Congratulations to chris christie on his landslide victory in NJ.   normally a 5% win margin isn't considered a landslide, but he actually won by 8% when you consider that the Washington Post, New York times and other official media mouthpieces of the Obama organization published polls showing Corzine winning by 3%.  

an 8% variance would have to be considered outside the margin of error for an election poll.   therefore it wasn't an error, but some sort of deliberate manipulation to mislead the public.

and if THAT relatively unimportant poll is being gamed, then what should we conclude from WaPo and NYT polls showing Obama himself at 53% - 3% above neutral?

i withdraw the question.  I'm sure that fine organizations like the WaPo and NYT would NEVER deliberately mislead us about the presidency, even though they're subject to manipulation by the president on governor election polls.   They'd draw the line and uphold their journalistic integrity right?

Slate - thanks for pre-publishing a "slatest" op ed piece yesterday exonerating Obama from any blame for the election losses.  At least YOU showed some integrity in admitting the election debacle the democrats faced.

And by encouraging the Obama organization to pre-deny the election results as totally unrelated to their own policies and behaviors, you're encouraging them to continue those behaviors.

Guess what that implies for 2010 and 2012?

Thanks slate - the republicans appreciate your help.

(full disclosure - BA lives in maryland, is not registered to either political party, and believes Chris Christie ran a poor campaign. Poor campaigns do not portend effective leadership styles.  Good luck NJ.  And corzine: you can stop the daily videos of yourself jogging now - this election was always about having the highest taxes in the nation, and the inability to say no to more taxes - not about christie's waistline.  you have crappy political handlers too - the ones in the White House pulling your strings on campaign positions)

That whole post was certainly built upon a tall stack of "what if" and "I think". Its hard to seriously entertain your conclusions when your premises are totally ungrounded. What evidence is there that the polls were tampered with? Then you assume the alleged tampering was from the White House. Then you assume the White House will protest election results. Next you'll explain how that proves Obama is an alien reptile.

There's a *major* flaw in your logic, mainly in that I don't think you understand what "margin of error" means.

First, just because a result falls outside the margin of error, doesn't necessarily mean that the poll was bad. It means that it is statistically probable that something was wrong. Just how probable depends on how far outside the margin of error the result was. So, while the result being outside the margin of error is a good indicator that there was a problem with the poll, it doesn't *prove* anything.

Second, even assuming the poll was flawed, it *doesn't* mean it was manipulated. It's far more likely that it was simply bad, i.e. poorly designed, poorly worded, used a unrepresentative sample, etc. These problems aren't necessarily intentional or manipulative.

I think you are justified in criticizing the poll as likely flawed, and even raising the *possibility* of manipulation. But asserting that it was definitely "some sort of deliberate manipulation to mislead the public" is an unsupportable statement.

my evidence is statistical - the 8% difference between the polls and voters is more than double the theoretical maximum "margin of error"

pollsters use many tricks to push results in one direction or another. among the most common -

- interviewing "adults" rather than likely voters. if you're interviewing adults who aren't registered to vote, who are registered but haven't voted recently, or don't even live in that state (commuting workers, college students) that is a nifty way to skew results

- "push polls", which begin with asking an inflammatory question like "do you think chris christie, who opposes mammograms for women, should become the state's next governor?"

- pollsters which are "on retainer" for one party or another, or work for a media enterprise which has endorsed a particular candidate

vincent - how do YOU explain an election result so far out of the margin of error?

on what "margin of error" means, and how its used <link>

when a pollster says his results have a margin of error of "X", and he's off by 2X on the election day, either he was lying about (a) the margin of error, (b) his sampling methodology, or (c) the actual poll results themselves

I'm sorry, but you don't seem to understand statistics, *and* you've completely ignored my points. And a Wikipedia article doesn't quite match up to years of work in statistical sampling.

Yes, depending on the confidence interval used (which isn't specified, but 95% is commonly used for polling), a result that is outside the margin of error is statistically unlikely *assuming* there are no flaws in the poll. But that doesn't make it *impossible*, just unlikely, even very unlikely (I would need to know the sample size to know *how* unlikely). Note this quote from another statistics page <link>:

"A 95 percent level of confidence means that 5 percent of the surveys will be off the wall with numbers that don't make much sense. Therefore, if 100 surveys are conducted using the same customer service question, five of them will provide results that are somewhat wacky."

What it should do is to alert you to look for flaws in the poll, not to *assume* them.

But even if we stipulate that the poll was flawed (which, again, seems to be likely, but isn't certain), that does *NOT* mean that it was "manipulated", and this was my main point (which you completely ignored). It is entirely possible that the flaws were due to accident, incompetence, or unknown factors. You have no evidence one way or the other. And that is the main fallacy of your post, you treat flawed as equal to manipulated, which it is *not*.