February, 2003
© 2000, 2001, 2003 Scott S. Emerson, M.D., Ph.D.
It is
necessary to refine the (usually vaguely stated) scientific
hypotheses into a form that lends itself to statistical analysis
It is
useful to consider the hierarchy of refinements to the scientific
question that are necessary to obtain a statistical
hypothesis
·Deterministic: Does it work?
·Probability model: What proportion of the time does
it work?
·Bayesian: What is the probability that it works most
the time?
·Frequentist: If it didn’t work most the time, would
we see data like this?