A few months back, I was called by a political polling organisation. I was rather busy at the time* and anyway I hate political polls because they are, frankly, full of shit. So I told the nice young lady on the phone that I wasn't interested in taking part in her charming poll and hung up.
Now, I'm not exactly normal, but I can't be the only one to do this. So I'm just wondering how many people might be telling the pollsters to, well, shove it, and what the distribution thereof might be.
One thing I remember from my introductory stats class in college is that this (along with people lying) is a maor issue in conducting any poll, and there's basically nothing statisticians can do about it, though from what demographic data they do have, there isn't that much correlation between refusing to answer and class or political alignment, though I suspect that older people are more likely to respond for the same reasons that they are more likely to respond to a fundraising letter (whether or not they give money).
Posted by: John A. Kalb at Monday, August 09 2004 11:24 AM (+AHyD)
But it does make you wonder if that is not the reason that they perform these polls on saturdays or at dinnertime or just when you are about to woo that special someone, hypothetically speaking, of course. Indeed, in those cses, only those that REALLY feel strongly about the issue will take part. That is what i have always taken from these Gallup polls, NYT polls, WTF polls, etc. that they are specially designed either through wording and subversive answer chioce, or timing to glean the result that they want to hear. That coupled with the fact that 89.423% of all statisics are made up, leads me to agree with the Reverend that in most all cases these polls are merely a way for an interested party to degauss their own coils in a seemingly official manner.
Posted by: tommy at Monday, August 09 2004 05:10 PM (pMauK)
I've been polled over the phone before, but they were never political polls -- just consumer trend polls.
The only political polls I've seen were sent to me through the mail from partisan sources (the Democratic Party, the ACLU... funny, I don't think I've gotten a "conservative poll" in the mail).
Anyway, as you can imagine, these polls were worded to have the politically correct answer(s) and the politically wrong answer(s). But the CORRECT -- to my mind anyway -- answer was never included.
Posted by: Tuning Spork at Monday, August 09 2004 07:38 PM (HCv3/)
I have read, multiple places, that pollsters have found that Republicans are under-sampled in recent years because they're more likely to screen calls or refuse to be polled, so all the polling houses, or all the big ones anyway, try to balance their results accordingly.
Not sure how exactly they do it though.
Posted by: Dean Esmay at Tuesday, August 10 2004 03:26 AM (T7xta)
It is called "a refusal", and the refusal rate problem is the big dark secret of the polling companies.
There is no way to tell whether there is any correlation between "refusing" and ANYTHING the poll is attempting to measure.
That fact destroys the statistical validity of the poll, which depends upon a random sampling of the population.
There have been discussions that the "refusal rate" may be as high as 50% in some US polls.
Consider also the use of answering machines and Caller ID units to screen all phone calls. Is that usage likely to be correlated to income, education, or other demographic data?
The error bounds of modern polls are much larger than the pollsters are willing to admit.
Posted by: Any A. Mouse at Thursday, August 19 2004 06:21 PM (kCb5q)
| Add Comment