8.8 Retrospective Power
In general the discussion above has focused on a priori power,which is the power that we
would calculate before the experiment is conducted. It is based on reasonable estimates of
means, variances, correlations, proportions, etc. that we believe represent the parameters
for our population or populations. This is what we generally think of when we consider
statistical power.
In recent years there has been an increased interest is what is often called retrospec-
tive (or post hoc) power.For our purposes retrospective power will be defined as power
that is calculated after an experiment has been completed, based on the results of that ex-
periment. (That is why I objected to the use of the phrase “post hoc power” in the G*Power
example—we were calculating power before the experiment was run.) For example, retro-
spective power asks the question “If the values of the population means and variances were
equal to the values found in this experiment, what would be the resulting power?”Section 8.8 Retrospective Power 239Figure 8.4 Main screen from G*Power (version 3.0.8)a priori power
retrospective
(or post hoc)
power