Rebuilding Iraq - Newsweek Poll - LOOK At The Disparity In Numbers...

Between the MSNBC internet polling vs. the actual phone polling they did. Almost a total opposite reflection. I know we see this all the time, but just who are they calling on the phone???

And since we've seen plenty an internet polls get freeeped, where the hell are they here???

And with an internet sample of 10,492 responses as of this post (as opposed to the 1,011 phone calls Newsweek made), is this STILL an illegitimate sampling of the public???

:shrug:

Results Link: http://www.msnbc.com/news/956235.asp#survey

Article Link: http://www.msnbc.com/news/956458.asp?0cv=CB20 (Poll link halfway down)

Hmmmm......

the first being: Who do they call for their poll responses?

the second is: Is this disparity defined by those who use the internet and are therefore more informed of the real situation?

and the third is simply a curiousity: What does "undefined %" mean?

that those options weren't in that poll.

Interesting, isn't it, if they weren't available on the actual vs internet poll. Brings up another question then. Why weren't they?

* has a 20% approval, 66% are not at all confident the Iraq situation will stabilize and 75% oppose the invasion

Yes, it presumably is, because the sample is self-selected. The legitimacy of a poll is not purely a function of the sample size, but of the relationship between the sample and the population to which you wish to generalize. We actually have no idea of who the respondents are in the online poll, while the pollsters who did the original poll presumably did a random sampling of a carefully defined population (e.g. likely voters) in order to draw inferences about that population. When you get over 1000 respondents in a well-done poll, all you do is tighten down the margin of error. Since the MoE decreases as a function of the square root of the sample size, you have to increase the sample size by a factor of 10 in order to halve the MoE. Above about 1000, with a MoE of about 3, it ceases to be worth the effort for most purposes, except maybe on election night when you wnat to get the MoE down to a VERY small percentage.

It seems lately more than a few 'legitimate' polls (let alone outright VOTING) have been called into question. Just wondering who\what to believe any more, ya know???

:hi:

On edit: Fixed my subject line for spelling, a first, LOL!!!

According to Newsweek 36% of the population does not approve of Bush. The population is about 280,000,000, but excluding children under 12 and the institutionalized, figure about 200,000,000. That means that 72,000,000 do not approve of Bush. If you conduct a poll which somehow chooses people from that very large group, then you can easily get far different results, even if you have a million samples. On the other hand 1,000 seems like such a small sample size that you could easily have a fluke and contact a majority from the 100,000,000 idiots who approve of Bush. It is like tossing 200,000,000 pennies in the air and then randomly picking out a thousand. Wouldn't it be very easy to pick out 700 heads instead of 500?

For example, the prevalence of answering machines and caller ID, and the fact that many people won't even pick up their phone unless they know it's someone they want to speak with, could be skewing the sample. So might the move to cellphones. You can't poll someone if they don't answer.

I haven't seen any statistical study of this idea -- so far as I know, it's still just speculation. But it does seem possible that the "random samplings" of phone surveys aren't random anymore -- that they tend to be comprised of the more conservative, less tech-savvy, and generally more out-of-touch.

Thanks!

:hi:

trouble getting through than for the 2000 election.

:kick: