That is essentially my point. "These assumptions yield a very close election". And what if those assumptions are a little bit wrong? Or a lot wrong? Saying Obama wins by 1.7% is essentially saying "I have no idea what's going to happen" because that's well within the margin of error.

That's the result you get if you throw out all the polling done and just apply voter models from previous years. The polling that's actually being done show that Obama is having a lead in the range of 3.5% nationally, and larger than that in many important states (Ohio is the most striking example, where his lead is well outside the margin of error in mostly all polls nowadays).

Anyway, do you have anything to back up your claim that the polls use 2008 turnout data?

Fixed? Nope
Here's an argument similar to the one I was giving. I'm trying to independently verify the claim of the 2008 model, but google doesn't seem to index the data I want highly .
By "data I want", I'm referring to the models polls use now and in 2008 to compare them.

EDIT: That link seems to be broken, which is odd because I just copied it. One moment.

EDIT: I see. The interguild is censoring part of the url because it thinks it contains profanity. Erm. Not sure what to do about that.

EDIT: Well, if you want to see what it is, here's the url http://www.d___ckmorris.com/why-the-polls-under-state-romney-vote/ and replace the ___ with an i, which isn't swearing or breaking the rules because it's a person's name, you jerks.

Censors gunna troll

OK, so you used a conservative source to back you up, that has no outgoing sources himself to back his claim up, and has made factually doubtful claims such as that Bill Clinton will vote for Romney but is held hostage by his wife, and that muslims in the Park51 mosque would celebrate the 9/11 attacks and killings?

http://www.d-----ckmorris.com/no-federal-funds-for-ground-zero-mosque/
http://newsbusters.org/blogs/noel-sheppard/2012/07/30/d-----ck-morris-i-guarantee-bill-clinton-going-cast-his-ballot-mitt-romne#ixzz22FMEV7DQ

I am sorry, but I do not find that to be a credible source. Could you please find another?

http://www.caintv.com/IsDickMorrisrightthatRomneyisr-462
Also a conservative source, but the NYT is a very liberal source, so...
http://www.colorado.edu/news/releases/2012/08/22/analysis-election-factors-points-romney-win-university-colorado-study-says
I don't happen to know what the bias is here, but it's probably conservative since they start by saying romney will probably win. Dunno though. There's also the issue that this was last updated in august, which while relatively current is still out of date.
http://www.huffingtonpost.com/2012/08/22/university-of-colorado-pr_n_1822933.html
This one isn't quite talking about the 2008 model and polls but is as close as I can seem to find.

I guess the take away is that the polls may or may not be accurate because they may or may not be using correct models. Sigh.

Hey! I found a primary source! This demands a double post.
http://sprblog.wordpress.com/2012/09/24/why-our-recent-polling-is-accurate-and-shows-a-close-election-for-president-by-james-lee-president-of-spr/

EDIT:
Oh, and one more which seems pretty balanced overall, is recent, and discusses the model structure
http://www.commentarymagazine.com/2012/09/30/obama-poll-surge-doesnt-jive-with-democrats-registration-decline/

So, does this count as meeting my burden of proof?

First source also doesn't have anything that it's based on and is purely speculative, so no points there. You get no points for the NYT being liberal either, because you're debating a poll consensus, not how this guy uses the polls. You were the one to claim that pollsters use wrong voter turnout data, so the NYT/538 is not relevant.
The second model predicts that Obama would lose Minnesota...I'm sorry, but I find it hard to believe that he'd do so given the current standings in the polls and the historical fact that Minnesota has the longest Democratic voting streak of all the states in the US. Moreover, it doesn't support your statement that polls use voter turnout data from 2008 - in fact, it doesn't mention it at all and is frankly irrelevant. It uses another way to come up with a prediction than the regular poll, but that's it.
Third link is just talking about the second link, and like you said didn't bring up the 2008 model either.

So far, you have no credible sources to back up your claim, so I feel pretty confident in that the pollsters are doing a good job this year, just like they've done most other years (538 predicted all states correctly in 2008 except Missouri and all Senate elections correctly in 2010, so they're pretty decent at their thing).

PEdit: Oh hey. I'll get to that.

The first link is pretty interesting, I'll give you that, but I will continue my research. I think it seems weird to decrease voter turnout from 18% in 2008 among 18-29 years old persons down to 10% however - that doesn't seem credible.
The second link though, just says that some polls may or may not be using a poor model, but it doesn't say which ones, and moreover says that Romney has undeniably lost ground, so...that link just says "Hey, maybe there's problems with the polling. Look, less people have registered as Democrats." and leaves it at that, mostly.

