Non-Gallup/Rasmussen tracking polls thread

[Lunar]

I gave it 9:1 odds of showing a lead, Sam only said it'd be at least tied.

Please at least quote me correctly next time.  Thanks. 

After the ten point movement the last four days, I am willing to bet even money that Zogby shows McCain within a couple of points or even ahead sometime before the election.

Another thing: Friday night/Halloween polling is bound to be horrible.  Do not trust it one bit.

Zogby polls (*supposedly*) during the daytime, so he's actually exempt from this edict.

Haha, I was just joshin'!

I wonder if Zogby is secretly using his internet numbers for his actual poll...  it'd be cheaper

He's certainly found a great ally with Drudge

[Lunar]

.... and the TIPP gets headline treatment by Drudge.

[Lunar]

http://www.washingtonpost.com/wp-srv/politics/documents/postpoll_110208.html

Obama +11%

[Lunar]

*slams head into wall*

[Lunar]

I'm slamming my head into the wall because I'm incredibly annoyed.

EVERYONE knows that if Obama underperformed the average of polls in Pennsylvania by even 0.75%, you'd be up there declaring the victory of the Bradley Effect.

Plls can be wrong for many reasons.  Cherrypicking and then averaging national polls to prove the Bradley Effect when a simple analysis of your Pennsylvania Bradley obsession proves you wrong.  Obama strongly overperformed the polls in the state that you were certain would contain at least a small Bradley Effect.

If you can't admit that polls, especially bad ones, can't overestimate the Democrat for any other possible reason besides race, you're...well...there are several dozen reasons why a poll could be off and doing this makes you look bad.

*slams head into wall*

[Lunar]

For every poll that Obama overperforms in, it's 100% due to race.

Ok, from now on, just to prove how idiotic this crap is, I'm going to assume that every time McCain underpolled it's also because of race.

Are Pennsylvanians lying, claiming to pollsters they shall vote for McCain, because McCain was white?

This is dumb.  *hits head against wall*

[Lunar]

So the main difference between the results of Gallup and Rasmussen is the robo-factor?  I assume most would say their weighting schemes....

And the main difference between robo-calls and live-interviewers is not "standardization vs. higher response rate," but rather people trying to not appear as racist?

Suddenly, for J.J., the Bradley Effect has switched from people lying to people lying to live interviewers, out of retroactive convenience.

*bangs bloody head against wall*

[Lunar]

Statistically Obama should overperform his poll margins 50% of the time

[Lunar]

Statistically Obama should overperform his poll margins 50% of the time

Actually, Obama should over perform about 1 in 20 times out of the MOE.  We have Zogby numbers on Gallup and ABC/WP.  Come on, a 2 point MOE and the polls are off 4.5 points.

Didn't Obama constantly overperform his polling number in MANY states by >4%?

What do you attribute that to?  Why is it that every time he underperforms it's because people are afraid of being racist to the interviewer?

[Lunar]

Pennsylvania?  Nevada?  New Mexico?  Michigan?


He beat the MoE of the polls in all of these states.  Why is it the fricking Bradley Effect whenever the polls are wrong in the inverse?

*hand trembles*

[Lunar]

Pennsylvania?  Nevada?  New Mexico?  Michigan?


He beat the MoE of the polls in all of these states.  Why is it the fricking Bradley Effect whenever the polls are wrong in the inverse?

*hand trembles*

PA:

From the 30th two Rasmussen, 1 ARG and a crap tracking poll  out of the MOE, 3 out of 12.

NV:

3 polls in the last week, two were out of MOE one was on.

NM:

3 polls in the last week, two in MOE, one out.  One overcounted Obama.

MI:

Five polls in the last week.  4 were in the MOE.

*suggest Lunar puts down the crack pipe*

So if 9 out of 12 polls are out of MoE, it's nothing notable and these polls have a solid methodology.  But if a few polls that show the inverse, it's caused by race.

