[blml] Chicago Casebook Posted [SEC=UNOFFICIAL]

[Marvin French]

> Concluding comments by Jeff Goldsmith:
>
> [snip]
>
> >Twice in this set, poll results rather surprised
> >me. I wonder about the procedures used to pick the
> >players for polls. In a regional AC case I heard
> >recently, a poll was taken and the results
> >astonished the whole AC. We asked who was polled;
> >it turned out that the four players consulted could
> >have been argued to be the four most conservative
> >expert players in the room. I don't claim that the
> >directors intentionally chose those four players to
> >get a desired result, but a representative sample
> >is not obtained with a poll size of four, and if
> >the sample is biased, poor results are obtained.
> >Again, I do not intend to imply that the results of
> >the polls in these cases are suspect; in fact, my
> >personal poll upheld the directors' poll results in
> >the one that surprised me the most. In that poll,
> >about half the players thought that the decision to
> >be made was not close. Those players were split
> >evenly between the two choices. AC members take
> >note: even if you have strong feelings that an
> >action is automatic, consider the possibility that
> >other players may feel very differently. Tournament
> >directors (TDs): try to avoid ACs with fewer than
> >five members to attempt to get multiple viewpoints.
>
A course in statistics isn't needed to understand that small sample
sizes have very little significance.  Player polls should never have
been instituted and should be abolished.

Marv
Marvin L. French
San Diego, California
www.marvinfrench.com