Commentary on Magnitude Matters
Dwight J Thé, Exercise Science, Syracuse University, Syracuse, NY 13244-5040.. Email. Sportscience 10, 66, 2006 (sportsci.org/2006/djt-inbrief.htm). Published Dec 20. ©2006. |
|
I enjoyed reading the In-Brief item and I think that this is another fine service that you provide. It is interesting that you emphasize the “buzz” surrounding the use of effect sizes, yet in what you say about the Publication Manual of the American Psychological Association (APA manual), I feel that you are understating the case. The 5th edition of the APA manual seems to be very enthusiastic in calling for the reporting of effect sizes. For example, the comment is made that “For the reader to fully understand the importance of your findings, it is almost always necessary to include some index of effect size or strength of relationship in your Results section” (p. 25). Similarly, a page at the APA website states “The 5th edition lists the failure to report effect sizes and the lack of congruence between a study’s operations and the discussion as defects in reporting research”. Also, I would have included the specific list of “15 ways to express magnitudes” cited in the APA manual. The topic of the slideshow itself was a
good idea. My impression is that the material
is quite interesting and also fairly accessible, and your arguments seem
sound. You have attended to the specific
points I raised, but we agree to differ on the question of your extension of
Cohen's scale for correlations. I
appreciate the sentiment expressed (i.e., that there are times when Cohen’s
scale should be extended to include more categories and descriptive terms)
and the initiative shown by proceeding with the extension of the scale. However, as a correlation of 0.90 indicates
81% explained variance (or that 19% of the variance remains unexplained), is
it really appropriate to refer to the result as awesome? To me, we should reserve this word for more
impressive magnitudes. Back to article
|