Introduction to Effect Size

Overview of effect size, the concepts behind it, and how to calculate it.

[vincegredo]

Your M and mew are kinda hard to tell apart, might wanna use x bar instead. Also what if it's a t-test and there's no population standard deviation present?

[xojurxu]

Hi Andy, in you recent video, you use 3dTcat combine con.nii and spmT.nii as the result of effect size, it seems differ from the formula you show in this video, how to understand this? Thank you time and so great video!

[michaelfresco]

I'm confused by this video. When I run a small experiment (n=4), and I test say the average height of the group, and still don't know the *real* underlying standard deviation of the complete population (say, n=10, 000 of all the college students in the university). The formula Andrew uses to calculate effect size (observed difference)/(population standard deviation), is also different from most other videos, where they use a pooled standard deviation. Is Andrew suggesting, that we should derive the population standard deviation, (based on the sample standard dev), and then calc the effect size? (even if we cannot measure the population standard dev directly?)

[drryangagnon]

Good video.  However, I disagree that you shouldn't calculate/report effect sizes if the relationship is non-significant.  

[davoodbayat]

thanks for your perfect illustration! a short Q btw; this video is about the effect size for two samples after doing the experiment with known data; what if we have more groups (like in ANOVA) and would like to predict effect size for sample size calculation? I think we can consider two of those cells in ANOVA which are thought to be of the least difference, so we would calc Cohen's D for them to get the most conservative n per group. am I right?

[tabatamusic]

Hey, thanks for this great video!
How do I calculate the standard deviation if I have a baseline and intervention study? Since the standard deviation of the baseline might differ quite a bit from the standard deviation of the intervention. Thank you!