Samir
May 3rd, 2004, 05:43 PM
In reliability / accelerated life testing I get Anderson Darlings Goodness of Fit (Adj) values for different distribution. While the distribution with smallest Aderson Darling value is best fit among all those distribution, how can we conclude that even that distribution is not the best fit and one should select Non-parametric distribution. Is there a threshold value for it?