I'm sure everything is on the up and up. Do you not think for a second there could logically be nefarious reasoning behind the decisions to do it? They are creating a false narrative with the desire of swaying the low info crowd into giving up.
Poll oversamples?
- Thread starter PriddyBoy
- Start date
I'm sure everything is on the up and up. Do you not think for a second there could logically be nefarious reasoning behind the decisions to do it? They are creating a false narrative with the desire of swaying the low info crowd into giving up.
I used it several years, but probably did not understand if you are correct. I am pretty sure that you are misunderstanding. "If you design the pool you want", you are prejudicing the outcome unless are choosing the pool via random selection of the entire population.
Simple two choices red and black with one of each, you assume drawing a red is 50:50. Correct? Now selectively add 5 black. The odds of drawing the red has changed to 1:6. Use the same for any quantities you have in the population you want that are representative of the population. If you add 5 to any subset, you change the statistics of occurrence in the general population.
I will give you the same response that Virginia Cross, accounting instructor at WVU, gave me. If you do not understand that, there is not a damned thing I can do for you.
You're assuming they're weighing correctly when increasing the sample size. And we're also assuming poll respondents are treating these polls like they would have a Romney or McCain, where there was no shame in admitting support.
I think you could argue whether or not the Clinton campaign would want to skew polls. On the want side, it may dissuade Trump voters from turning out because they may feel like their vote won't matter. On the down side, it might make folks with a Clinton lean decide to stay home since their vote isn't needed.
It's all a moot point in my line of thinking. Pollsters are nerds who make money from giving accurate results. It's not in the pollsters interests to skew things one way or the other at the request of anyone. They want to be able to point to their polling results, compare them to the actual results, and be able to say that they made an accurate prediction. If the campaigns were running the polls, I can see how they may want to skew things - possibly. The campaigns can make whatever suggestions they want to pollsters though. Have you ever tried to sway a mathematician or a statistician about something related to their field when you aren't in their field? It's a hard conversation to watch. I've watched several of those. I've been part of a few of them too.
I think you're right that a skewed poll could both help and hurt a particular campaign. Even Nate Silver blew his prediction on Trump in the primary, but not out of any desire to move the needle.
What's completely ignored in this discussion however is Podesta's motivation for his email.
The arithmetic isn't hard. If I want to base this on a pool of 1000 people, but I want to decrease my error bars on certain segments of the population, I can pool more folks from certain segments - the example was 500 African Americans to get a more accurate result than 136 - then I reduce all of my individual segment results to 1 person (divide by each sample size) then multiply each by the proper number per 1000 people to extrapolate the results to the overall population. It's not rocket surgery - it's a simple weighted sum. This isn't a tactic that is new to polling either. It's been done this way for a long time. How have polling results worked over the last several elections?
Another point is that several people/groups have been doing poll averaging. Nate Silver is the most famous of the folks, but he's not alone in the field. You essentially are doing the same thing again with poll results, basically increasing your theoretical sample size. Those folks have done extremely well in the last several elections.
That's not the same thing, not by a long shot. You are talking about random draws from a single fixed set. This is random draws from several fixed sets, normalize the results, then combine. I could poll 1000 people from each representative group, normalize the results to the correct proportion of 1000 people, and give you accurate results. I can still do that with random samples.
If I wanted to break it down to your example, split the red and black balls between 3 bins: A has 10 red, 20 black, B has 20 red, 10 black, and C has 20 red, 20 black.
If I did several samples from each bin, I converge on the true result. Look at it from the perspective of selecting a red ball from all of the balls:
P(red) = (1/3*30 + 2/3*30 + 1/2*40)/100 = 1/2.
I'm not disputing the polling method. I completely understand how polling works, and as much as I hated statistics I did manage get through the class. What I'm suggesting is that the possibility exists (not saying it's actually happening) that it's not being done correctly.
From 538's own blog:
"Certain types of polls are not assigned a weight at all, but are instead dropped from consideration entirely, and not used in FiveThirtyEight’s forecasts nor listed in its polling database. from the firms Strategic Vision and Research 2000, which have been accused – with compelling statistical evidence in each case – of having fabricated some or all of their polling, are excluded."
No matter how one feels about sampling and even if no pollster is taking Podestas advice the fact remains that he is telling the media how he wants the polls done and why.
He's not telling the media how he wants the polls done, he's telling an internal polling company to oversample more democratic subsets so that the campaign could pinpoint ads (media).
Last edited:
Campaigns rely on polls to know where to travel and spend their time and money. They want polls to be accurate.
Campaigns pay for polling that you never see. They quote media pols that are favorable to them to shape opinion.
Thanks for proving to us again what a moron you are. Would you like to wager 500 dollars on the election outcome since you are so convinced Trump will win?
Where did I say trump would win moron? I realize it is difficult for a simpleton like you to be told how things are but dont flatter yourself by believing your little brain can speak for me.
I'll take that as a no. Look at the thread; you are the one accusing the Clinton Campaign of fudging numbers. I think you have also chimed in on the "media conspiracy".
You will take what for a no? You said it not me. If you cant see a media conspiracy you are more stupid than a lot of people know you are.
It has nothing to do about the actual election results. Damn, just when I think you can't be any more ridiculously dumb, you prove me wrong.
Once again, it's about being able to sample and hype select statistics to shape a false narrative. Christ, it's like dealing with my sister's retarded kid when seeing you on this board.
If it's such a false narrative then Dave will gladly put his money where his mouth is, or do you want in on the action?
![[laughing]](http://l.yimg.com/j/assets/img/emoticons/classic/laugh.r191677.gif "Laughing [laughing]")
It is funny. You want it so bad but at the end of the day you are who you are. A Marshall grad.
I would explain the logical fallacy of your point but you will never understand.If it's such a false narrative then Dave will gladly put his money where his mouth is, or do you want in on the action?
I'm not the one making claims I can't back up. You're an idiot. You claim the polls are bogus but won't accept the bet.
What claim did I make that I cant back up? Where did I say the polls were bogus? I stated just the opposite several times. It is not my fault that you cant understand words.
Neither of those quotes suggest that polls are rigged. Once again you are simply clueless and too stubborn to quit while you are behind.
I can let you off easy, all you have to do is not respond. Both of those posts clearly demonstrate you think the polls are bogus, "they quote media polls that are favorable to them to shape opinion". But when you were challenged by a bet, you puss out. You are just like Trump; you try to say you didn't say something but there is proof to the contrary.
Nothing in thise posts suggest that polls are bogus. It is common knowledge that campaigns pay for their own internal polling. The media pays for its own polling. When a media poll result is favorable to a candidate they will tout that poll. Even if they know the poll is different from their own polling. Its just common sense. I will let you off easy. You can just not respond.....were you dropped on your head?
So you think Trump is going to win? Simple question, yes or no.
You sound just like Dick Chaney in 2012.
