More recently Quesenberry has shown that control limits obtained from samples
of this size produce SPC procedures with Run Length (RL) distributions significantly
worse than would be expected. A number of articles (Nedumaran & Pignatiello(2001),
Tsai Et al (August 2005) and Albers & Kallenberg (December 2000,December 2003))
have since appeared, each proposing a different method of calculating chart control limits
for Shewart Ccharts that will produce desirable in-control RL characteristics while
minimizing training sample size. The out-of-control performance for the above plans,
however, was only addressed in one of the articles. In addition, the different authors
employed differing performance measures. Among these are percentiles of the
conditional RL distribution, percentiles of marginal RL distribution, and exceedance