2016-08-11, 10:43 | Link #21 | |
Part-time misanthrope
Join Date: Mar 2007
|
Quote:
Your post literally didn't have an argument. You stopped reading before making one. I'll gladly discuss the matter with you once you actually bother typing something related to the issue. |
|
2016-08-11, 16:37 | Link #24 |
Not Enough Sleep
Join Date: Nov 2003
Location: R'lyeh
Age: 48
|
latest polls out and Hillary is leading every Atlantic Seaboard State except SC and she is only 2pts behind there. Come Nov, Hillary could swipe every Atlantic State.
Hillary also penn a nice op-ed in Deseret News for Mormons. http://www.deseretnews.com/article/8...m-and-the.html
__________________
|
2016-08-11, 23:18 | Link #25 |
Banned
Join Date: Jan 2005
Location: Montreal, QC, Canada
Age: 40
|
No wonder why a number of Republicans have questioned Trump's mental health. What's even more embarrassing though is that all other GOP candidates yet found a way to lose against such a loose cannon, and thus allowed that embarrassment to reach worldwide proportions. There's plenty of food for thought for the next few decades when asking how and why such a crazy person is allowed to be that close of the highest political position in the world.
|
2016-08-12, 01:54 | Link #27 | |
Banned
Join Date: Aug 2013
Location: Holy Terra
|
Quote:
Sorry but I do not trust polls, never-mind if they told me if Clinton or Trump have 10+ advantage. What I trust is how american people feel, the majority that is. And right now Trump is much closer to him, even if some despite him, than Hilary who is rarely seen even on the news. |
|
2016-08-12, 02:34 | Link #29 | |
My posts are frivolous
Join Date: Nov 2008
Age: 35
|
Quote:
__________________
|
|
2016-08-12, 04:18 | Link #30 | |
Senior Member
Join Date: Nov 2011
|
Quote:
In addition, the polls a few weeks after the conventions have picked every winner (popular vote, but not the margin) since Nixon (source: bottom of http://fivethirtyeight.com/features/...r-than-trumps/). There's a strong predicative power of both the polls and (some) polling models that you can't just throw away without good reasoning. |
|
2016-08-12, 05:11 | Link #31 |
Banned
Join Date: Aug 2013
Location: Holy Terra
|
Here are some good reasons:
-Polls usually take samples anywhere form 500 to 2000 people. That's not even 1% of 320 million potential voters. So saying that all, or half of, all white stones on a pile are actually black because there are few black ones is not the true representation of how the things actually are. -Polls take samples from people who say who will they vote for. Those same people can change their vote on the day of the election. In short - people can give false data when asked during sampling. I know this from personal expiration as I did lie about my vote as many others did during out last elections. The result was that the opinion polls just before the elections were off for some 10% in some areas. -Modern day polls mostly take samples form the internet, while most of the US ( 286,942,362 people who have access ) population has internet access it's a good question how much of the muse internet for following politics and actually participating in official polls. -Latest example of how pools can be wrong is Brexit vote. While the popular opinion was tied polls have shown the majority of times how remaining was 2 or up to 8 percent higher than leave vote. Something that was far from the truth when the real results kicked in. -And lastly: there is first time for everything. Same goes for pool predictability, even if they predicted every outcome right so far there is always a chance that they can be wrong. In the end, if someone will win or not depends on how much effort they are going to put into campaign and debates. Also, from what I understand the polls are actually much more closer to the real deal some week or two before actual voting. So we will know the true results only after the 8'th of November, not before. And all of this is not going into possibility that some of the polling are deliberately set up to favor their own candidate or outright payed up - just like some video game reviews. I guess we shall wait and see but I am not putting my bet on statistics. |
2016-08-12, 09:27 | Link #32 |
AS Oji-kun
Join Date: Nov 2006
Age: 74
|
I've refrained from commenting much in these threads, but your comments are so full of misinformation that I'm forced to chime in.
First, it is apparent you have never taken, or didn't pay close attention to, a course in statistics. Do you understand how the error of estimate is inversely proportional to sample size? A poll of 400 people is twice as accurate as one of 100 people, but to halve the error again would require interviewing 1,600 people. Notice that all these formulas are indifferent to the size of the universe being sampled. When the population is much much larger than the sample size being drawn, only the sample size matters. Moreover with the sheer volume of polling being conducted, the compiled results across pollsters make for a better estimate than any individual poll alone. Next, it's not true that most polls are done over the Internet. Of the 269 general election polls archived at Huffington Post Pollster through yesterday, 125 used Internet methods, 103 relied on live telephone calls to landline and cell phones, and the other 41 used methods like "interactive voice response" over the phone. I've been running analyses of these data and find no significant differences in the results of Internet polling compared to live calls. (The other methods show some differences, but because they are used by only one or two pollsters it's not possible to determine how much the method itself was the issue. One of these organizations is the infamous Rasmussen Reports. Rasmussen's polling for his Republican clients led them to the mistaken belief that Romney would win in 2012. His polls this year average over seven points more pro-Trump than the consensus.) On Brexit, the Economist had the two sides tied in its final polling survey with nine percent undecided. That is hardly inconsistent with a final 52-48 result. While it's obviously the case that polls conducted soon before an election will be more accurate than ones conducted farther out, still one can learn a lot from the overall patterns and trends over the course of the campaign. For instance, in 2012, simply counting up the number of swing state polls in which Obama or Romney held the lead proved a very accurate predictor of the states Obama would ultimately carry. By the way, that same methodology applied to the 2016 race shows it is highly unlikely that Trump can carry Pennsylvania, Michigan, or Wisconsin, key states in his "rust-belt" strategy. I have many issues with polling, largely because of problems with response rates. These are especially problematic in primaries with small electorates like New Hampshire or Iowa. In those races with small electorates and a large number of polling organizations, it's quite likely many respondents were interviewed at least twice and some perhaps as many as five times. That probably gives too much weight to the views of the most committed, and often most extreme, respondents.
