Apropos of nothing...

[gwyn]

In a weird little bit of serendipity, I stumbled on this cool icon from psychodragon82, using lyrics from A Perfect Circle's 3 Libras. I did a Spike vid last year, which I showed at Escapade in February, to this song, although few outside the con or those who picked up our (me and feochadn) tape/dvd will have seen it because I'm not an online vidder (and if you're not an online vidder, you don't count! Oh, did I say that? Bitter? Moi?).

And I see that a poll is being discussed indicating that 75% of Americans favor the war. The poll was made up of 605 people. HOW many people do we have in this country? And they think 605 is representative? Scary thinking this is what will be used to help decide your future -- "see! most of 605 people are in favor of things! And stuff!" Gah.

Keeping my fingers crossed that the Oscars will go on so my little Oscar soiree planning will be worth it. I love snarking over the clothes and the hairdos with a few equally inclined bitchy friends. Makes life worthwhile.

Comments

Polls & Stuff

[melinafandom.livejournal.com]

Man. I studied social science in college, so I spent an unbelievable amount of time learning about crap like sampling and statistics. Ick.

Anyway. If the poll is made up of 605 randomly selected people, then it probably is accurate to within its margin of error, as far as it goes. It's been proven many times that relatively small sample sizes, if properly (randomly!) selected, can yield a statistically accurate sampling, as unbelievable as it sounds. 605 sounds a little small (Gallup usually uses 1100-1200), but that just increases the margin of error, it doesn't render it unreliable -- IF the sample is randomly selected. If it's not, the whole thing goes out the window. Web polls that people volunteer to answer, for example, have no statistical validity at all. (Look! 99.99% of visitors to the FOX News web site are in favor of the war! Well, no shit, Sherlock.)

Being statistically sound doesn't rule out bias, of course, or other problems, such as conducting the poll in a manner that doesn't lead to a truly random sample, or bias inherent in the way the question is worded. Asking "Are you in favor of abortion on demand" will yield a significantly different result than "are you in favor of a woman's right to choose abortion" among the same sample. Professional companies like Gallup, the Field Poll, etc. generally do a decent job coming up with the most "neutral" questions possible. Advocacy groups, whether right or left, usually do the worst. So whether I gave any credence to this particular poll would depend greatly on who conducted it.

I guess it's possible that 75% of Americans really are for the war, I just don't know very many of them. Then again, 40 million people watched Joe Millionaire, and I don't know any of them, either. (At least I think I don't.) It's a truly depressing number in any event.

Re: Polls & Stuff

[gwyn-r.livejournal.com]

Holy cow! That's scary, girl. I went back to check the story and it was changed, so I don't know for certain, but I believe it said CNN/USA Today and one other entity... maybe Gallup, I just don't know. But it's no longer clear in the story, and they're talking about other polls now, with even fewer pollees(?).

I just keep wondering where they find all these people. The Nielsen people, the Gallup people... I don't ever meet them. Ever. I know Seattle's liberalville, but still...

who are these people!?

[melinafandom.livejournal.com]

It's true. I was called by Gallup once, I think -- they were hired by Bridgestone/Firestone to measure how badly people thought about them after the whole Ford debacle. It was kinda interesting. The guy was very professional and I didn't think the questions led me one way or the other.

I think my parents did an Arbitron radio diary for two weeks one time. I've never known anyone who was a Nielsen person.

I went looking for something that would explain sampling in understandable language, instead of academia-speak. I found this at the ABC news site, kinda interesting: http://abcnews.go.com/sections/politics/DailyNews/POLL_EXPLAINER.html