Subject: Re: Unfair TV Ratings - Ron Reagan show (TP/30Something)
From: golchowy@alchemy.chem.utoronto.ca (Gerald Olchowy)
Date: 1991-08-28, 16:37
Newsgroups: rec.arts.tv,alt.tv.twin-peaks

In article <1991Aug28.222140.18463@casbah.acns.nwu.edu> phil@eecs.nwu.edu (William LeFebvre) writes:
> >In article <56693@apple.Apple.COM>, unknown@Apple.COM (Matt Ackeret) writes:
> >
> >|> I'm not saying that the Nielsen families ARE a statistically valid
> >|> sample.. they probably aren't.
> >
> >I don't want to get in to the position of defending A.C. Nielsen, but
> >for what reason do you blithely assume that the Nielsen families
> >"probably aren't" a statistically valid sample?  Do you have any idea
> >at all what method the company uses to select those homes?
> >

The problem with the Nielson sample is probably not the size, but
that the sample is not changed rapidly enough...to have statistical
relevant information, you cannot sample the same miniscule subset
of the population over and over...considering the size of the
population, to be sampling properly, one would expect essentially
zero overlap between the subsequent samples. 


The other problem is whether the sample they choose is really
representative of the TV audience, or only particular segments
of the viewing audience.

Gerald