情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,...

Anonymous @ 8:58a

Re: 2004 Exit Polling

You are correct about the 2004 exit polling in the Presidential race showing a popular vote victory for Kerry, and many have raised concerns about the discrepancies, which I will not address.

Of interst is this report on efforts to explain the discrepancy. However, if you focus on Appendix C at pp. 28-29, it appears that the exit polling controversy is a tempest in a tea pot, as the only states where the exit poll winner and the counted vote winner differ are IA and NV, and the differences between the exit polls and counted results are well within the MoE for those sub-samples.

Further, even if you flip those states to Kerry, Bush still wins the Electoral College. Game, set and match.

The 2004 exit polling discrepancy is a good intellectual exercise, but has little or no relevance, since the overall discrepancy was within the exit poll's MoE. As we have bashed about at length in this thread, when you are within the MoE, you cannot take anything definitive from a poll.

Counsellorben said:"jsh1120:As a mathematically in...

2008-07-02T10:40:08.205-04:00

Counsellorben said:

"jsh1120:

As a mathematically inclined lawyer (now software developer/lawyer), I must take offense! Fie on you, sir. <~grin~>

Seriously, lawyers are trained as advocates for our clients, and to take a position and defend it unto death. Also, as a general rule, most lawyers have strong analytical skills and relatively weak mathematical skills, a fact which I exploited often when I was a full-time practicing lawyer.

None of which has anythign to do with this thread. Pardon the interruption..."

I see no reason that we can't hijack the thread, Ben. It's not like we're the first to do so.

You're right, of course. And what I always found interesting about law students was the strong "analtyic" skills combined with the tendency to make generalizations based on unique, antecdotal information.

In fact, I used to have a quotation from Justice Vinson, I believe, in which he ridiculed evidence in a case argued before the Supreme Court because it ONLY met the confidence test of 95%. :)

The " National Council On Public Polls" link is in...

2008-07-02T10:10:07.341-04:00

The " National Council On Public Polls" link is incorrect. I assume you meant it to be this ( http://www.ncpp.org/node/4/#12 ).

jsh1120:As a mathematically inclined lawyer (now s...

2008-07-02T10:01:17.604-04:00

jsh1120:

As a mathematically inclined lawyer (now software developer/lawyer), I must take offense! Fie on you, sir. <~grin~>

Seriously, lawyers are trained as advocates for our clients, and to take a position and defend it unto death. Also, as a general rule, most lawyers have strong analytical skills and relatively weak mathematical skills, a fact which I exploited often when I was a full-time practicing lawyer.

None of which has anythign to do with this thread. Pardon the interruption.

Points taken. I can't speak to specific "liberal" ...

2008-07-02T09:58:52.437-04:00

Points taken. I can't speak to specific "liberal" and "left-leaning" bias in the polling except that I know you have to have a true representative sample for accurate numbers. If your sampled group has more liberal representative members than the actual population, and you have a result showing a 5-10 point Obama edge, that's pretty much a useless result. I can no more prove that "left-leaning" people have been over-sampled than you (or anyone) can prove that they haven't. It'll show up on Election Day when the proof is in the pudding. I just know (and this is biased too, but it's a stat, nonetheless) that at least 75% of hard-line Democrats I'm familiar with will NOT be voting for Barack Obama. Even if the national figure is much below that, can he really overcome that kind of a loss with "new" votes? No one had a counter to my argument that exit polling showed Kerry winning in '04. How is that explained?

Anonymous @ 8:08a said "Obama should be very afrai...

2008-07-02T09:45:04.936-04:00

Anonymous @ 8:08a said "Obama should be very afraid of his chances, because he's up against a battered and bruised GOP that has numerous issues working against it and very liberal polling showing him with only a slight edge."

"Liberal" polling? Public polls may reflect a bias of the pollster or the organization commissioning the poll, but that bias is as likely to be to the right as to the left. Are the polls commissioned by Fox (Not Really) News "liberal" polls? (Though I see someone else made this exact same point while I was typing away)

If you can offer hard evidence of a particular bias among the universe of polls Nate uses, then please do so, as that will only help to improve this site, the purpose of which is to present an unbiased model of this election cycle.

I admit to my liberal/progressive bias, but I am here because the discourse generally is fact based.

If you are "right-leaning," then you are likely to believe in the wisdom of markets. In 2004, the predictive markets favored Bush for all but a few days. Currently, all of the predictive markets favor Obama, with contract prices representing figures very close to Nate's "Win Percentage" tracker.

You are correct that things may change, but are you willing to bet your own money on this election?

Others have noted the most obvious "bias" in CNN's...

2008-07-02T09:30:36.565-04:00

Others have noted the most obvious "bias" in CNN's (and other media's) headline writing. They all, especially the cable news networks, have a vested interest in maintaining the picture of a close race, shifting momentum, fading frontrunners, emerging underdogs, etc. If reality can be stretched to support such headlines, they'll do it.

I'll only add a personal observation from having taught statistical sampling and questionnaire design at a couple of major universities some years ago.

