A final comment: My focus here has been on Cohen’s qualitative labels of small, medium, and
large. His focus in the books cited was on power analysis; small, medium, and large
conventions “when no others suggest themselves” was something of an aside.
Now, 50 years later, it is worth reflecting on the dramatic changes Cohen has caused in social
science. He came of age at a time when analysis of variance and multiple regression were
separate worlds, hardly on speaking terms. His 1968 Psychological Bulletin article changed that
(and—unseen by users—computer programmers quickly realized that, with a little matrix
magic, common routines could undergird the computations required by various statistical
procedures; think general linear model).
A few years earlier (1962), Cohen had rattled the social sciences with his revelation that articles
claiming no effects (i.e., no statistically significant ones) were too often under powered—and so
a power analysis became de rigueur for all grant and dissertation proposals. But now that the p
< .05 “cliff” (you can only talk about effects below the cliff, as journal editors and reviewers are
prone to admonish), has been partially demolished (thanks in part to Cohen, 1990, among
others), we can realize that power analyses only matter when statistical significance is the only
criterion by which results are judged. This makes only more prescient Cohen’s emphasis on the
size of effects.
References
Bakeman, R. (2005). Recommended effect size statistics for repeated measures designs. Behavior
Research Methods, 37(3), 379–384. https://doi.org/10.3758/bf03192707
Cohen, J. (1962). The statistical power of abnormal-social psychological research: A review. Journal of
Abnormal and Social Psychology, 65, 145–153.
Cohen, J. (1968). Multiple regression as a general data-analytic system. Psychological Bulletin, 70, 426–
443.
Cohen, J. (1969). Statistical power analysis for the behavioral sciences. New York: Academic Press.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences (Revised ed.). New York:
Academic Press.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
Cohen, J. (1990). Things I have learned (so far). American Psychologist, 45, 1304–1312.
Kirk, R. E. (1996). Practical significance: A concept whose time has come. Educational and Psychological
Measurement, 56(5), 746–759.
Thompson, B. (2007). Effect sizes, confidence intervals, and confidence intervals for effect sizes.
Psychology in the Schools, 44(5), 423–432. https://doi:10.1002/pits.20234
A personal note: in the mid-1960s, when I was working for the Yale Computer Center, some
people in the Psychology Department, showing me formulas in Weiner (1962; Statistical
Principles in Experimental Design), asked me to write a program for repeated-measures designs,
which were not yet included in the fledging statistical packages of the time. The program
(written in FORTAN) was brought to Georgia State University from Yale by a Yale post-doc GSU
had hired and installed on the GSU computer system. When I asked someone where the
program had come from, I got a vague, oh-we-get-them-from-here-and-there response—
unaware they were talking to the author.