[B5JMS] And So It Begins...

[b5jms at cs.columbia.edu]

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
From: Robert Perkins <rob_perkins at hotmail.com>
Date: Wed, 28 May 2003 05:08:37 GMT
Lines: 41

Coupla comments, directed more or less at everyone, in the hopes
there's an answer or two:

On 27 May 2003 23:45:20 GMT, jmsatb5 at aol.com (Jms at B5) wrote:

>First, the sampling is both small and highly anecdotal.  

Why isn't that the case with a sample size of 900 or so in a
population of several million voters, when polls are the fodder of
headlines leading up to the election? (A practice which I abhor, by
the way. The polling should take place at the polls, not at the news
desks of television broadcasters or newspaper editors.)

>Further to the point, where the reporters may or may not vote one way or
>another, the people who *control* those publishing and TV companies are highly
>conservative.  Witness the lack of liberal news commentators/talk shows and the
>propensity of conservative ones.

People say that it's been tried, without success. Donahue had a go of
it just recently, and I recall that it didn't get ratings or
something, for example. For some reason liberalism doesn't sell, or
something, goes the logic. But if, for crying out loud, you can't tell
which side of the fence remarkably influential people like Peter
Jennings and Dan Rather are, you're just not lookin'. (It's either
that or those two are just overwhelmingly condescending and elitist, I
can't tell which, since every time *I* watch them I get the feeling
they think I'm really stupid or something. And how could that be true
if I'm a B5 fan? ;-) )

>So bottom line...it's a dubious statistic, from a tiny sub-set, which does not
>have any provable associated bias in past situations, and is thus for the most
>part fairly meaningless...

'fraid JMS is right on this one, folks (at least IMO). But the same is
true for most reported polling, (also IMO). Unless someone can explain
differently *why* tiny samples of a couple thousand are supposed to
represent the Whole Country. 

Rob



=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
From: jmsatb5 at aol.com (Jms at B5)
Date: 29 May 2003 04:54:34 GMT
Lines: 41

>>So bottom line...it's a dubious statistic, from a tiny sub-set, which does
>not
>>have any provable associated bias in past situations, and is thus for the
>most
>>part fairly meaningless...
>
>'fraid JMS is right on this one, folks (at least IMO). But the same is
>true for most reported polling, (also IMO). Unless someone can explain
>differently *why* tiny samples of a couple thousand are supposed to
>represent the Whole Country. 
>
>

You're talking apples and oranges.

First, a correctly done scientific survey has controls in the way that the
questions are asked, so that the do not skew the data one  way or the other;
they are often numerically weighted (on a scale of one to four, how do you feel
about a, b or c?), and there is a statistical range of error depending on the
size of the sample universe.

The more people, the smaller the possibility of error; the fewer the people,
the larger the possibility of error.

Asking your buddies down the hall, in an uncontrolled survey, with lots of
variables, isn't a valid survey by any stretch of the imagination.

jms
b.a. clinical psychology
b.a. sociology

 jms

(jmsatb5 at aol.com)
(all message content (c) 2003 by synthetic worlds, ltd., 
permission to reprint specifically denied to SFX Magazine 
and don't send me story ideas)