Quote :

"i'm not trying to compare the validity of CNN versus drudge"

Yes, you are.

10/3/2008 7:50:11 PM

No, I'm not. Actually I am (after my edit at the end of the last page). I'm saying neither of their polls are likely worth a damn

I haven't even seen the drudge poll, but I heard it said that Palin won the debate which isn't a surprise given the source...but don't focus so much on the validity of the drudge website or you'll keep missing the bigger picture:

How can CNN asking 611 people what they thought be representative of 100+ million voters?

[Edited on October 3, 2008 at 7:57 PM. Reason : .]

10/3/2008 7:51:14 PM

.

[Edited on October 3, 2008 at 8:03 PM. Reason : nvm]

10/3/2008 8:03:14 PM

Good luck with that theory.

10/3/2008 8:06:49 PM

seems like common sense that 611 people's answers couldn't accurately represent 130+ million voters opinions, but thanks for not explaining it to me

i've been asking repeatedly for an explanation, but nobody seems to be able to explain it...hmm

[Edited on October 3, 2008 at 8:18 PM. Reason : .]

10/3/2008 8:13:51 PM

How come I knew last night...before watching the debate...that the FOX text message poll was going to overwhelmingly say Palin won....and I also knew that the CNN poll would say the opposite?

It's sort of like how I'll know that if I go into a synagogue and ask all those who believe that Jesus Christ is their lord, to raise their hands....the number is going to be overwhelmingly small....where as if I went into a Church, the number would be the opposite.


and TT...yes, I meant high not low

10/3/2008 8:26:03 PM

If McCain deserves to lose for any reason, it's picking Palin. It's like he believes that his republican supporters and prospective independents are too stupid to spot how unprepared she is from a mile away.

Attn Mccain: your supporters aren't fooled by you hiding Palin. If she can't stand the heat of a campaign, she sure as hell isn't going to stand the heat of the presidency. Instead of pandering like Obama, you should have picked someone who was qualified and picked someone who would have helped your ticket for more than 1 week after the convention.

10/3/2008 8:26:46 PM

Here's a link to the debate transcript:
http://elections.nytimes.com/2008/president/debates/transcripts/vice-presidential-debate.html

Biden's arguments are fairly focused and can easily be picked apart. Palin, however, has a sort of rapid-fire talking-points technique that is somewhat overwhelming due to the volume of arguments she conveys in a single statement.

10/3/2008 8:27:57 PM

^^^^ Although I don't put much stock in polling data so soon after an event such as a debate I do agree with your assertion that 611 is far too small a number to represent such a large size. To me it's akin to using the Young Earth Creationist movement to represent all of Christianity. With this methodology by CNN it is hardly surprising why people question their impartiality.



[Edited on October 3, 2008 at 8:32 PM. Reason : For Twista]

10/3/2008 8:30:07 PM

I didn't look at those transcripts.....but I'm just sayin that based on what the trusted newsource CBS does with it's 'transcripts' I'd be weary. I mean hopefully they are accurate but it wouldn't surprise me if they weren't

10/3/2008 8:30:56 PM

Palin was a huge mistake that showed a clear lack of common sense by John McCain. Good luck fellas.

10/3/2008 8:31:09 PM

^and what does that have to do with the current discussion you were engaged in regarding polling? Or did you give up that one

10/3/2008 8:32:17 PM

Provide any evidence that the poll was unreliable or unscientific with a primary source and we can discuss further. Otherwise, I have nothing more to say on that point.

10/3/2008 8:34:11 PM

Here is the Bosniac discussion. I had it wrong, it was actually PBS

http://talkingpointsmemo.com/archives/221642.php

10/3/2008 8:37:54 PM

why cant anyone explain how polling 611 people, when in 2004 ~130,000,000 people voted, could possibly be scientific?

10/3/2008 8:38:10 PM

It's statistics. I don't know how they do it, but they do it and it is very accurate usually within 3-4 points.

10/3/2008 8:39:09 PM

Quote :

"I don't know how they do it"

me neither which is why i keep asking

and no offense but "It's statistics" isn't a very convincing answer

10/3/2008 8:40:03 PM

If you go to any reputable polling place they will have their methodology. There are different variables that the different polling places use.

10/3/2008 8:41:39 PM

thats fine, but on the cnn poll that everyone has been discussing, there isnt any methodology explained, at least not that i've found

are you saying the cnn poll isnt reputable? cause i've kind of been arguing that none of the polls are reputable, at least none of the cable news channel and website polls

10/3/2008 8:42:45 PM

^9 or so
The only errors I saw were the occasional typo and once attributing a Biden quote to Palin. That particular instance was fairly glaring, since Biden was referring to himself and Obama and the question was directed towards him.

