Experience with a criterion for the classification of johnson distributions

Abstract

We use simulation to investigate the choice of percentiles when using the criterion given by Bowman and Shenton (1989) to select between Johnson's SB and STJ distributions for a data set. We also look at the choice of a tolerance level in classifying distributions as Johnson's SL-

Keywords:

• Slifker-Shapiro
• Johnson distributions
• criteria based on percentile differences