American Statistical Association
Offshore Oil Spills, Gulf of Mexico, Anderson-Darling Test, Weibull, Goodness of Fit, EDF Methods
Eschenbach and Harper (2006) analyzed offshore oil spills in the Gulf of Mexico with extensions to the northern seas of Alaska. This involved multiple methods including assessing what statistical distribution adequately fits the data. Empirical distribution function (EDF) statistical procedures are powerful goodness of fit tests and also provide for good visual assessments. The most powerful of the current EDF methods is the Anderson-Darling test. This paper focuses on the Anderson-Darling EDF goodness of fit procedure for both the two and three parameter Weibull distribution that is often used in reliability analysis. Excel VBA code has been developed to compute this test statistic and also the associated p-values to allow statistical significance tests. The Excel routines are available free at http://faculty.otterbein.edu/WHarper/. The functions are illustrated with Gulf of Mexico oil spill data.
Harper, William V.; James, Thomas R.; and Eschenbach, Ted G., "Reliability Goodness of Fit for Oil Spills in the Gulf of Mexico" (2009). Mathematics Faculty Scholarship. 19.
Harper, William V., Ted G. Eschenbach, Thomas R. James, (2009), "Reliability Goodness of Fit for Oil Spills in the Gulf of Mexico", JSM Proceedings, Statistics and the Environment Section, pp. 618-626, American Statistical Association, Washington, DC.