It was fairly accurate for the most part, even correctly quoting candidates when they mispoke, from what I recall watching the debate.

10/3/2008 8:48:08 PM

If you can show me that they are flat out wrong, please do so. I would like to know the truth, but until then you are asking me to believe that all of the polls are meaningless, which is a little bit coincidental considering your candidate doesn't favor well with them. I'm open to the idea, prove it.

10/3/2008 8:48:24 PM

well...um.....since these polls are 'random' and since there is no way to verify that the people answering the phone are indeed registered to vote, or for that matter, of voting age...and since the population in the United States in somewhere around 305 million....and just for shits and giggles, let's say that 100 million of those people aren't able to speak yet.....then you are suggesting that .0003% (that's 3 ten thousandths of one percent) is in any way reliable. And if you actually believe this and I have to provide evidence of why that is not reliable, then well....I hope you don't plan on procreating

10/3/2008 8:51:35 PM

I don't know what equations they did. There isn't much deviation between the equations. That's pretty much all I remember from the class I took on polling.

^I promised theDuke I wouldn't get into it with treetwista anymore,, but that doesn't go for you. Polling methodologies have been tested time and time again and are typically close to the truth. Arguing that that is not true shows your ignorance. There are plenty of criticism of polling, but sample sizing isn't really one of them.

[Edited on October 3, 2008 at 8:57 PM. Reason : .]

10/3/2008 8:53:01 PM

um what now? So you're saying that if I go to Spring Break in Cancun and I walk around in some slutty outfit and ask random people whether or not they think I'm hot as shit....and of the 611 people 600 of them say that I am, then I could make the statement that 98% of all Americans think I'm hot as shit?

Now while this may indeed be true...exactly how accurate is this poll?

10/3/2008 9:02:11 PM

Quote :

"why cant anyone explain how polling 611 people, when in 2004 ~130,000,000 people voted, could possibly be scientific?"

Because they select those few people randomly. You can then calculate how likely it would be to hit the same opinion over and over. For example, say the country were evenly divided between McCain and Obama. In the poll in mentioned, the odds against selecting all Obama supporters would be 2 to the power of 610. Now, no poll yields completely certain results. In addition to the margin of error, most use a 95% margin of error. So there's at least a 5% the poll's significantly off. You can always roll a one.

Quote :

"So you're saying that if I go to Spring Break in Cancun and I walk around in some slutty outfit and ask random people whether or not they think I'm hot as shit....and of the 611 people 600 of them say that I am, then I could make the statement that 98% of all Americans think I'm hot as shit?"

Americans in Cancun during Spring Break would not accurately represent the country. Plenty of Americans don't go there for various reasons. That would bias this poll.

[Edited on October 3, 2008 at 9:06 PM. Reason : Cancun]

10/3/2008 9:02:56 PM

Quote :

"well...um.....since these polls are 'random' and since there is no way to verify that the people answering the phone are indeed registered to vote, or for that matter, of voting age...and since the population in the United States in somewhere around 305 million....and just for shits and giggles, let's say that 100 million of those people aren't able to speak yet.....then you are suggesting that .0003% (that's 3 ten thousandths of one percent) is in any way reliable. And if you actually believe this and I have to provide evidence of why that is not reliable, then well....I hope you don't plan on procreating"

You can't be serious. That is your example of proof?

10/3/2008 9:06:41 PM

Quote :

"um what now? So you're saying that if I go to Spring Break in Cancun and I walk around in some slutty outfit and ask random people whether or not they think I'm hot as shit....and of the 611 people 600 of them say that I am, then I could make the statement that 98% of all Americans think I'm hot as shit? Now while this may indeed be true...exactly how accurate is this poll?"

In your stupidity, you completely failed to acknowledge the aspect of equations going into the methodology.

10/3/2008 9:08:28 PM

Quote :

"There are plenty of criticism of polling, but sample sizing isn't really one of them."

i beg to differ...how can you not look at these 2 numbers and think sample size isn't a possible criticism?

overall size of group to be represented: >=130,000,000 people

overall size of sample group polled: 611 people

conservatively, thats sampling 1 out of every 200,000 people

how can you not question the validity? you can say mccain supporters would be skeptical if the polls didnt serve him well, but you could also definitely say that obama supporters might be willfully ignorant to the accuracy of the polls because the cnn poll DOES serve obama well

10/3/2008 9:09:42 PM

Quote :

"how can you not question the validity? you can say mccain supporters would be skeptical if the polls didnt serve him well, but you could also definitely say that obama supporters might be willfully ignorant to the accuracy of the polls because the cnn poll DOES serve obama well"

Fair point, but the burden of proof resides in your court, not mine. I don't have to prove the validity, you must disprove it to back your assertion.

