Attribute vs. variable data

[Steven_S_Prevette@RL.gov]

> QUESTION:  "In a production line, they measure the efficiency by first
> counting how 
> many bags of product were produced in a day, how many bags they should
> have 
> produced according to the machine specifications, and then divide the
> real 
> into the ideal amount of bags produced "  Is this a binomial (p or np
> chart) model?
> 
> 
> There are two reasons that this does not fit the binomial model 
> 
> 1.  It is possible to exceed 100% efficiency.  In a binomial model, no
> result can even theoretically exceed 100%
> 
> 2.  In a binomial model, each trial is independent from each other.
> We do not have a "specification" bag number one, and did we produce it
> or not, then a "specification" bag number two, and did we produce it
> or not.  Instead, we produce all of the bags (in sequence) that we
> produce, and the remainder (difference between specification and
> actual) are all not produced.  I believe this violates the assumption
> of independence between trials.
> 
> The true test is what happens when you plot the data and use the
> p-chart control limits.  I bet you will find the data is either "out
> of control" or is too tightly grouped around the average line.  I
> would use the x-chart.
> 
> Steve Prevette
> FDH ESH&Q Planning and Evaluation
> ASQ Certified Quality Engineer
> steven_s_prevette@rl.gov
> 509-373-9371
========================================================================