On Sunday 24 Mar 2013 14:23:32 Steve Dougherty wrote: > The t-test part of the results I linked in IRC is not valid. I forgot about > the condition that the population be normally distributed, and the reject > percentage distribution plot on the stats page shows that it is not. The > means are still valid. If there were an equivalent probe with an > identifier, we might be able to run the t-test when pairing the samples as > the population becomes the differences.
:( I still think including both reject stats and an identifier is risky. Keeping stats and then returning them for historical periods might be less so, dunno. However we won't see the full story until it's mandatory anyway?
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Devl mailing list [email protected] https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