Check.

Your move.

10/3/2008 9:13:39 PM

If we are willing to throw out the validity of polling, then we must rule that the entire metrics used by advertising firms to be completely bullshit and without merit. We all know that is not true in the least.

10/3/2008 9:13:59 PM

^^why do I have to prove the validity? in this instance, isnt it CNN's job to prove their polling validity?

^obviously we can all agree that polling isnt any kind of definitive answer and might do a good job of figuring out a general opinion on a particular topic, but the precision of that opinion gets more acute as the sample size is increased

besides, if you are an advertiser and you only rely on 611 peoples' opinions, you havent done enough research to validate marketing a product

[Edited on October 3, 2008 at 9:16 PM. Reason : .]

10/3/2008 9:14:26 PM

years of history prove their point.

10/3/2008 9:15:35 PM

"years of history" has nothing to do with the poll from last night

if you'd prefer to criticize a fox news poll or something so you wont have a conflict of interest i'll gladly change the topic to one of those polls as the source, as the results have no impact on my argument whatsoever

[Edited on October 3, 2008 at 9:20 PM. Reason : .]

10/3/2008 9:18:13 PM

Quote :

"how can you not question the validity?"

Randomness. If you truly select randomly, six hundred people will yield solid results. Confidence interval of four or so at 95% confidence level. The overall population size hardly matters. Come on, this isn't particularly advanced math. Even I understand it.

10/3/2008 9:29:53 PM

Quote :

"The overall population size hardly matters"

how can you say that? 600 people would probably do a great job of giving a representative sample of an overall population of 12,000 people...but we're talking about 130,000,000 people

could a poll of 600 chinese people accurately represent the opinions and views of 1,300,000,000 people? in the CNN poll, 611 people versus the voting public is the equivalent of going to 4 different NC State football games with sellout crowds of completely different people and asking 1 total person their opinion on something

[Edited on October 3, 2008 at 9:35 PM. Reason : analogy]

10/3/2008 9:32:24 PM

Quote :

"how can you say that?"

Statistics. You know how manipulate simple probabilities, right? If you flip a coin twice, there's a one-in-four chance it'll come up heads both times. Go to three flips, and it's one-in-eight odds for straight heads. See the pattern? It's exponential. Two squared, then two cubed. Extend that out to six hundred. The odds of results dramatically different form the average become mind-boggling. That's the logic behind polls.

[Edited on October 3, 2008 at 9:40 PM. Reason : logic]

10/3/2008 9:39:42 PM

Simple statistics will tell you that the greater the sample size, the more precision of your results...if we were trying to ascertain information about a group of 1,000 people...and we asked 100 people their opinions, I think that would be a good representative sample...however asking 900 people would be a much better way, and only asking 10 people would be a lot less accurate assessment of the entire 1,000 person group

The CNN poll is a similar situation...except their group isn't 1,000 people, its 130,000,000 people...and they're not asking 10% or 90% or even 1% like the example in my first paragraph...they're asking roughly 0.00047% of the population...how can people live in this world and not realize how many differing opinions and outlooks people have and trust a poll that asks such a minuscule percentage of the population? And in this case I AM simply arguing statistics. The inherent biases of various websites and news outlets comes later. I'm talking about 600 out of 130 million. We're talking winning the lottery type odds here.

[Edited on October 3, 2008 at 9:46 PM. Reason : i edited]

10/3/2008 9:44:13 PM

Quote :

"Simple statistics will tell you that the greater the sample size, the more precision of your results..."

Yes, but that's mostly independent of overall population size.

Quote :

"if we were trying to ascertain information about a group of 1,000 people...and we asked 100 people their opinions, I think that would be a good representative sample..."

No, that would actually be significantly worse than your example of 611 for 130,000,000. I'm not making this shit up. See for yourself:

http://www.macorr.com/ss_calculator.htm

Again, the accuracy comes from the exponentially increasing odds. 2 to the power of 611 beats 2 to the power of 100. The overall population size hardly matters.

