Trump leading the polls when factoring in polling bias

inquir.io/2016/08/05/donald-trump-leading-new-polls/#.V6ONw6KUZ8E

>CTR shills on suicide watch

Other urls found in this thread:

longroom.com/polls/methodology/
fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/
projects.fivethirtyeight.com/2016-election-forecast/
theblaze.com/stories/2012/08/23/electoral-college-model-predicts-romney-will-win-big-in-2012-and-its-been-right-since-1980/
fivethirtyeight.com/features/is-donald-trump-blowing-it/
twitter.com/SFWRedditGifs

lol, "polling bias"?

You haven't noticed that the trump train is crashing and burning?

>LongRoom is also giving the average polling bias to Hillary Clinton with 4.2% being biased in favor of Clinton. It is easy to see how the polls are skewed in favor of the Democrats given the number of Democrats sampled in a majority of these polls is much larger than the number of Republicans.

The CNN poll not only heavily oversampled Democrats, it literally removed the entire 18-34 age demographic.

hi payed shill!

>literally and unabashedly "unskewing" the polls after what happened in 2012

You people never learn

Here is the actual poll result at the moment. Hillary did indeed gain a lot of ground from the convention and Khan controversy, but shes still behind Trump on average.

By what metric, may I ask, are they determining these poll biases?

the amount of democrats polled vs republicans. one poll had ~800 democrats but only ~700 republicans, CNN also likes to use as little independents as possible because trump leads among them. their latest poll had ZERO people from the 18-35 demographic, because they know hillary does terrible with young people.

Not zero, just not enough to statistically accurately show breakdown percentages within the 18-35 demo. That is what N/A means on a poll.

like i said, zero.

Of course they poll more dems, there are more democrats than republicans. Splitting it 50/50 or doing a simple random sample does not accurately describe the population with the number of data points they are taking. Thats why they use stratified random sampling based on models instead

Here's how you can tell this guy is full of shit:

>barely intelligible writing
>no equations, numbers or any actual details about methodology

longroom.com/polls/methodology/
longroom.com/polls/methodology/
longroom.com/polls/methodology/
longroom.com/polls/methodology/
longroom.com/polls/methodology/

Wow, this thing is even more garbage than I initially suspected

>the number of Democrats sampled in a majority of these polls is much larger than the number of Republicans
maybe because there's more democrats than republicans in general ?

There's 4% more democrats than republicans you auto-buttfuck math queen

it's crashing and burning if you only follow liberal publications and the words of shills

There are more democrats than republicans and of course the pollsters take that into account when doing the numbers.

>mfw I don't give a shit about polls because the butthurt will be sweeter when Trump wins and libtards go "b-but the polls!"

Enjoy your rigged and biased polls, though.

The current sampling models do not take primary turnout into account. The +/- D/R sampling ratios are the same as they were in 2008 and 2012 despite the fact that the Dem turnout was -20%

>2012: unskewing polls
>2016: unbiasing polls

I'm gonna go ask President Romney how that works out.

The difference between registered Republicans and registered Democrats is about 3%. Most of these polls sample Democrats at +10% or more

Registered democrats outnumber registered republicans, but not by the numbers reflected in the sample numbers. There are far more democrats polled than there actually are. Furthermore, Republicans are more likely to vote overall than Democrats in most elections, and in this one in particular.

There's no point in arguing with them. They've cast their lot with this conman and are desperately trying to validate their decision in the face of every sign pointing to Trump self-destructing.

Just like the bernie supporters, these people never learn

>Hey guise! Guise! I realized that if you pretend there are more Republicans than Democrats, than the Republican nominee is more popular than the Democratic nominee. Isn't that something guise? Guise?

It's usually the opposite and 2008 and 2012 were hugely out of the norm

There is no statistically significant correlation between primary and general turnout, so why would they take it into account?

I looked at it, it looks fine. Fuck off shill.

What looks fine? The part where he doesn't actually describe how he calculates any of his numbers?

I'm not saying he's just making shit up out of his ass, but I have no reason to think he is credible or even knows anything about stats.

What makes the other polls more credible?

