[Vision2020] Polls 'oversampling' Democrats?

[Art Deco]

As someone who has acted as a consultant in years past to various polling
attempts, I can safely say that almost every poll fails to meet the
standards set for using the Central Tendency Theorem, the
mathematical/statistical basis for accurate sampling.  Some pollsters do
not seem to really understand it at all.

The most obvious error is the lack of a random sample in these times due to
several factors.  The problem is sometimes so bad that the degree of
randomness is not even able to be reasonably predicted.  Therefore, the
confidence value and range of error cannot be accurately calculated.

I haven't done it, but if someone were to Google several of the most
recent, so-called major polls, I'd guess  there would be a range of
predictions.

w.

On Mon, Nov 5, 2012 at 5:14 PM, Scott Dredge <scooterd408 at hotmail.com>wrote:

>  Not surprisingly is that right wing radio / media has been complaining
> about liberal pollsters supposedly 'oversampling' Democratic voters so make
> it look like Obama is winning.  I'd think that polling is fairly accurate
> these days and if the pollsters track record is poor, they wouldn't get
> much credence or attention until they made the necessary adjustments to
> have their predictions more closely matched to results.  I'm looking
> forward to watching all the fun tomorrow night!
>
> My can't miss prediction:  Meet the new boss.  Same as the old boss.
>
> -Scott
>
> =======================================================
>  List services made available by First Step Internet,
>  serving the communities of the Palouse since 1994.
>                http://www.fsr.net
>           mailto:Vision2020 at moscow.com
> =======================================================
>



-- 
Art Deco (Wayne A. Fox)
art.deco.studios at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.fsr.com/pipermail/vision2020/attachments/20121105/e4f68c92/attachment.html>