What SD numbers do you use for this?I agree they should be different, but I've not seen much researchevidence on what the correct numbers are.
Also SDs are about expected errors and do not cover blunderprobability