The liberal media bias is so terrible today, America probably has the worst propaganda in the world. And the bad kind, not the cool kind.

For comparison, here's Nate Silver describing his methodology:

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

The poll aggregator in the OP predicted Obama winning when the other polls showed otherwise.

Except you can't verify that they're registered democrats or republicans. Polls can only say if someone is LEANING democrat or republican.

If Trump does something that makes people less likely to identify as republican, then it would make sense why polls show more democrats or independents as opposed to republicans.

those polls look retarded, its not the same polling site, the one in the OP has correctly predicted the last few elections.

...

The "Unskewed polls" guy factored in polling bias as well in 2012 check out how accurate his prediction was lel

When dead people and mickey mouse and donald duck are voting then it doesn't really matter what hte fuck the polls say huh CTR?

Is 0.9% a good lead?

kek

No, especially when it's not even true.

projects.fivethirtyeight.com/2016-election-forecast/

A poll has to lay out how the numbers are calculated and test for faults. You usually try and adjust the polling results with different well known correlations that can compensate for low sample size.

For example, polling via land lines means you will get older age people. This wouldn't be an issue if old and young people had the same voting patterns, but they don't. So you can try and increase your sample size, which is expensive, or you can use past data on voting bias to adjust what you polled.

That is one example, but these pollsters are doing a major disservice by not doing this statistical adjustment. (Ie, only showing 19% of population are independent, where as it has been +50% for decades.

The model apparently predicts past elections. The thing is that anyone can make a model that can predict things that have already happened. They tend to be extremely inaccurate when trying to predict the future.

Just look at the model that predicted a romney win based on 5 past successful "predictions". All the models that have actually predicted future elections all point to Hillary.

Is it possible that more democrats are sampled because there are more democrats in the country?

Nobody trusts (((Nate Silver))) you fucking kike.

yea but so has nate (((silver)))

>paid shill
>payed shill
>payed
>pajeet
I should have known

Why are independents so underrepresented when they're, by far, the largest voter base in the country?

Exactly. Interpolation is pretty easy. Extrapolation not so much

Of course not user, it couldn't be that simple. Its much more likely that Clinton/CTR/illuminati are actively conspiring against Trump

>when factoring in polling bias
What did he mean by this?

(((polling bias)))

Thank you for correcting the record.

How do you explain Fox News then?

>.05 cents have been added to your account

>longroom.com/polls/methodology/
Fuck off shill. It makes perfect sense and they accurately predicted the last 3 elections within 0.3% margin of error

>This faggot is a shill who merely posts lies to keep you from looking at the methodology and deciding for yourself.

18-34 is massively pro-Hillary you idiot. So if that was factored in, she would probably be +10. They're only including pro-Drumpf polling bias so that you racists don't go killing yourselves when you see how DOMINATED you are getting. Utah, Arizona, Texas (YES FUCKING TEXAS) will all go blue. Drumpf is unqualified to be President.

That article is bunk. Some other user noticed in a previous thread that the website was most likely only set up last year, meaning they've only back tested their techniques and this is the models first real test.

theblaze.com/stories/2012/08/23/electoral-college-model-predicts-romney-will-win-big-in-2012-and-its-been-right-since-1980/

i was just reading this
>fivethirtyeight.com/features/is-donald-trump-blowing-it/
>natesilver: I just want to resist the interpretation that any of this is coming from a rational part of Trump’s brain, whether as part of some brilliant 12-dimensional chess strategy, or even a strategy at all really.
nate are you posting in this thread?

to be fair though this goes to show how big of a faggot romney is

longroom.com/polls/methodology/

it predicted the last few elections even though some polling places had romney to win

No one is being paid to shill for Hillary by the way. It's a Republican-funded lie and a pathetic attempt to smear her campaign.

The model hasn't actually predicted those elections. It's just backtested like the model I posted.

Like I said, backtested models aren't very accurate, especially considering it was only tested on three elections. The model in the link I posted "predicted" every election since 1980 and said Romney would win by 300 electoral votes. It ended up being spectacularly wrong.

Four years after the fact. This website is less than a year old.