You know what disproves you more than anything else?  The fact that 538's model, which weights polls by their accuracy during the primaries, and that used all of the national polls, simply averaged them and ran an regression model and *gasp* came up with exactly McCain's number and 0.1% off Obama's number.  Why did they end up with the perfect national number without accounting for the Bradley Effect. Compare that to your hilariously stupid rule of just adding 1.5% to McCain's number because people don't like to appear as racist.   What you do is you twist and twist the numbers and cherrypick where you like to support your predetermined conclusions.

Whenever you're wrong, you just keep digging.  Nothing phases you.  You make hilariously bad predictions time and time again and your analysis is so bone-headed and hackish, but you never back off it, you just get more and more convinced that you're right when the evidence piles up against you.  Your Pennsylvanian prediction was only what, 9% off? You were still listing the race as 50/50 a week out?  Your predicted margins (copying Josh's but shifting North Carolina) were one of the worst in the forum.  There were at least a dozen predictions you made during the primaries that did not happen.

I give up.

Let me say this one time:  Lying to pollsters to not appear racist is a source of error but it's not the dominant source of error and nothing you have ever said gives evidence otherwise.        I could list two dozen things that are logically more likely to be a source of error.

*blocks for a week in order to maintain my own sanity* 

[Lunar]

No, you can tell me which polls are "complete crap" now.  You don't do an experiment and then toss out data points afterwards.  Why would you?  That serves no purpose whatsoever other than to potentially introduce bias.

I'm not reading the other posts, but there should be a standard for what polls are complete crap too, right?  538 doesn't toss out any polls, it just weights them on historical [primaries] accuracy...

I'll read the rest when I've cooled down in a week or two.

[Lunar]

I think it's time to retroactively throw some polls out.

[Lunar]

Still waiting to find out why a three-day tracker is different than a three-day poll, btw.

...wait, you mean that polling accuracy fluctuates up and down, possibly due to poor weighting, and it's stupid to apply analysis to tiny amounts of error without statistical significance, especially if it doesn't stand out from the overall sample?

[Lunar]

and there's still that pesky 538 site that managed to predict both Obama's and McCain's exact numbers (not margins) by weighting the polls based off of historical accuracy.  Well, it was 0.1% for one of them.

0.1% reverse Bradley effect?

[Lunar]

What, are you saying there might be another reason McCain overperformed in Arizona besides race?

[Lunar]

I actually think it's the left-handed effect.  Obama is left-handed and most people aren't, so they lie to pollsters but only in Arizona and Iowa.

It makes perfect sense, shutup!

[Lunar]

Yeah whatever is statistically insignificant is what I'm going to rely on to base my opinions on.

Idaho here I come!

[Lunar]

Don't debate the historical bits, even if they are arguable.  That just gives him leeway and the ability to dodge the fundamental facts. 

Lack of information is the basis of his argument.  Whenever there are less polls and lesser quality polls (say, Arizona '08 or all of the SUSAs '06 which all favored the Democrats by 5%, black Republican or not), that's when polling error is most likely to show up.  Iowa's overpolling could largely be explained by the fact that McCain completely ignored the conservative base there and some small percentage of them felt compelled to answer "Undecided" to poll questions -- or that they were afraid of being labeled racist!

[Lunar]

was the Bradley Effect even ever intended to be applied to black Republicans?

J.J. would often like to argue that black Republicans might lie and say they are voting for the socially acceptable candidate (Obama), but blacks have always been pressuring their social group to be Democratic.  I think it's more likely that long-time black Republicans would claim to be for McCain when actually secretly for Obama than the inverse.

[Lunar]

oops, my mistake

Well J.J. says it doesn't matter whether he's Republican or Democrat.  If he overpolls and he's black it's because voters lie to pollsters motivated by race.


Also, it doesn't apply to primaries.

[Lunar]

Something just hit me as to how even more inane J. J.'s arguments are. His three referenced examples of the Bradley Effect in 2006 are all states where no one would argue it happened in 2008. So basically apparently these states changed so much in only two years that the Bradley Effect disappeared. But it then popped up again in states like Iowa and Arizona for some mysterious reason.

That is beyond laughable.

