Ichingcarpenter
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 08:59 AM
Original message |
About that poll from Washington Post showing Obama behind in Colorado |
|
Edited on Fri Jul-25-08 09:09 AM by Ichingcarpenter
18-54 age group had 626 polled vs 55+ had 763 polled The age and race demographics don't match the state's reality. They over counted the 55+ group also numbers assembled by Dan Meyers, the editor of KCFR's Public Insight Network and the producer and host of the show Colorado Matters, reveal that between December 2003 and December 2007, the margin of registered Republicans over registered Democrats shrank by 57,430. In the same period, the margin of registered Republicans over registered unaffiliated voters shrank by a whopping 123,375. This doesn't even take into account the growth that happened after December 2007 http://upload.wikimedia.org/wikipedia/commons/3/3c/USA_Colorado_age_pyramid.svghttp://www.eredux.com/states/demographics.php?id=1134http://www.washingtonpost.com/wp-srv/politics/interactives/campaign08/battleground-polls/battlegrounds_co_072408.html
|
sinkingfeeling
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 09:01 AM
Response to Original message |
1. Polls don't mean very much. |
Ichingcarpenter
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 09:05 AM
Response to Reply #1 |
2. I'm trying to find the last poll they did on Colorado and looking into |
|
the demographic weighting of the poll.
|
Virginia Dare
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 09:35 AM
Response to Original message |
3. At this point I'm paying attention to the fundraising.. |
|
more than the polls and Obama is blowing McCain away.
McCain is vastly outspending Obama and he can ill afford it at this point. Not to mention all the free pub that Obama is getting..:rofl: Any way you spin it, the McCain camp is in serious trouble right now.
|
Ichingcarpenter
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 09:42 AM
Response to Reply #3 |
4. I agree but I'm just saying that the polls need examined |
|
for reliability and honesty. The polling sample use in Colorado is not honest or reliable.
|
burythehatchet
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 09:59 AM
Response to Reply #4 |
5. WADR, the polls should be ignored. Their ONLY purpose is to allow the |
|
media to justify its' horse race narrative
|
Ichingcarpenter
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:25 AM
Response to Reply #5 |
12. I not ignoring statistical scientific bullshit when I see it. |
|
and neither should the public
|
ProfessorGAC
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:31 AM
Response to Reply #12 |
|
Either is statistical science or it's bullshit. It can't be both! Based upon the stratification as outlined in the OP, i choose the latter.
The Professor
|
burythehatchet
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 11:19 AM
Response to Reply #12 |
17. Agreed. It is important to expose the monkey business. My point is that |
|
we need to ignore polls in larger sense, i.e., stop taking them seriously
|
spanone
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:00 AM
Response to Original message |
6. the polls are being skewed to pretend we have a horserace. otherwise the media would be worthless |
|
which we know they already are
|
Ichingcarpenter
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:03 AM
Response to Reply #6 |
7. I just provided the proof that they were skewed |
|
I doubt the media will do this.
|
ProfessorGAC
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:18 AM
Response to Reply #6 |
10. Exactly What I Think As Well |
|
I think the media is desperate to make this look like a contest because they live for this stuff. No horse race, no ratings. They're freaking out that this might be a runaway, so they're doing everything in the power to make it appear to be razor thin. The Professor
|
On the Road
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:04 AM
Response to Original message |
8. Never Seen Such a Skew Toward Older Voters |
|
but you don't know how they weighted the raw results. If they weighted by party identification, that would probably overstate the Republican vote.
|
librechik
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:12 AM
Response to Original message |
9. Thanks for spotting this--Bastards! |
Ichingcarpenter
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:20 AM
Response to Original message |
11. The Wisconsin Poll is overly weighted in the 55+ favor |
|
Edited on Fri Jul-25-08 10:31 AM by Ichingcarpenter
|
TahitiNut
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 10:27 AM
Response to Original message |
13. "UNWEIGHTED" polling numbers rarely match the demographics of the populations sampled. |
|
That's why professional polling organizations normalize the polling results by 'weighting' the results against known demographics. Indeed, it's in this normalization process that biases of the polling organization are often introduced. For example, Rasmussen is often cited as having a 'Republican bias' due to their alleged normalization assumptions.
|
Ichingcarpenter
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 11:08 AM
Response to Reply #13 |
15. I know how polling works the Margin of error is suppose |
|
to take into consideration for the unweighted portion.
I say it doesn't in these polls.
|
Time for change
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 11:17 AM
Response to Reply #15 |
16. Do you know if the poll result showing Obama behind in this poll is |
|
weighted or unweighted?
If it's weighted, how would we know how the weighting was done? That information doesn't appear to be available from the numbers in the OP.
|
Ichingcarpenter
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 11:26 AM
Response to Reply #16 |
18. You really have to dig into their methodology |
|
which was given at the link. There is a statistical formula that you can use with the margin of error and the unweighted poll sample.
Now the margin of error is supposed to clear up the unweighted sample. However, I don't see that the margin of error is an accurate representative of the sample population.
|
woolldog
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 01:04 PM
Response to Reply #18 |
19. how do you know they didn't reweight it according to demographic |
|
Edited on Fri Jul-25-08 01:04 PM by woolldog
info?
|
Time for change
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 02:00 PM
Response to Reply #18 |
20. A margin of error will generally not clear up an unweighted sample |
|
Edited on Fri Jul-25-08 02:09 PM by Time for change
There are two things in polling that can produce results that are not on target with the actual status of the population being sampled.
One is chance. That is the least important of the two. Its role can be assessed with statistical tests, which give you a margin of error.
The more important reason for poll inaccuracy is bias. That's related to the kind of issue that you bring up in this OP. Bias can be produced, for example, among other possibilities, by over-sampling Republicans or certain age groups. If that was done, the only way to clear that up is to weight the poll for the over-sampling of certain categories. A margin of error will not take care of that.
That being said, I thought that political polls are generally obtained by taking a random sample of the population of interest. If a sample is truly random, then sampling bias does not exist, no weighting is needed, and all of the inaccuracy in the poll can be accounted for by the margin of error. For that reason, it is always best to obtain a random sample if that is possible.
If what Tahiti Nut says is correct, that would mean that the reason for obtaining a non-random sample would be simply that the practicality of the situation makes a random sample impossible to obtain (otherwise there would be no reason not to obtain a random sample). So, in order to eliminate as much bias as possible, the sample would have to be weighted.
Edited to add: I can't tell by looking at the link (I assume you're talking about the 3rd link?) what method they used to weight this poll, if any. All I see is "unweighted" numbers. Does that mean that the poll was not weighted at all? I can't tell. If it wasn't weighted does that mean that they believe that they had close enough to a random sample that it wasn't necessary to weight it? The age and party disparities that you point out suggest that the sampling for the poll was biased rather than random, and that it should have been weighted accordingly. But what they actually did, I can't tell.
|
TexasObserver
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jul-25-08 02:16 PM
Response to Original message |
21. Bingo! I posted earlier this poll would have bad methodology! |
|
I don't recall the thread, but I said that skewing a poll is easy. All you have to do is alter the mix, so that old people are disproportionally represented, and you will give McCain a significant boost in the outcome.
This group of polls was GAMED! They wanted to produce an outcome showing McCain improving, and that's what they got.
Too many Republicans in the poll. Too many old people in the poll. It's INVALID.
|
DU
AdBot (1000+ posts) |
Tue Apr 23rd 2024, 08:49 PM
Response to Original message |