As previously, statistical significance is merely a function of sample size, and in
essence indicates that one was utilizing a large sample. Effect size, on the other hand, is
not affected by sample size. As an example, if one has a sample of 17,000 participants
and a correlation coefficient ofr¼0:03, the result is statically significant atp,0.001.
However, by examining the effect size (r
2
),Xonly explains approximately 9 percent of
the variance in Y, indicating that 91 percent of the variance is unexplained. From a
practical perspective, the relationship is null, meaning theXwas not associated withY
in any meaningful way. This dilemma was present throughout the quantitative studies
published in theJEAandEAQ, which did not report an effect size, or reported effect
size (s) but failed to interpret the reported effect size. A plausible explanation for
researchers reporting but not interpreting effect size is that most computer statistical
packages report multipleR
2
for multiple regression and most researchers include this
effect size without understanding what this really means. Although it is helpful to
report, it would be more useful if the effect size was understood and interpreted by the
researcher to assist the reader in understanding the results in a more complete manner.
Note that the statistics reported in both Tables I and II are subsumed under the general
linear model, which are based on correlations. In turn, variance accounted-for effect
sizes are easily calculated. However, reporting and interpreting effects size was scarce
in the studies examined.