Calculating statistical power using G*Power (a priori & post hoc)

This video explains how to calculate a priori and post hoc power calculations for correlations and t-tests using G*Power.
G*Power download: www.gpower.hhu....
Howell reference: Howell, D. C. (2012). Statistical methods for psychology. Cengage Learning.

[@virmacreations]

Boom!!!! All I was needing!! Gracias!!!!!

[@tevfiksabuncu4623]

nice explanation. thanks

[@danserban8648]

Thank you! It helped me quite a bit:)

[@nelhuens7590]

How do you calculate the correlation betweens groups?

[@peteradu5103]

can u help with ANCOVA calculation?

[@papschulo]

From F-test, What is the differences between r2 deviation from zero and r2 increase?

[@zorawessely3638]

Thank you for this video! Related to post-hoc power calculations for logistic regression in G*Power I don't understand how to determine Prb(Y=1 | X=1) H0. I found these two formulas: Odds1SD=p/(1−p)∗OR(Sx) P(Y=1|X=Sx)=Odds1SD/(1+Odds1SD) So I was wondering: 1. Can I use these 2 to determine Prb(Y=1 | X=1) H0? 2. If so, is OR(Sx) the odds ratio reported in the study? Do you happen know the answer to them? Best regards, Zora

[@ainaazeezbabatunde748]

Hello, i would love to have an answer to this question as well. Can you reply back.Thanks

[@peteradu5103]

Please I am doing a network analysis for my phd thesis how can the G power help me calculate my sample size ?

[@PsychEDD]

A network analysis like a social network / authorship network? I think the network size will exist as it is - perhaps any effects you're looking at (such as measures of centrality) will have some type of effect size and you can do a post-hoc power analysis instead of a priori.

[@peteradu5103]

@@PsychEDD Thanks for the response. Please a further explanation will help or material to help me. What I know is that post hoc is done after I have collected my data but i have not collected the data yet.

[@l.s.7682]

Thank you for the helpful video! I wonder if the number was right in the coefficient of determination p2 box at 12:22. You said the correlation is 0.5. So I guess what should have been put in was 0.25 rather than 0.5?

[@kappios78]

that's what i though, yep

[@samiakhalid6966]

very helpful for me, thank you

[@angelica3744]

Thanks for your help!

[@muratgul6449]

For the Paired t test , I will have a question as follows : Effect size 0.5, Alpha=0.05, 2-Tails, power=0.80, while the Total Sample Size was calculated as 34. But since there are two groups, should I take these 34 people as 17 people in each group, or should I take 34 people in each group ? I can't decide this. Thank you so much for your help and your response. Kindest regards

[@PsychEDD]

Two groups of n=34 at both time points? Or n= 17 ppts taking the measures twice?

[@pastelqueenwudel]

@@PsychEDD I think he means if the result of g*power is n = 34 for the paired test - that means we should take 34 people in the group we have to measure on two different times. But what if we have a control group here? Should the control group also have 34 people in it or is it 17:17 on each group?

[@yang6745]

@@pastelqueenwudel Control group also have 34 people in it!!!!

[@rustyjeanz]

Can we input the data series from excel?

[@DonsHerr]

*screams* are you South African??!

[@PsychEDD]

Yep!

[@DonsHerr]

@@PsychEDD It's so dope to hear a familiar accent on KZbin. Hope the lockdown is treating you well! Thanks for the help this video provided - from Rondebosch, Cape Town :)

[@tanikapatricemckissick260]

@@PsychEDD How do you report this?

[@Wiwwelien]

thanks man!

[@gene12586]

Hi, what is the song in the intro? Also, thanks for the video, very helpful

[@phoenix9999]

It is the intro to Angelica by Lamb, it is on the Between Darkness & Wonder Album (released 2003).

[@sevinchkhonnajmitdinova1555]

Debussy - Clair de Lune

[@Alex50702]

Very helpful video, thanks! Keep it up