🚀 go-pugleaf

Whoa! Now everyone is asking  "h o w  do I apply transformations
in order to use an ANOVA", without asking "what hypothesis are we
testing?" It seems to me that everyone is trying to reduce the
data to a form in which differences between groups in means
can be examined, but that the first interesting that we found out
is being ignored, viz: that the groups differ in other respects
too! Can I suggest that a test for differences in means may not
be the best way of extracting  p a t t e r n  from the data...
So we're back to the researchers with questions about the model
underlying the process they're investigating. Or, at least, that  {
would be my first reaction. What if the means are all the same but
the distributions are different shapes? Does anyone care? Will this
simply embarass the investigators? And why does a mirror reverse your
left and right side but not your head and feet?
I guess I better get them to review my medication...

r