JASP/Excel - Mediation Analyses Example
HTML-код
- Опубликовано: 22 авг 2024
- Lecturer: Dr. Erin M. Buchanan
Missouri State University
Spring 2017
This video covers how to:
1) Power analyses for mediation in G*Power
2) Run data screening in Excel and JASP for regression
3) Baron and Kenny steps for mediation
4) APA style reports
Lecture materials and assignment available at statisticsofdoom.com.
statisticsofdo...
I just discovered your videos. They're great! Thanks for sharing them.
Thanks! Glad to be of help. :)
You are just awesome. Thanks for your great tutorials
Thanks for this video! Should I use Z scores for regression and mediation analysis? And also, can I stop the mediation analysis if a path and b path are not significant? AND if the c and c' are significant (but sifferent p value) can I do Sobel test to see if the two sig. P values are different each other?
Not sure on the z-scores - if standardization makes sense sure, but normally it's in the scale of the data. I would probably make a case if a and b aren't sig, then mediation might be difficult to argue for. However, you could use still use the Sobel test to determine if the indirect test is greater than zero yes.
Hi Dr Buchanan thank you for this very helpful video. I have two groups coded 0 and 1 and I completed the Baron and Kenny steps by linear regression, similar to your video. After looking at the data the results make sense however I checked them by mediation on SEM on JASP and on the process macro on SPSS followed by bootstrapping. The effect size for all paths apart from the b pathway has doubled exactly. My mediation effect size also doubled. The results from the Baron and Kenny steps by linear regression look more sensible but is there a way to explain the difference? should k number of groups be factored in somewhere? Thanks so much. Your videos are awesome. : )
You should get the same results from each one of those things, so I would carefully check they were all performed the same way.
Hey Dr. Buchanan! Awesome video. I was wondering if you had an analysis suggestion for a case of data that is showing the c' path to have a greater magnitude than the c path, however moderation analyses are showing nonsignificant p values. The p values for the mediation analysis had a range of >.001-.009 dependent on the path/coefficient in question.
Hmm that implies that the M variable is controlling for or suppressing the error in Y for the X variable to be "more predictive". Maybe check out the correlations with X, Y, and M?
Thanks for the extensive explanation. I need some advice: In my study one of the variable is non-normal. So my question is whether we can do mediation analysis on non-normal data?
The assumption isn't if the data is normal but if the residuals are normal - you can just be careful with interpretation.
@@StatisticsofDOOM Thanks a lot for the prompt reply.
Respected mam can you please explain how to represent graphically the mediation effect through spss or excell
Not sure what you mean...something different than the triangle picture?
This is great thank you so much! I need some advice: What happens if the the c-path is significant, but a-path is not significant - however, the c'-path is significant when the mediator is included? Also, the Sobel test shows a significant difference between c and c'. Any help/advice/insights would be hugely appreciated
I think the answer here depends on your theory/hypotheses. Generally, people want the a path to be important (meaning the effect is there, however you define that, whether it be significance or a specific effect size or coefficient)....otherwise it sounds like maybe b is doing all the work, which wouldn't imply a redirection of the effect through the mediator.
@@StatisticsofDOOM Thank you, that makes sense. I appreciate the insights.