RE: "Paddle Effect"

["Prevette, Steven S" <Steven_S_Prevette@rl.gov>]

What we are dealing with here is an issue of mechanical sampling versus
random sampling.  In true random sampling, each bead in the incoming
supply would have an equally likely chance of falling into any given
hole in the paddle.  However, this is a mechanical process, and there
are significant mechanical effects that can occur.

1.  The red beads and the white beads are different.  Due to whatever
property gives them their color, they may have a different surface
texture or a different density.  If the surface texture is different,
this may affect the likelihood of a bead to fall in a hole.  If the
densities are different, the heavier beads will gravitate to the bottom
of the bead supply, and if the bead supply is not remixed during the
session, this could effect the rate of red beads.

2.  REAL STORY (I observed this myself) - the red and white beads may
collect static charge differently.  One morning I observed Phil Monroe
getting 100% white beads over and over on trial runs.  The beads did not
behave "properly" until someone found a spray can of anti-cling fabric
spray.

3.  The paddles involved have been made of different materials.  The
interaction between the paddle and the beads may vary.

A point to the story is that for each combination of paddles and beads
the average was stable and predictable.  But as you changed
combinations, the average changed, and the differences between the
averages were significant. 

Dr. Deming usually told the "ore ship" story after the red beads, about
sampling an ore transport after crossing the Pacific.


Steve Prevette
Occupational Safety and Health
Fluor Hanford, A Fluor Government Group Project
ASQ Certified Quality Engineer
steven_s_prevette@rl.gov
509-373-9371