Deming Experiments

Variation using the Quincunx Experiment
    This experiment is conducted with a series of pegs.  Balls have an equal (50%) chance of falling each direction.  The bottom line is that overall the long haul, half of the balls with fall each direction at each direction, and the shape of the final output will be bell shaped or normal.  The normal curve has some very nice properties which allows us to make statements of chance or 'probability'.  If we know only two properties of the outcome, the measure of central tendency, called the mean, and a measure of variation, called the standard deviation, we can make statements about the outcome of a process.  A big thing is that every process has some inherent variation, and we should not correct for it.

• variation will follow the normal curve
• do not correct for random variation

In this experiment workers are told exactly how to do a job (do not pick out red balls) but they inevitably pick out some red balls by chance.

    points to ponder

• variation exists in systems, and if stable, can be predicted different people pick out differing numbers of red balls simply by chance
• all the variation in the production of red beads, and the variation from day to day came entirely from the process itself the difference in numbers of red balls chosen was inherent or part of the process
• numerical goals are often meaningless I could scream at you not to pick red balls or reward you for not picking a red ball, but this was completely by chance and it is sending the wrong message to reward you
• management is responsible for the system 85% off quality problems can be traced back to management.  I would have to change the process (let you look at the balls when picking) to improve the overall quality level.