On goodness-of-fit tests for the two-parameter Weibull distribution derived from the stabilized probability plot

Abstract

Michael's statistic D sp , based on the stabilized probability plot, is considered for the assessment of goodness of fit of a two-parameter Weibull distribution. This statistic is shown to have high power when the parameters are estimated using a procedure due to Blom (1958). The correlation coefficient of the stabilized plot, R sp , is also proposed as a test statistic for this situation. Finally a family of transformations is obtained to reduce the null distribution of R sp to standard normality.