Effect size calculation in G*Power

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
mmvar
PLS Junior User
Posts: 1
Joined: Mon Sep 19, 2022 2:51 pm
Real name and title: Miguel Rosado, PhD Student

Effect size calculation in G*Power

Post by mmvar »

Hello.

Please bear with me as I am quite inexperienced in statistics.

I want to calculate the power for my experiment, these are the general parameters I used:

n (controls) = 5
n (experimental condition) = 6
average (controls) = 0.000026042
average (experimental condition) = 8.70E-06
pooled standard deviation = 1.5329E-05

I opted for the Mann-Whitney test since I am comparing means from two independent variables which do not follow a normal distribution. Because I have different sample sizes I used the option on top in the calculate effect size menu in G*Power. Thus I used the pooled standard deviation (blue outline) instead of each group's standard deviation.
Correct.png
Correct.png (54.59 KiB) Viewed 828 times
As you can see I obtained very low power. However, if I use the other option to calculate effect size (the one assuming the same sample size for both groups) I get a much better power value of about 0.97. I have seen guides online in which people use this option even when the sample size of each group is different.

My question: Is it valid to calculate the effect size (and consequently power) in this way, even if the sample sizes are different?

Thank you for your input.
Post Reply