There is 0% change that polling in Iowa happened because of dissatisfied Republicans in the primaries claiming to be undecided or bad polling weights and 100% chance it happened because Obama was back and people were afraid of appearing racist.

Arizona, I don't have any idea what could be up with that.  I suppose with any same size of 50 there are guaranteed to be outliers, especially those polled by only a few firms and most of them are god-awful, creating automatic error.  Other than that, any ideas as to why Arizona might underpoll McCain?  Any at all?  Any kind of connection with McCain they might have to sway undecideds?  Ok, nevermind, I'll assume it's because of race.

[Lunar]

McCain:  UT, AZ*, KY*, GA*, NH*, IA*, ND, AK*

Obama:  MA, VT, PA*, IN?*, NM*, CO*, NV*, MI*

Ok I brought myself to read your posts again.

Do you notice that the odds of Obama underpolling in any given state are completely equal to McCain?  Despite the fact that McCain is white?  You cited an equal number of outlier cases for both candidates.  Why can McCain underpoll for non-racial reasons but not Obama?  Why can Gore or Bush underpoll or overpoll for non-racial reasons?  Why are Obama's reasons racial!!

Is it possible that this is just statistical error or methodological error THAT HAS NOTHING TO DO WITH VOTERS LYING?

AHHOGHOHOIGOHIEHOIFHOEOHFHOEWOHFE

IT'S A COINFLIP. A NON-RACIAL COINFLIP




STOP CHERRYPICKING

[Lunar]

So Alaska and North Dakota have Bradley Effects but no state in the south except Georgia does. And the Bradley Effect has magically vanished in the three states it supposedly appeared in in 2006. Yeah, that makes perfect sense. This is clearly the Bradley Effect as there is no other possible explanation for polls being wrong.

There was a Bradley-Effect in almost every Southern State. Take a look at this:

https://uselectionatlas.org/FORUM/index.php?topic=88407.0

I'll do a full analysis in the coming weeks, when I have some time.

For example there was no Bradley-Effect in California, maybe a Latino-Effect, because the very accurate SUSA poll showed Latinos at 22% of the electorate, but they only made up 18% and they supported Obama by a slimmer margin than the poll predicted. But Whites supported Obama exactly by the margin the poll predicted.

First off, many polls make weighting errors!!!!!!!!


Yes, that is one aspect of the Bradley Effect insofar as undecideds break strongly for the non-white candidate?  But the better question is if white undecideds in the South broke strongly for Bush in 2004!  If we are to assume this has to do with race (which any assumption of the Bradley Effect involves), then we need to compare this to a white non-Southern, recent candidate.  What did they do in 2004? 

If we can acknowledge that undecided voters often break one way or the other, then we can acknowledge that this sometimes happens because of reasons not associated with race.  If undecided voters broke towards Bill Clinton in a few states, there's at least fifty million ways to rationalize that without resorting to foolish logic.  But J.J. chooses to emphasize race as the driving variable here despite all evidence and pre-election predictions (which were based upon logical demographics -like Pennsylvania).  One asks why Obama’s has to do with race and if normal irregularities combined with standard state-associated traditional undecided-breaking schemes can explain everything within statistical significance.

And I’m not sure if the B.E. inherently includes an exception clause that says “oh but if blacks turn out it doesn’t happen.”  That’s sort of an added on claim.

And how can you separate Obama’s race from his message?!?!  His message of change, of being different?




*And of course remember that error increases if we're using exit polls versus normal polls and it increases if we're isolating crosstabs.

[Lunar]

Might I add that the scientific method involves making predictions and then testing them!

In order to be scientific, one cannot simply look backwards and cherrypick data that disagrees with the original hypothesis to prove the premise behind the original hypothesis true (people lying to pollsters).  If one starts out pre-election stating that Obama should actually result significantly (he'll actually end up with 1.5%?)  below his polled result for X and Y and Z reasons and those all are proven wrong by the actual result... to go into a less-polled state (out of 50 - bound to be some outliers!!!!) and claim that oh, you were wrong initially but your argument actually applies here and here is so intellectually dishonest it's mindblowing.