Introduction to Measuring Effect Size (2 of 2)
Next section: Variance explained in ANOVA
Assuming job satisfaction is measured on a seven-point scale, how big an
effect is one that is 1.5 units on the scale? Is it big enough to be important
either theoretically or practically? There is no clear answer to this question.
The most common approach is to interpret the size of the mean difference
relative to the differences occurring within each group. This involves
defining the size of the effect in terms of the degree of overlap between the
groups. Thus, an experiment finding a difference of 1.5 with very little overlap
between groups would be said to have revealed a larger effect than one finding
the same difference of 1.5 but with greater overlap between groups.
Measures
of the size of an effect based on the degree of overlap between groups usually
involve calculating the proportion of the
variance that
can be explained by differences between groups. At one extreme is the case
in which the only differences are differences between groups with all scores
within a group being equal. At the other extreme is the case in which the
group means are equal. In the former case, 100% of the variance is explained
by differences between groups; in the latter case, 0% of the variance is
explained.
Next section: Variance explained in ANOVA