[thelist] flash detection

[Mike Hardaker]

> If you want valid stats, you have to get a representative sample.
> If you offer people $5 to participate in a 3-hour survey about wines,
> you'll find that Mad Dog Fifty Fifty is a great value in wines.
>
> Take a look at how they got their sample, and you'll see why
> their numbers are so screwy.

This is sadly familiar - anyone who's dealt with pollsters and their ilk
will know how carefully they dig out what you *want* to hear before
formulating their questions.

------------
Hypothetical Example - We want to know how US citzens feel about Dubya's
approach to China.

Option 1: Should the US stand up to threats from other countries, such as
China? ("83 percent of US citizens support the President's current foreign
policy, according to a recent poll, CNN reported yesterday")

Option 2: Should the US be prepered to engage in nuclear war with China over
Taiwan? ("87 percent of US citizens are opposed to the President's current
foreign policy, an NBC poll exclusively reveals")
------------

This is a glib exaple, because it's so easily dismantled (even if it isn't
that far from the truth). However, in the business - rather than political -
world, my experience suggests that just such loaded questions are
commonplace.

<tip>
If you're a European, be careful about posting messages to evolt late at
night. Many of the list members are just finishing work in California,
caffeine-fuelled and sober as a Judge. You, however, may be burping Scotch
fumes and only at the keyboard because you're trying to avoid "the spins".
This may affect your logic and/or typing ability. You will (should) regret
much of what you have to say when you wake up. On the other hand, you can
*rule* shortly before breakfast...
</tip>

-------------------
Mike Hardaker
Founder & Publisher
AngloINFO
www.angloinfo.com (Web)
wap.angloinfo.com (WAP)