| View previous topic :: View next topic |
| Author |
Message |
CopperTop
Joined: 13 Jul 2006
Posts: 127
Location: South of Next Tuesday
|
 Posted: Tue Aug 08, 2006 11:38 pm Post subject: |
 |
|
My guy's father helped build the World Trade Center towers. The design of those buildings was such that if the center of the tower somehow was damaged to the point of destablization, they would fall inward, as if they were imploded. That would explain why people get the impression they were taken down with explosive charges.
Any of you that have gone out an visited http://www.isagittarius.com/photo.html can see that Ralph takes a lot of photos. He had his camera with him on September 11, 2001 as he was riding the express bus to work in Manhattan. From the top of the Gowanus Expressway, he noticed the odd sight of an airplane where it shouldn't have been and took a photo. Moments later, the plane smashed into the WTC.
_________________
The two most abuntant elements in the universe are hydrogen and stupidity - Harlan Ellison |
|
| Back to top |
|
 |
Teh Digital Dragon
Joined: 09 Jul 2006
Posts: 1888
Location: THE WORLD OF LARNING.
|
| Posted: Wed Aug 09, 2006 2:54 am Post subject: Re: DEMOCRACY, BAD FOR DEMOCRACY?? |
|
|
| Sam wrote: | 
|
I love the way it's totally not bias if you phrase it as a question.
“Is the Bush administration exemplary, or merely outstanding?”
“The Democratic party, a tool of Satan?”
“Michael Moore: Fat because HE EATS NEWBORN BABIES!?!?!”
_________________
"You cannot run away from a weakness; you must sometimes fight it out or perish. And if that be so, why not now, and where you stand?" |
|
| Back to top |
|
|
WheelsOfConfusion
Joined: 09 Jul 2006
Posts: 11223
Location: Unknown Kaddath
|
| Posted: Wed Aug 09, 2006 3:00 am Post subject: |
|
|
| Silly Dragon, it's not bias if you are providing a balance against Teh Liburul Mediyuh. |
|
| Back to top |
|
|
mouse
Joined: 10 Jul 2006
Posts: 15622
Location: under the bed
|
| Posted: Wed Aug 09, 2006 11:09 pm Post subject: Re: so basically a fairly reliable sampling method says we r |
|
|
| Marik wrote: | | Quote: | | so i'm feeling like those surveys are being stretched pretty thin. |
The sample bias is indeed totally worth taking into consideration ('One in two americans who had a residence with a landline that we could contact them at and were stupid enough to participate in phone surveys think that Iraq had WMD' etc) but a four-digit size is sufficient to narrow the response accuracy down to a goodly small margin of error. |
if the sample is carefully selected and properly weighted. careful selection will give you a truly representative sample, but with a small sample size, it takes a huge amount of work to get weights that will be representative, and yet not blow you out of the water on confidence intervals. i know - i am currently working on a population survey for just california (~11% of the us pop) with about 15000, and i'm lucky to get a 3% confidence interval on the question "do you smoke now?". and that's after spending a month working through the weighting issues.
| Marik wrote: | Statistics is pretty fun like that, although it seems counter-intuitive on the surface. These sorts of polls, while far from perfect (making them not unlike any other macro data collection tool for public sentiment) are still a testably worthwhile means towards getting 'snapshots' of public mentality. They'll give fairly actionable results.
A good acid test of their not being totally bupkis is that marketing and political firms use them constantly as a means of data collection. It helps them sell product, helps them figure out what to pander to, or how ahead or behind they are, and all that. |
now, see, i would put the marketing and political use as more of a bupkis indicator. marketing uses a lot of things to tell them what to do - focus groups, and the like. and they can pick pretty tight targets - 18-34 year old males they might be able to sell a truck to, for example.
and the political polls show the critical importance of exactly what questions you ask, and what answers you allow - vide the whole "values voter" thing.
admittedly, i'm going a lot on my gut here. think of all the people you know - do you really believe 1/3 of them (or whatever the number was) believe our own government blew up the towers? i don't know, maybe i live in a dream world - but even the people i know that i believe are certifiable idiots don't go that far.
of course, i really don't know any fox news watchers....i mean, not ones who watch it seriously......
_________________
aka: neverscared! |
|
| Back to top |
|
|
Major Tom
Joined: 09 Jul 2006
Posts: 7562
|
| Posted: Wed Aug 09, 2006 11:19 pm Post subject: Re: so basically a fairly reliable sampling method says we r |
|
|
| mouse wrote: | i know - i am currently working on a population survey for just california (~11% of the us pop) with about 15000, and i'm lucky to get a 3% confidence interval on the question "do you smoke now?". and that's after spending a month working through the weighting issues.
|
just curious -- how many levels are you working with in your weighting scheme?
i think the most elaborate i ever played with was 3 -- age w/in gender w/in ethnicity. that was probably a shopper study for a paper so the geography was fairly tightly defined. |
|
| Back to top |
|
|
Sam
Joined: 09 Jul 2006
Posts: 8883
|
| Posted: Wed Aug 09, 2006 11:21 pm Post subject: or being depressed about. |
|
|
The marketing stuff is a pretty hefty investment, especially for small groups. But it does provide actionable information, otherwise it would have been phased out as an expense -- I'm wagering.
