Discussion Groups / English Usage / January 2009

I am sorry that I have posted this again.  I wondered if my first
question was not accepted.

LP

They are the same, in reality.  If you understood statistics, you
would
know that any attempt to "cook the books" by forcing what you believe
is "representative" is to introduce a bias and make the "cross-
section"
less than "random".

GFH

I don't understand those who say these are the same.   They might be
sometimes but in the area of political polling, in the USA and I would
expect everywhere, random sampling is not as often used anymore,
because pollsters found they could get better results, those that more
closely matched the results on election day, if they made efforts to
find a more representative sample.   And though they only once in a
while go into the details of this, most election projection polls are
not just random.  AFAIK, most other large polls are not either, such
as the ones that ask "Is the economy looking better next year than
last year?:", etc.

Polls taken in person are liked because not everyone has a phone.  But
where does one put poll takers to sample every sort of the population.
Regarding phone calls, randommness might allow a poll taker to ask
queestions only during the normal 9-5. M-F working hours, but unless
one is able to talk to people at work at the same rate as people at
home, that will skew the results.  Randomness might allow polltakers
to call whenever they want to.  Let them sleep late if they are
inclinded, take off at lunch hour to feed themselves or their
children, take a nap in the afternoon.

Instead the method for the last 30 years and probably much longer has
been Ithink to make sure a certain percentage of calls are placed
during each time period, maybe 9-11, 11-1, 1-3, 3-5, 5-7, 7-9, and a
certain percentage to each phone exchange, counting area code the
first 6 digits.

Now that one in 6 homes don't have land lines in the US, other methods
are trying to be found to catch cell-phone only users.

Just dialing phone numbers at random, like in sequential order, might
work for this but I think since so many numbers are not in use, this
isn't done.  Not sure.

This isn't just a statistics questoin but a polling question.   There
is plenty written on it, and probably a whole Yahoo list on it.

Even if the two kinds of polls would give the same result, and even if
one disapproves of one of the two kinds, they are not the same.

P&M