What is it with people not being able to read polls?

[enki23]

I see it again, and again, (and again, and again, and again, and...). Someone posts a poll showing some difference that, given even the briefest glance, is *clearly* within the margin of error. Or some poll that has changed by a percentage point either way from a poll taken the day before is touted as showing "movement." Given 95% confidence intervals that routinely span a range of 6 to 10 percent or sometimes more, it should be *obvious* that this is statistically no different from the day before. And that assumes there are no problems with the sampling methodology in the first place.

And not only do people not get it, nobody even seems to notice if you point it out. If it's some Obama/Clinton thing, both sides just merrily go at it, seemingly oblivious to the fact that what they're arguing over doesn't even mean what the asshole who posted it *says* it means. Point it out, and you will simply be ignored. Nobody, I mean damned near *NOBODY* on this site knows even the barest basics about how to read a poll. The closest thing I ever seem to see is somebody saying some variation of "polls suck." Which is at least a sort of naive skepticism. Anyway...

/Rant

[TZ]

1. Its the same deal with people reading scientific journals/studies

They don't bother to look at details like sampling size what statistically is significant etc..They just pluck out the parts that push their agenda.
How many times have we seen a study that shows a slight increase in tumor rates in rats trumpeted as "this substance is teh poison!".
Its all about agenda...Objectivity is non existent in many here.

[salvorhardin]

2. Truth vs. factual

I submit that for many people, "truth" is a moral position and unconcerned with facts.

[LeftishBrit]

3. Yes. They use it quite similarly to the way that it's used in religion.

Faith, not proof.

[salvorhardin]

6. Exactly.

Bush is evil, therefore everything he says, and everything in connection with him (including far flung federal agencies like NIST) must be untrue. It's the same shortcuts in reasoning everybody takes but backed up by moral certitude.

[realisticphish]

4. truthiness

[salvorhardin]

5. That's something a little bit different I think

Truthiness is employed by those who want to manipulate others based on the moral positions with which they ascertain truth value. It's what Harry Frankfurt was talking about in On Bullshit; statements which are said sincerely and sound true but are totally unconcerned with the actual truth.