In the political world, both parties are addicted to phone polls. They use them all the time as measurements of their support. The data you get back will have some soft spots, but generally will mean something. If a phone poll gets back one in three respondants saying 9/11 was an inside job, or one in two respondants believing that Iraq had WMD, then it's worth paying attention to. |
|
| Back to top |
|
|
Major Tom
Joined: 09 Jul 2006
Posts: 7562
|
| Posted: Wed Aug 09, 2006 11:26 pm Post subject: Re: or being depressed about. |
|
|
| Sam wrote: | | The data you get back will have some soft spots, but generally will mean something. If a phone poll gets back one in three respondants saying 9/11 was an inside job, or one in two respondants believing that Iraq had WMD, then it's worth paying attention to. |
the amount of attention that should be paid is entirely dependant on the rhetorical structure of the whole survey.
even if a single, issue-driven, pivotal and newsworthy question is written and delivered with the utmost integrity and balance, if the rest of the survey (especially the portion that preceeds the pivotal data) is biased, however subtly, the entire exercize is a partisan tool rather than a trustworthy source of indicators.
[edit] the methods of analysis and reporting also have a great potential to inject bullshit into otherwise meaningful information |
|
| Back to top |
|
|
Sam
Joined: 09 Jul 2006
Posts: 8883
|
| Posted: Wed Aug 09, 2006 11:31 pm Post subject: support among males named sam was like maybe ten percent |
|
|
Reminds me of what I saw in the latest Harper's index.
Polled support among Republicans for "Hillary Clinton" was something like 16%.
Polled support among Republicans for "Hillary Rodham Clinton" was in the mid 20's. |
|
| Back to top |
|
|
mouse
Joined: 10 Jul 2006
Posts: 15622
Location: under the bed
|
| Posted: Thu Aug 10, 2006 9:35 pm Post subject: Re: so basically a fairly reliable sampling method says we r |
|
|
| Major Tom wrote: | | mouse wrote: | i know - i am currently working on a population survey for just california (~11% of the us pop) with about 15000, and i'm lucky to get a 3% confidence interval on the question "do you smoke now?". and that's after spending a month working through the weighting issues.
|
just curious -- how many levels are you working with in your weighting scheme?
i think the most elaborate i ever played with was 3 -- age w/in gender w/in ethnicity. that was probably a shopper study for a paper so the geography was fairly tightly defined. |
we do a two-level one for adults - region X gender and race X age X education (and then those two have to produce reasonable totals when combined). 18 regions, 4 races, 3 groups each for age and education.
and as i think i mentioned before - response rates for phone surveys are dropping like stones (in large part due to those self-same political polls that people are getting tired of) - then you add in answering machines and the fact that you can't do phone surveys on cell phones (which is starting to really bias the younger demographic, as people that age increasingly have just a cell, and no land line)...and well, the population you are talking to is starting to get .....well, unbalanced.
as can be seen by the poll results.
_________________
aka: neverscared! |
|
| Back to top |
|
|
Lasairfiona
Joined: 09 Jul 2006
Posts: 9683
Location: I have to be somewhere? ::runs around frantically::
|
| Posted: Thu Aug 10, 2006 9:37 pm Post subject: |
|
|
Wouldn't email polls work if the emails were only sent to official emails like university accounts? Just an idea. (I don't have a land line and haven't since... 16).
_________________
Before God created Las he pondered on all the aspects a woman might have, he considered which ones would look good super-inflated and which ones to leave alone.
After much deliberation he gave her a giant comfort zone. - Michael |
|
| Back to top |
|
|
mouse
Joined: 10 Jul 2006
Posts: 15622
Location: under the bed
|
| Posted: Thu Aug 10, 2006 9:49 pm Post subject: |
|
|
well, but then you only get the part of the population that is associated with a university - which will tend to be more educated (we devoutly hope), probably better off financially, and skew towards youth (if you include all the student accounts). fine for some things, but if you want to get a picture of what the american population as a whole thinks, you gotta include those poor old stupid people as well.
polling on the internet might work if you could reliably determine the demographics of the people taking the poll, and if you can get around the self-selection bias. seems to me one political poll sam brought up a while back did that - they sent people letters, directing them to the poll website, so should have gotten a reasonably random sample of the population - and hopefully, they collected reasonably accurate demographics as well.
_________________
aka: neverscared! |
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
