Please help: I need to calculate the g*power for my study. I have two predictor variables and one criterion variable. All measured as continuous scale.
Three research questions: Rq1-Does mindset predict GPA? Rq2-Does SES predict GPA? Rq3-Does mindset predict Gpa while moderating for SES?
I’m thinking that rq1&rq2 needs a simple linear regression/bivariate regression.
Im thinking rq3 needs a moderated multiple regression.
Will I run gpower once or twice? (For Two seperate test?)
What type of gpower is best for linear regression with two predictor variables and one dependent variable?
Hello Adrian,
I agree with your intuition that RQs #1 and #2 could be addressed either by ordinary Pearson correlation, or by simple (one predictor) linear regression. Whichever RQ calls for the smaller target effect size (say, r-squared) should be the one that guides your selection of sample size for these two RQs.
For RQ #3, there are actually two considerations: (a) is the overall R-squared noteworthy? and (b) does the mindset by SES interaction significantly add to the explanatory power of a base model that has only the IVs of mindset and SES (with GPA as the DV)? I suspect that (b) is the key aspect for you in order to be able to assert moderation effects. So, in this instance, use g*power for linear multiple regression: fixed model, Rsquared increase (adding one more IV, the interaction, to a model which has 2 IVs in it to begin with). Again, the question is, what is the smallest R-squared increase that is noteworthy to you (that's the effect size, at least before converting it to a Cohen's f-squared, which the program will do for you).
Good luck with your work.
Adrian Hendricks have a look: https://www.youtube.com/watch?v=veaEDIM13Ks&t=15s
Hello!
I just defended my dissertation and one of my RQ's had included GPA. Are you using a letter grade or the % amount of the grade. Get clarification as I had to go back to IRB to get approval, because the results were clustered around the letter A and could not get the results I wanted, so I had to use the % of the grades.
Hello,
I think the way to help you with this would be to go https://statistics.laerd.com/. Its not bad. My Chair recommended it when I had a similar question without giving me the answer lol. I was grateful because the site was able to help and explain which statistical test I should do. Once you figure out which statistical test to run for your study, the G*power analysis will be more accurate. I believe GCU has a g*power software (outside of SPSS) you can download and use to run the analysis. You should only need to run the g*power analysis once to get your sample size requirements. Hope that helps.
Hello Adrian!
In my opinion, you could try to use RStudio or R program to conduct the power analysis. You can use pwr package. It is also pretty straightforward. For example the code looks like this: pwr.t.test(d = .24, sig.level = .05, power = .80, alternative = 'two.sided'). This means the effect size is medium and because my previous research is between subject I used t test.
You just need to figure out which statistical test you want to run and the effect size.
Good luck!
Best,
Edita
I just did this for my dissertation work at GCU, I would use a simple Pearson correlation for what you're describing .
Given that you have multiple RQ's you have to consider the bonferroni effect as well. so if your alpha is 0.05 and you have 3 RQs your alpha would be 0.0167.
Power should be 0.8. Correlation pH1 should be 0.3.
I took almost an entire night at residency watching YouTube videos on this topic to ensure I understood it. make sure that you understand all the different terms (alpha, power, Cohens effect size, correlation, bonferoni, etc....) there's some good resources out there. https://www.youtube.com/watch?v=JCzh5jqMKO0
Best of luck
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size