__________________
Last edited by SeijiSensei; 2016-08-12 at 09:48. |
2016-08-12, 10:45 | Link #33 | |
cho~ kakkoii
Moderator
Join Date: Nov 2003
Location: 3rd Planet
|
Quote:
Utah: "We’re having a tremendous problem in Utah. Utah is different."Trump ran his primary campaign the way he had because how the polls were reacting to his provocative rhetoric. Primary is over, and he consolidated his base but he is still running his campaign the same way he knows: words and without substance. The first debate will only help to highlight how lacking in substance he is. His campaign was considering not to attend some of those debates or skip them all together if he was leading in the polls prior to these debates. That is no longer possible.
__________________
|
|
2016-08-12, 10:49 | Link #34 |
My posts are frivolous
Join Date: Nov 2008
Age: 35
|
The main issue with the polls this time has to do with the possibility of non-stochastic error, particularly with sampling issues and self-selection bias. Many polls tend to use some variation of strat sampling in order to maximise statistical efficiency, but the voting demographic has varied quite considerably since 2012, which would invalidate the strata selection criteria that was used in 2012. Fivethirtyeight, as well as several other political pundits, was left red-faced when they got pretty much all of their most important primaries predictions wrong in 2016 after lauding their achievements in getting the general election right in 2012.
That said, while the chance of non-stochastic error is likely to be quite high in terms of predicting the actual percentage of votes, the statistic that is likely to be more accurate is the change in votes, which simply requires an assumption that the sampling methods have not changed since the previous poll a few days/weeks back, and that the voting demographics have not changed too greatly since then. Here, the polls are all saying the same thing: Trump got a bump after the Republican election, and then lost it. It may not be fully accurate to draw conclusions about who's currently ahead based on the percentage of votes, but I think it's fair to say that the last few weeks have been bad for the Trump campaign, as seen from the change in vote percentages.
__________________
|
2016-08-12, 11:07 | Link #35 |
AS Oji-kun
Join Date: Nov 2006
Age: 74
|
Statistically, my regressions show that the size of the bump after both conventions was about equal in magnitude, some three percent on the overall lead for Clinton, though obviously with opposite signs. Trump's problem now is that the post-convention polling shows Clinton still on that advantaged path.
__________________
|
2016-08-12, 11:15 | Link #36 |
Senior Member
Join Date: Sep 2008
Age: 38
|
http://www.politico.com/story/2016/0...sarcasm-226947
So now Trump is backtracking on the "Obama is the founder of ISIS" statement by saying it was all just sarcasm. Even though he doubled down on it and kept repeating it. And even when someone tried to give him the benefit of the doubt that he meant Obama created the power vacuum, yet he still insisted the Obama literally founded it. |
2016-08-12, 11:16 | Link #37 | |
My posts are frivolous
Join Date: Nov 2008
Age: 35
|
Quote:
__________________
|
|
2016-08-12, 12:16 | Link #38 | |
✘˵╹◡╹˶✘
Join Date: Nov 2006
Location: Australia
|
Quote:
It's almost a joke how the fear of the opposite party winning , now destroy American political thinking. As they literally could flip between two sides of the same argument depending which one fit their agenda more during that specific moment. I can bet that those same guys who call on Hillary winning right now, will be the same one who call on polls to not be trusted if Trump started to gain back the leads
__________________
|
|
2016-08-12, 12:40 | Link #39 | |
Banned
Join Date: Aug 2013
Location: Holy Terra
|
@SeijiSensei
I must say that your comment was very educating and quite interesting to read. I commented based on common sense when regarding this things but you nailed few things with particular sat of data. I will reply to only one thing: Quote:
Statistics are interesting but I always found them inappropriate when it comes to trying to determine future developments. In my opinion there are just too many unknown factors to combine in that it's very hard to get any accurate results. I mean sure, I can use current statistic to predict will McDonald's still be profitable in 20 years from now. And then other successful companies may kick in on the market, scandals might happen, economy might collapse etc... To me personally, statistics will always be useful for providing clear picture after analyzing already finite amount of data of the even that already happened. As for predicting future events... not so much. |
|
2016-08-12, 12:56 | Link #40 | |
cho~ kakkoii
Moderator
Join Date: Nov 2003
Location: 3rd Planet
|
Quote:
Polls should continue to widen slowly between the two candidates in favor of Hilary as November 7 draws closer.
__________________
|
|
|
|