Esteemed list members:
I like to think that I'm pretty adept with statistics of description, but
well know that I know little about statistics of inference. Thus, can anyone
explain the rationale for
1) obtaining an opportunity sample from a hypothetical (and indeed undefined)
population, while 2) going on to use the calculus of probability --- which
(as I understand things at least) assumes a *probabilty* sample from an
existing *known* and *finite* population --- in order to 3) compute p -
values and then 4) to use those of 0.01 or 0.05 for identifying meaningful
outcomes from which 4) to impressionistically generalize to some some reality
supposedly "out there"?
I've seen that done (and wondered about it) a *lot.* But --- as you may
have guessed --- I've now got a specific research report in mind One which
I intend to criticize on purely common sense grounds, but would like to know
if it can be faulted on the aforegoing as well.
Harley Upchurch
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|