How do you interpret the values between 0.01 and 0.06 for a sample size = 17? small or medium? in my sample size, there is no significant difference between the mean sores two groups. the values of eta squared are between 0.02 and 0.09. are these values statistically correct?
Hello Zeinab,
First, effect sizes are meant to convey information about the magnitude of a difference or relationship, that is independent of the sample size (whereas statistical significance depends on both).
I presume you're referring to Jacob Cohen's 1988 text, Statistical power analysis for the behavioral sciences (2nd ed.). He gives guidelines for interpreting the magnitude of effect that can be helpful if you don't have substantive knowledge or understanding of the variables, population, and context. But, they are only guidelines.
The answer to your question, based on Cohen's guidelines, depends on which (there were many from which to choose!) effect sizes you are reporting.
If Cohen's d: "small" = .20; "medium" = .50; "large" = .80
If correlation coefficient (r): small/medium/large = .1 / .3 / .5
If variance explained (like R-squared or eta-squared): s/m/l = .02 / .13 / .26
...and many more.
Good luck with your work!
David Morse Thanks for clarifying which of Cohen's books that was. I don't work in the behavioral sciences. Best, D. B.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size