I remember seeing a sign to this effect at a Mondale rally in '84.  Walter
Mondale was trailing behind badly in the polls behind Reagan and this supporter
of his was using this sign to encourage other supporters into believing the
cause wasn't hopeless.

Of course we can say that this person was a fool. But I was just reflecting
recently on just how many polls I've seen, some daily with hundreds of
questionees.  Should I have been polled by now?  I'm wondering what exactly
would be the mathematically expected rate of phone-ringing for a poll that the
average person should expect.  The rate expected before one can start to believe
the polls are biased against some part of their demographic (such as having no
one home during business hours, being in a non-"Neilsen" neighborhood, etc.).

Post a reply to this message

gregjohn wrote:
> But I was just reflecting
> recently on just how many polls I've seen, some daily with hundreds of
> questionees.  Should I have been polled by now?  I'm wondering what exactly
> would be the mathematically expected rate of phone-ringing for a poll that the
> average person should expect.

I suggest that almost all statistical surveys and polls use only one 
small set of humans. These people answer questions all day so that the 
rest of the general population can do *productive* work. And that is how 
people can afford to commission so many surveys. Rather than go out and 
gather a representative sample of people, they just print out the 
questions, walk down the hall and hand them to the room full of 
question-answer-people.

Also has the nice property that you can decide what result you want, yet 
you can still claim that, legally speaking, it's a valid statistic. :-D

Post a reply to this message