Lo and behold.
http://fivethirtyeight.blogs.nytimes.com/2012/09/23/sept-23-does-omaha-matter/

(and besides, you haven't shown that the other polls use improper data from 2008, just that this one used data from 2004 and 2008 - like Rasmussen, which coincidentally has a Republican lean as well, both historically as well as now)

@ Decreasing to 10%
That's totally credible, as young people typically poll at even lower rates than that.
source

Also, here's a boat load of poll data, if you really, really want to analyze it yourself.
http://www.realclearpolitics.com/epolls/2012/president/us/general_election_romney_vs_obama-1171.html

EDIT:
To clarify, the claim isn't that they do something like oversample, which is obvious, but that when weighing the responses they get that they just use turnout data from the last election, which is normally an acceptable practice, but not when the last election was so unusual.

This is Begging the Question. You can't use the consensus to show that the consensus is true.

I stand partially corrected about the youth turnout then, but you do realize that the poll data RCP uses is the exact same as the one 538 uses, right?

PEdit: Yeah, I know. For all we know, the poll you found may be the correct one - but that means that a significant number of polls are completely wrong, whereas it's more likely that this, the statistical outlier, is the wrong one. We can't tell until election day, but until then, I'm siding with the majority of the polls instead of the sole poll showing a tighter race.

Didn't FiveThirtyEight correctly predict the result of every state minus Indiana in 2008? Seems legit to me.

Anyway, watching some of the debate Romney won pretty comfortably in my opinion. Worrying, because I absolutely cannot stand the idea of having Romney being president however debates have shown time and time again they don't make a damn difference nowadays.

Everyone except Missouri, not Indiana. McCain won Missouri with just 0.1% of the vote, so it was extremely tight.

I'd just like to take this time to laugh in the face of Jorster for being a climate change skeptic. I really don't care nearly as much about climate change as a lot of people, but at least I acknowledge it's existence. 99% of Scientists agree with it. The other 1% don't deserve to be called scientists since most are hired by the various oil and car companies, the others are just terrible scientists.

-snip-

-snip-

I did say I apologise if I'm making this up, but I definitely remember it somewhere... Maybe I'm getting confused with someone from another forum. If it isn't in this topic I'll snip it.

Snipped. I absolutely swear I saw a post about that somewhere, though...

Since you snipped yours I'll snip mine

*Looks up what snip means*

I site:interguild'd on Google your -snip- stuff Guyguy, and it does exist, but it honestly looked like you were taking it out of context from what I can tell. I might be wrong, but I think you misunderstood? Whatever. That is all I'm saying about that.

I only caught some of the debate live and caught a few snippets on the news. From what I could tell, it seemed pretty boring. And Romney totally creamed Obama. I don't think it will have too big of an effect on the race. Maybe if this happens in the rest of the debates it will, but not just the first one.

This also depends on what you mean by "correctly predict". If you mean that they generally favoured the person who won, then sure. If you mean they got the percentage of victory accurately, then no. And then there's the question of when. Polls change a lot over time, sometimes cycling several times. So, even if we assume that the polls are totally accurate right this instant, there's no historical reason to believe that the predictions being made now will be accurate in the future.

Also, I reread what I was saying, and I think I need to say that I don't believe the pollsters are intentionally rigging their polls, rather that the practice of approximating this year's turnout by basing it (initially) on last year's turnout is not a valid practice in elections following atypical turnout years.

I think the bulk of my objections were based on just that one website's predictions, which have Obama wining by double digit percentages, which is insane. If you look at the average of polls, the numbers look more like Obama in the lead by three points, as shown in this link I gave earlier. That number seems way more realistic.

538 predicted Obama to win with less than 5%, the exact moments prior to the start of the debate, and now predict that we will see polls lowering that number. There's no "double digits" in there, you're reading something wrong.



Moreover, the prediction they made that was 49/50 right was made on Election day, which to be fair, is absolutely the easiest day to guess who's going to win where. As it seems right now Obama is more or less certain to pick up Ohio and is favored in Florida, but just some months ago, that was one of the swing states with the most favorable polling for Romney, steadily showing a tight race at best and most commonly a Romney lead. However, what makes 538 unique is the analysis they make up until Election day, by sorting through the polls and applying their model. They weed out statistical noise from signal and helps you identify weird polls that should most likely be rejected outright, like one showing Obama to be behind Romney in Florida by double digits, at some point in August, but also identify house effects - which are useful, and which is why you shouldn't trust PPP or Gallup as much as...CNN, or YouGov. Even Fox News polls are more reliable than Gallup or PPP ones, and by significant percentages.

Oh, it was their "chances of winning" thing. Moral of the story is: don't think about anything after your bedtime.