[Edited on October 3, 2008 at 9:50 PM. Reason : worse]

10/3/2008 9:49:48 PM

Your example is correct in mathematics, but not with peoples' opinions and viewpoints which are much more dynamic and unpredictable. Flipping a coin or having some type 50/50 question has 2 finite answers per trial. The poll itself had a number of questions, most of them with 50/50 (Biden/Palin) answer selections. The reason these polls aren't accurate, is because voters do more than ask themselves 6 or 8 questions and decide who to vote for, its much more complex, its the human thought process.

And so while each individual "Biden or Palin" question might be representative of how the country might view the candidates on that specific issue, thats oversimplifying many other dynamics of how people are different

[Edited on October 3, 2008 at 9:59 PM. Reason : .]

10/3/2008 9:54:30 PM

Quote :

"TreeTwista10: Can we really know anything, maaaannnnnn?"

Did you totally talk to a professor this one time that told you statistics was bunk?

10/3/2008 9:58:09 PM

Okay, well that's a different argument, Twista. I'm glad you've off the last road you were on. It's hard for many folks to understand that a poll of five hundred or a thousand people can accurately reflect the entire nation. However, the math works out. Think about it this way, everybody: When you taste a well-stirred soup, do you use a bigger spoon for a bigger pot?

If you want to question polls, question the randomness. Who knows if pollsters manage a truly random sample? Don't question the entire basis of a field of study.

[Edited on October 3, 2008 at 10:05 PM. Reason : meow]

10/3/2008 10:04:49 PM

[Edited on October 3, 2008 at 10:12 PM. Reason : don't feed the troll. ]

[Edited on October 3, 2008 at 10:18 PM. Reason : foo]

10/3/2008 10:11:53 PM

^ This strikes me as an overly caustic response to poor Trap.

10/3/2008 10:14:11 PM

There. Life is too short and it's Friday night.

[Edited on October 3, 2008 at 10:19 PM. Reason : foo]

10/3/2008 10:16:50 PM

*Official* Vice-Presidential Statistics Debate Thread

10/3/2008 11:48:36 PM

I think Palin won. She clearly had the bigger flag lapel-pin.

10/4/2008 1:49:04 AM

Using my example of Spring Break in Cancun, this just proves my point. If I wanted to hear how hot I thought I was I probably wouldn't go to a Big, Black and Beautiful convention and ask those people the same question because chances are good that number would be significantly lower. But you see...I didn't go to the BBB convention, I went to where I knew there would be lots of drunk retarded guys who's sole purpose is to fuck as many chicks as they can. Now if I'm trying to convey to TWW how most people think I'm hot shit, I'm not going to tell you that my random sample was really just a random sample of drunk horny retards am I?

If I'm CNN who is a liberal media outlet, and I want to know who is going to vote for whom, I'm probably going to 'randomly sample' people in a blue state...perhaps those people in a certain area code and prefix where black people are more likely to live....why? Because I'm going to get a pretty good showing for Obama and see...I 'randomly' sampled them. Much like if Fox did that same poll, they wouldn't 'randomly sample' anyone in the 415 area code because chances are pretty good that a liberal is going to answer the phone...so I 'randomly sample' people in 'red' area because I'm going to get a pretty good showing for McCain.

Now see...both of those were 'random' but one poll leans 80% for Obama, the other 80% for McCain so who gets to make the claim that their statistic is reliable?

Do you really think it just so happens that the polls from last night reflected this? Really??

10/4/2008 2:19:14 AM

Except neither of those examples would be random

The CNN poll -was- random. The only evidence you have to the contrary is a vague assertion that they're liberal. Or left of whatever nutjob pundit you pay attention to.

[Edited on October 4, 2008 at 8:50 AM. Reason : ]

10/4/2008 8:45:29 AM

October 3, 2004

Kerry 238 Bush 296



October 3, 2008

Bams 338 McCain 185 Ties 15



The 2004 election turned out exactly like that map, except for Wisconsin and NH going to Kerry by the smallest of margins. The bottom line is polls turn out right far more often than not, and trying to pretend that McCain isn't in serious trouble right now based on polling is just being ignorant.

10/4/2008 8:50:40 AM

but but... it's not a valid poll unless it samples 100 million people, and is conducted by Jesus!

I love how at least two people in this thread love to pretend they're unbiased, yet when their candidate is losing, they resort to questioning the basic fundamentals of polling, rather than admit that they're losing.



[Edited on October 4, 2008 at 9:04 AM. Reason : ]

10/4/2008 9:02:12 AM