ProUCL is a good tool. As far as I am aware however, ProUCL outlier identification assessment is undertaken using Dixon or Rosner test methodology (depending on size of dataset).
ESI stats calc uses the Grubbs test
Is there any particular reason why any of these methods should be used over any other? I like ProUCL - mainly because it is free but it seems sensible to be able to use various distributions other than normal - I've never come across anything that is normal. ProUCL seems to be more flexible - it can handle non-detects and the graphical histograms, Q-Q plots etc are very good at visualising the data distribution and spottng those statistical outliers.
|