Can someone calculate Cohen's d for me from these data?

F(4, 42) = 39.37, p = 0.001, eta squared = 0.48

There were 3 groups measured on 3 times (pre-treatment, post-treatment, follow-up; I don't think SD is pooled for any of the SDs). There were three groups, with the third group being the control group:

          Pre-treatment  Post-treatment       Follow-up
Group 1    37.10 (4.04)    16.10 (2.51)    17.88 (2.66)
Group 2    34.40 (5.50)    18.30 (4.62)    18.25 (4.83)
Group 3    34.30 (6.54)    25.55 (3.55)    28.14 (2.41)

You cannot calculate a $d$ effect size for more than two groups. For this, you would need to use Cohen's $f$. To calculate this, you will need to specify what type of variance proportion you want to use (e.g., $\eta^2$ or partial-$\eta^2$. Once you have made this decision, you can extract the necessary info from the ANOVA summary table. Then you can use the formula $$f^2 = \frac{\eta^2}{1-\eta^2}$$

  • $\begingroup$ I have The n^2 in which is 0.48. And I just need the effect size of how group 1 and 2 do compared to group 3. I don’t have a computer program to do the math. So I could use some help hehe. I think The variance proportion I need to use is n^2. $\endgroup$ Apr 4 '18 at 14:47
  • $\begingroup$ Or maybe I need the effect size of group 1 versus group 3 and then the effect size og group 2 versus 3. Then I could compare the two effect sizes, right? $\endgroup$ Apr 4 '18 at 14:49
  • $\begingroup$ If you are just comparing two of the groups, use the MS-error from the final ANOVA summary table, the square root of this can be used as the pooled-standard deviation. Then, divide the difference between the two groups of focus by this value, and you can report these pair-wise comparison $d$ values. $\endgroup$
    – Gregg H
    Apr 4 '18 at 15:48
  • $\begingroup$ Is it possible to compare all there groups? And how? $\endgroup$ Apr 6 '18 at 7:53
  • $\begingroup$ Yes, the Cohen's $f$ compares all three groups at once. $\endgroup$
    – Gregg H
    Apr 6 '18 at 12:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.