do not attempt to analyze polling data gleamed from 311 respondents...

that is what is known in the trade as a focus group.

gleaned.

And while you are at it, call one of your polling people you name drop and ask them to explain to you margin of error.

It even works when the party doing it is responsible for a disastrous, unending war and a global economic meltdown.

The Democrats have got to answer these attacks quickly and decisively, and then they have to move against Palin. They can't make the mistake of thinking that she's just going to make a fool out of herself the more she talks. While that's partly true, her attacks are focused and relentless, and they appear to be sticking among independent voters. (Very recent tracking polls have shown a trend towards McCain.) The public is NOT going to see through them--remember, that's what Kerry thought about the Swift Boat Veterans' lies, but the election proved him disastrously wrong.

In addition to responding directly to Palin's attacks, the Democrats need to go after her traitor husband and her own support of treason in the form of the Alaska Independence Party.

That Sarah needs to stop putting on airs

With this whole Ayers travesty.

if you look at any negative ad, it will drive down the positives of both the target and creator of the ad...

there is no evidence that there will be any long term negative impact on the target.

The new eskimo jihadists?

Okay, statistical significance is a can of worms in itself, and not worthy of discussion here except to say that it's a total fallacy in almost all instances.

That said, however, when someone reports a finding as "statistically significant at the .8 level", it usually means that their findings were so weak that they're essentially meaningless.

Is there an effect here? Perhaps. It needs a higher-power study (with a greater sample size) if you want to say anything meaningful. The basic result to be gleaned from all this is that the researchers in question don't know how to conduct a proper experiment to save their own lives.

do the due diligence...

http://www.dailykos.com/story/2008/9/17/10152/8380/600/601474

according to Media curves and HCD

all negative attacks are effective.

But that is what HCD is there to do, tell their clients their ads will work.

As political ads have short shelf life, there is no way to hold these companies accountable for cooking their data.

I've been going on about this in other threads all day, but could we please have some coverage of the assassination rhetoric going on at these McCain/Palin events?!

Supporters or rabble-rousers or nut-jobs in the crowds shouting out "Traitor!" or "Kill him!" in reference to the potential next POTUS. And McCain/Palin smirking, but otherwise ignoring a real, dangerous sentiment.

One the one hand, it's like Borat come to life (I keep wondering if the crowd will spontaneously break out in "Throw the Jew Down the Well" or its anti-Obama equivalent). On the other hand, McCain/Palin's rhetoric could very plausibly get Obama killed, or at least targeted. It's not a game. And McCain/Palin should be held responsible for such incitement.

Whoever conducted this poll is pulling a fast one. The accepted standard of statistical significance is 95% or greater. Shoot the messenger on this one. The poll is baloney.

The conventional wisdom on attack ads is they do damage to both attacker and target. The attacker gambles that the target gets hurt more. This supports that belief. Now if these studies could tell targets how to blunt the attacks. Obviously McCain picked the Ayers attack on purpose.

Though I still maintain thee personal attacks have already done their damage, or almost all. McCain has nothing left. Obama it appears still has dirt on McCain, so even if this is all McCain has, I think it's a loser for him.

only Christians need apply for public office.

This is in direct conflict with the Constitution.

Instant reaction upticks negatively on Obama.

Viewer learns afterward from numerous other sources that Sarah Palin has lied and obfuscated beyond all credibility and perception is so adjusted.

Instant reaction upticks in terms of "phew, I smell a rancid old man and his lying chunk of witless eye-candy."

Viewer has a chance to see more of this and has that opinion reinforced until the sight of John McCain and Sarah Palin triggers an instant gag-reflex.

I'm just sayin". Time and events will weigh far more heavily on McNasty and the foul mouthpiece.

"Statistically signfiicant at an 80% confidence level" is a very fancy way to say, "Not statististally significant."

The standard level that people pay attention to is 95%. For example, those margins of error cited for virtually every poll are the 95% confidence interval.

99% is considered a strong case. 99.9% is considered a very strong case.

90% can be mentioned, but is generally considered nothing more than a hopeful sign that a stronger result can be found later. It's a generous measure, one that allows effects with the potential to prove meaningful to come to people's attention.

Get it?

99.9% = very strong

99% = strong

95% = credible

90% = there might be something there, so try again and let us know

80%? 80% that it did not pass any of these thresholds. Any serious report on this reseasrch (e.g. article in a journal, conference presentation, class assignment) would call that a failure to find statistically significant confirmation.

80%? 80% is a joke. Claiming that it is meaningful should kill your credibility.

Mind you, these are not *my* standards. In fact, I would be a bit stricter than this. You see, the %'s reflect the chance that this result is not simply a function of good or bad luck, and assume ideal conditions and true randomness. Real world conditions actually make it worse -- which is what the folks at FiveThirtyEight.com refer to as pollster introduced error. And so, one in twenty that pass the 95% level actually aren't true. But I don't set the standards.

Let this be a lesson to you, Alex. Keep these numbers in mind, else your own credibilty might suffer, as well.