Introduction to Measuring Effect Size (1 of 2)
As explained in another
section, a
probability value is not a measure of the size of an effect; it is
a measure of how unlikely the data or more extreme data are under the
assumption that the
null hypothesis is true.
If the
dependent variable is measured in units
that are meaningful in their own right, then the measurement of the size
of the effect can be as simple as the difference between means. For example,
if an experimenter were interested in how long it takes people to figure
out how to use various designs for an automated teller machine (ATM),
then measurement of the size of the effect would not pose much of a problem.
If, on average, people can figure out how to use one design for an ATM
in 100 seconds and an alternate design in 145 seconds, then it is clear
that the effect size is 45 seconds. Or, if one wished, one could say that
the latter design takes 45% longer than the former for users to figure
out. Unfortunately, most of the dependent variables used in behavioral
research are more difficult to interpret.