[lit-ideas] Re: interaction of polls and public opinion

  • From: "John McCreery" <john.mccreery@xxxxxxxxx>
  • To: lit-ideas@xxxxxxxxxxxxx
  • Date: Sun, 12 Nov 2006 13:55:34 +0900

On 11/12/06, Eric Yost <eyost1132@xxxxxxxxxxxxx> wrote:

Seems that if you couldn't first isolate the process,
assuming it exists, you wouldn't be able to discuss how it
interacts with the other contexts and spins that you've

Of course. Not saying it can't be done. The experimental design would
be the one used in test marketing: Choose two distinct audiences
sufficiently isolated from each other to preclude interaction but
otherwise as identical as possible in other relevant respects. Then
broadcast the same programs with the poll included for one but not
included for the other, then look for some measurable difference.

But even running through these criteria shows how tricky this can be.
What, for example, happens to the time that the poll occupies in one
version of the broadcast but not in the other? Will the filler,
whatever it is, skew the results in some unexpected way?

Hard to say. That is why these types of designs are mainly used for
simple will-it-fly? tests of, say, a new variation on McDonald's
hamburgers, where failure in a test market is cheaper than failure in
a nationwide roll-out or to pre-test commercials, e.g., a new proposal
vs. existing commercials or two new proposals against each other.


John McCreery
The Word Works, Ltd., Yokohama, JAPAN

Register to Vote in '06 Elections
To change your Lit-Ideas settings (subscribe/unsub, vacation on/off,
digest on/off), visit www.andreas.com/faq-lit-ideas.html

Other related posts: