A couple weeks back I received and email from the dialoguepolitics group regarding a PBS run Sarah Palin poll. They were asking all the democrats to go in and vote no that she is not qualified. For some reason they felt this would make it true. They did manage to bring her yes leads to a tie of 49% yes and 49% no, but did not manage to make it an overwhelming majority did not find her qualified.
At that point I was going to do a blog on why polls don't work, but thought if I did I would get comments that it was just sour grapes, etc, since I am a McCain/Palin supporter. My husband teaches statistics and while I can't explain it as well as he can it comes down to how you ask the question, of whom you ask the question, and sometimes where you ask the question. You can form a question in such a way that you can manipulate the answer. Like - have you stopped beating your wife? So,in the meantime....
...the AFA countered by running a poll on whether Barack Obama is qualified. The poll at this point stands at 12,848 yes votes, 263,336 no votes and 2,521 not sure votes. And so now we get this email from dialoguepolitics stating the following.
The problem with both polls is that they allow the same person with the same or different emails to vote more than once (I voted twice on each just to find out, and it seemed as though I could have continued), so neither has any real validity, other than as a propaganda tool.
When I went to the AFA poll, 93,000+ had voted that Obama was not qualified, and some 4,000+ felt that he was (I forget the not sure vote).
Plus, and this applies to both polls, what, exactly does "qualified" mean. My guess is that people on both sides are voting whether they want the candidate in question, qualifications or lack thereof aside."
So I want to thank them for making my point about polls. They aren't reliable, they aren't true, and they can be skewed by who is doing the polling. Please keep this in mind with future polls and don't be so quick to believe it just because someone at a polling place says it's true. Thanks dialoguepolitics for making this point for me. I agree the polls are rarely if ever accurate due to the many ways of skewing the responses.
Subscribe to:
Post Comments (Atom)
3 comments:
I just taught a unit on graphs and tables. One of the lessons that many of the other teachers said they skipped because of "time constraints" was "Misleading Statistics". I asked how could they do that because we need to know how to discern those.
You are absolutely right about the particular internet polls discussed. Every internet poll is skewed because they have a limited sample, i.e. only people who use the internet.
But what you described points out something else. On the test I asked what was wrong with an online poll about whether to have online school classes. Some didn't have the foggiest idea but of those who had reasonable answers most gave the expected one that it's an internet poll about the internet. But one girl wrote that because it was online you could vote more than once. And I marked hers correct too.
We also talked about the fact that the old fashioned polls over the phone no longer have an adequate sample either since so many young people, like my son Ian, only have cell phones, and likely internet, but no land lines.
But between the differing polls, there are many ways to have polls that work better than the ones you describe. And although I still think there is SOME margin of error, it is 95% less than the polls you are quoting here. And the CNN poll of polls is even more accurate. Ask Keith. He TAUGHT ME statistics. LOL.
But the bottom line is that it comes down to sample size and random samples, and the real polls take all of those things into account.
Yeah, I'm hoping Keith will have time between classes and office hours to post something on this one, since he was the one who explained it all to me. And he can say it better than me, but I got the general idea of what he was saying. I thought he might respond eot my football and chess one too :-).
PS...now everyone knows I married your college professor! perhaps we should explain it was some 2o+ years after you graduated :-).
Post a Comment