Alternative Approaches to Interpreting Effect size (2 of 2)
Keep in mind, however, that, just as with any measure of effect size,
it is not possible to specify how big the effect must be in order to be
important without considering the context in which the effect is going
to be used. Just as with the measures of variance
explained, there are cases for which an effect that has very small
d' can have a substantial practical effect.
Graphical Methods
It is obviously difficult to summarize the size of an association with
a single number. A better approach is to use an appropriate graphical
display to indicate the size of an effect. When differences in the central
tendency of two or more groups is at issue, then side by side
box plots is an excellent way to portray the group differences. The
differences between measures of central tendency and the amount of overlap
among the groups is readily apparent. Scatterplots
provide a useful way to portray the size of a relationship between two
quantitative variables. Too often researchers
rely solely on the magnitude of the correlation
coefficient without viewing a scatterplot in their assessment of
the size of a relationship.
Other Resources
There are many issues involving effect sizes that are beyond the scope of this
book. The reader is referred to the book by Rosenthal,
Rosnow, and Rubin for more information.
End of book!!!