by defining the sample, and that's the most important assumption that you make.

TL;DR There isn't a way to be terribly accurate with demographic assumptions about voter turnout.

I would say the uncertainty in the results of the polls are much higher than what's reported, because their uncertainties are CALCULATED, based on quantifiable aspects of their analysis. But there is NOT a number to factor into uncertainty for assumptions like enthusiasm and "willingness to register for a major party." I mean, you could qualify it, but it's a representative number, not a measured number. Sort of like the Pain Scale doesn't have any units. Because it's subjective/made-up bullshit.

>putting NV, OH, IA, and FL all in blue
>States where Trump is literally winning right now

God who invited fucking Brazil to talk about American politics?

Are those bots posting those tweets? It's word for word..

Sometimes the pollsters just aren't aware of what is happening.

The reason why Clinton was +20 in the Michigan primaries was because the pollsters did not realize that Sanders was popular with young voters.

>iowa blue
>even rcp average has it +3.5 trump

fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/

Obviously its theoretical

yes i'm sure that if trump loses all his supporters will come here just to get humiliated by shills and shitskins

You're a statistics undergrad and you're asking fucking Sup Forums?

What the fuck man?

Okay, i read the 538 user guide, and it's bullshit.

You can read it and still not understand how the calculations are done, because what's in there isn't a complete, detailed methodology. There is too much stuff being simply referenced, but not shown/detailed.

I want to see the math, the data, and the program, but specially the math.

I asked elsewhere but it's hard to get answers.

Here i get answers, although most of them are low quality or not even answers.

Try this: github.com/fivethirtyeight/data

It's not open source. You should not expect them to show you everything. Quite the opposite, indeed.
Still, they do say a lot. If you know what they are talking about, you can replicate their results.

>doesn't know the difference between editorials and polls

the last fucking article even explains exactly that

Thanks. Will check it at some other time.

>currently wiining all those states, even if you presume he does get the proportion of independents even the (((polls))) says he has.

[citation needed]

except for cases like predicit it where most of the money betting is based on the odds given by the polls.

Doesn't, fuck you shitty android, I am going back to my blackberry.

What the fuck are you learning if you can't comprehend this.

You take representative polls, have some kind of error and a certain distribution around your mean value. From there you just simulate the election.

Reminder that this is absolutely not true.

yes

very few people actually understand what probability is

>didnt even read the file name

what's the website, I want to place a bet on Trump.

Sup Forums really needs to do more, less than 2 weeks now until the election

Bovada is the only one that takes Americans.

They receive the correct numbers by mail from CTR.

Well exactly. So even if a poll is as honest and skillfully done as possible, there's no way to have all the information, and so the result is always, at best, a guess, and at worst, an advertisement.

if you did statistical modeling in any class you'd know it's a bunch of bullshit filled with loads of assumptions. stop accepting what nate silver says as gospel.

the model is outlined on his website: fivethirtyeight.com/features/how-the-fivethirtyeight-senate-forecast-model-works/

i read that a few years ago. it's literally as generic as it gets. of course, he doesn't define parameters, or distributions

>certified MS in statistical techniques of the GLM variety

in case any autists are lurking: fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/

They aggregate a bunch of polls, weigh them according to how they personally feel about them, make a ton of assumptions, then run simulations.

So how are internal polls different from the polls we see?
Are they any more accurate, and if they are, how?

>HURR MONDALES HOME STATE VOTED FOR HIN THAT MEANS IT WILL NEVER EVER GO RED REEEEE

fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/

Right here you fucking retard.

Personally I think both the betting markets and polls are correct in a sense.

If it was a regular election or if you expected the same turnout as in 2008 then for sure Hillary would be a massive favorite.

The problem is that this is not a typical election. This is the exact opposite of that.

Let's not forget Trump holds the record for most votes ever by a Republican in the primaries. And that was also in the biggest field of candidates in the history of the GOP.

Just putting those two facts together is mindblowing and a glimpse of what sort of popular support Trump has.

We're talking people who have never voted in their lives or who haven't voted for 30 years. Then there's also the shy Trump voter who can't express his opinion for fear of social shaming and marginalization.

The problem is there is no scientific way for polling agencies to account for this. Their options are to take vague guesses and expose themselves to risks or just play it safe and use old turnout data. They chose to play it safe.

There will for sure be a huge turnout for Trump. I don't know if it will be big enough to win but it will be a close race. Way closer than polls or betting markets have it.

If you ever plan on betting on Trump now's the time to do it, because his odds are only going to shorten as the election approaches.

Shit odds, what site is that?

Weird how suddenly Trump is doing well in the polls days before the election.

It's as if the polls are rigged and the media is trying to save their reputation post-election when people find out that Trump actually had a better chance than what the polls said.

I'm a burger so I got stuck with Bovada. I placed that bet before the first debate, but Bovada matched that deposit, so I really only had to put up 250 dollars.

Bovada still matches deposits but only 50% of them. Their odds are a bit worse than the other sits, but better than most when you consider the matching deposit.