Most of my students were social science undergraduates and graduate students. Not an especially mathematically inclined group. But since I'd shared that perspective as an undergraduate, I knew where they were coming from and was more successful than most in getting through to them.

There were, however, two groups that were especially resistant to the inferential statistical methods: law students and journalism majors.

I concluded that each group suffered from the same weakness: their inclination and training inclined them to make generalizations based on antecdotal evidence. In the case of law students it appeared to be an over-reliance on precedent. My students were often willing to cite a case of a medieval dispute involving the ownership of a cow as evidence in making a case, but strongly resisted the notion that 95% certainty in a sample result was sufficient to guide a decision.

In the case of journalists, the preference for an antecdote was tied to the need to "tell a story." A convenience sample of four voters interviewed at length with suitable quotes was far more persuasive to them than a random sample of a thousand voters.

In general, I was impressed by the fact that these perspectives seemed uncorrelated with intelligence. Law students were among the brightest I taught. And journalism undergraduates, while not the best, were frequently bright kids. (Espcially compared to those majoring in "education.")

Their resistance wasn't based on an inability to understand statistical inference as much as an unwillingness to depart from single cases to broader samples.

anonymous @ 8:07Define "very liberal polling." If ...

2008-07-02T09:26:57.231-04:00

anonymous @ 8:07

Define "very liberal polling." If the pollster asks "If the election were held today, would you vote for Barack Obama, the only man capable of saving our country from a robot invasion, or John McCain, the guy who shoves old women down stairs and eats babies?" is that very liberal polling?

Does Fox News do "very liberal polling"? Even their last poll showed Obama up by four points.

If Obama(or even McCain) had been consistently ahe...

2008-07-02T09:09:23.830-04:00

If Obama(or even McCain) had been consistently ahead by about 10 points and had been for months and was still ahead by ten points, but yesterday it had been eleven, CNN's pretty girls and boys would still be reading off the teleprompter "The race tightens...more after this" in the hope that the viewer didn't have a remote control immediately to hand.

So much easier to have a horse race than do any real reporting about the issues reflected by the poll numbers.

I'll try and post again, since my original post wa...

2008-07-02T09:07:14.051-04:00

I'll try and post again, since my original post was deleted by the user. I'm not surprised given my "right-leaning" with the hard "left-leaning" of this site, though. Anyway, I was just pointing out that even exit polling (done AFTER the '04 election) showed Kerry had won some of the states he really lost and that he had a 2-4% lead overall nationally. Yeah, we see how that turned out. Obama should be very afraid of his chances, because he's up against a battered and bruised GOP that has numerous issues working against it and very liberal polling showing him with only a slight edge. He will lose by a considerable margin to John McCain come November. Mark my word.

Hey, I know it's off topic, but is there a way to ...

2008-07-02T08:40:52.868-04:00

Hey, I know it's off topic, but is there a way to have the percentage win (the stuff over on the left) with a graph showing changes since Obama won the nomination?

Coupled with the polls I think that would be really useful, and even shed some light on the dead heat brou-ha-ha. I'd love to see a running graph of how Indiana, say, has moved.

Any hope for this?

Thanks for jumping on this. I thought that CNN an...

2008-07-02T08:38:30.357-04:00

Thanks for jumping on this. I thought that CNN analysis was absurd! Unfortunately (or perhaps fortunately for CNN -- they love a horse race), most people don't know enough about polling to question the idiocy of their "analysis."

If we're just going to add 3.5% to McCain's number...

2008-07-02T08:34:48.235-04:00

If we're just going to add 3.5% to McCain's number and subtract 3.5% from Obama's then yeah, McCain's leading 48.5-46.5.

But if you're going to do that, you can also do the opposite and say Obama's leading 53.5-41.5.

There's nothing very scientific about either of those calculations and they're about equally (un)likely to occur. But you won't see many people spinning this poll as a "statistical 12 point lead" for Obama.

I hope people start making up their minds for Obam...

2008-07-02T08:09:33.200-04:00

I hope people start making up their minds for Obama or McCain soon. The closer the race, the easier it is to rig an election where electronic voting or electronic vote tabulation takes place. Hence, tight races could leave much room for foul play, and history has shown which party that usually favors.

Comments on FiveThirtyEight: Politics Done Right: And Here I Thought John Kerry Lost...

酒店經紀人, 菲梵酒店經紀, 酒店經紀, 禮服酒店上班, 酒店小姐兼職, 便服酒店經紀, ...

情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,...

Come on now, you know what they mean. I agree, 5 p...

It's okay, counsellorben. I kind of like for you t...

counsellorben,Don't preach to me about your sound ...

You're right, of course. And what I always found i...

If it's within the margin of error, it's certainly...

Why has no one commented on the fact that this CNN...

Anonymous @ 11:04aYes, you are the champion, that'...

Great job on your site Nate. One thing I would li...

"Game, set, and match"? Boy, I love it when someon...

Anonymous @ 8:58aRe: 2004 Exit PollingYou are cor...