On Friday, February 14, 2003, at 01:52  PM, Wirt Atmar wrote:

> Fred writes:
>
>> On Friday, February 14, 2003, at 07:31  AM, Tom Brandt wrote:
>>
>>> The validity of on-line polls using self-selected responders: 0%
>>
>>  I agree with the idea that they're validity is suspect. One doesn't
>>  need to toss the results into the trash. One can accept them without
>>  believing that they are (even close to) perfect. The 0% you assigned
>>  implies that they're absolutely false. Is that the result of a
>> 1-person
>>  poll? :-)
>
> Beyond the case that robots are responding to the polls, which I'm
> sure is
> not the case, asking about the cross-section of people who are
> responding to
> the poll is to miss the point of the comparison. How polling is
> conducted is
> an obviously legitimate question when only one group is polled, but it
> becomes much less important when multiple groups are polled, so long
> as the
> same methods are applied universally.

Agreed. Randomness is also an important factor. Also, in what the
pollsters consider as random.

For instance, if one does a telephone poll (even if the phone numbers
are randomly generated) the poll would be flawed due to the non-random
distribution of telephone owners. Many seniors, for instance, don't
have phones or live in senior communities which share only one or two
phones.
Similar imbalances exist whenever several people share a single phone.
Also, polling pedestrians minimizes the participation of those who are
stay-at-homes or travel from place-to-place by car. Polling by email or
websites minimizes the participation of the poor.

Even though all polling suffers from these kinds of imperfections, I'd
still like breakdowns by sex, age, education, race religion or whatever
the pollsters decide to provide.

FW

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *