A question about polls in general.

[Liberal Veteran]

I know I am one of those people that simply NEVER answer the phone. I have CallerID and an answering machine and voice mail and I simply don't bother to answer the phone figuring that if it's important they will leave a message.

I know a lot of people who do the same or at the very least never pick up unless they know who is calling.

Then we have the younger generation who only have cell phones now.

What sort of impact is that having on the overall nature of polling?

[NMDemDist2]

1. lots and lots IMHO n/t

[unblock]

2. they can skew the results, obviously

however, responsible polling companies occassionally do very thorough door-to-door polling over a longer period of time and correlate the results with simultaneous phone polls, to see if the results significantly differ.

if they do, on certain kinds of questions, they make statistical adjustments to the results of the phone polls.


alternatively, you correlate the results to the census and make demographic adjustments. e.g., say you know that 18-25 year olds are 15% of the entire population and 45-65 year old are also 15% of the population. your poll contacts 200 45-65 year old but only 100 18-25 year olds. well, you simply normalize the results so that the 100 youngsters carry equal weight to the 200 older folks.

there are many good ways to control for built-in poll bias.

now, the real question is: ARE there any responsible polling companies out there???

[lapfog_1]

3. Depends on whether or not the population of people

like yourself, who either don't answer polls or are otherwise
not participating in the polls are significantly different
than the population that participates. They TRY to take care
of this in the "margin of error". But I suspect that it isn't
that easy. See, polls do better with more participants OR with
more representative participants, and the margin of error attempts
to account both for the sample size AND the variance of
representation. It is also the case that the more participants,
the more expensive the poll (more people calling, more data to
enter, etc), so sometimes these things get out of whack.

Truth is, I don't think they really know. I think the margin
of error could be 50 percent bigger than they state... so instead
of a +/- 4 percent, it's more like +/- 6 percent. I'm sure the
reputable polls try to get the statistics right, but it's just
damn difficult without spending a lot more money.

[Shoeempress]

4. Don't forget unlisted phone #s