JASP 0.12.2 Tutorial: Mediation Analysis (Episode 17)

[@champ88888]

What does the R² in the Path plot say? I mean what does the 0.08 at the box "A_T" or the 0.25 at "BKL" or the 0.9 at "Age" say?

[@AlexanderSwan]

This is the value of measurement error in each measured variable in the analysis.

[@marianandrade38]

I don't know what happened but when I ask for the path plot the error values are really high (55, 19, 45) but everything else is fine in the analysis. My sample is near n=700. Please help! Why is this happening? Thanks.

[@AlexanderSwan]

Hi Ana, it's hard for me to help without actually looking at your data and output. It sounds like there is potentially an error somewhere in the code, so in order to figure that out, I would submit a bug ticket on JASP's support page.

[@marianandrade38]

@@AlexanderSwan thanks. Could you please advise me on what to check in the data, in case I am missing something? Thanks.

[@AlexanderSwan]

Not without looking at your data, unfortunately. You're welcome to email me everything and I'll quickly take a look, but it's not something I can do from minimal info in a YT comment.

[@abubakor6491]

Thank you very much for making this video...

[@lhisc6]

thank you for making this video! will you be making one on moderation analysis?

[@smiileish]

Thank you for the video! very good explanations. How would you report the findings in an APA table/text?

[@AlexanderSwan]

You report it much like a regular linear regression. Talk about the model fit to the data, so the result of each path, I directs and the direct. Couch these numbers in the theory and variables

[@wenzhenli6751]

Hello! I was having the same issue and was wondering if you were able to figure out how to report findings in APA text?

[@Bioexploration360]

Well-done sir much appreciated...

[@fairysaurus7557]

great video. how do you prepare the data beforehand? for example, how did you get the values for the mediator? Are they just values from a specific questionnaire

[@AlexanderSwan]

Yeah, it would depend on your design and which variable you set as a mediator. As far as setting up the data in your CSV for JASP, the mediator is just one of your variables, set up as a column like the others

[@fairysaurus7557]

@@AlexanderSwan thanks a lot :) does the data have to be standardized?

[@AlexanderSwan]

@@fairysaurus7557 No, unless you are are doing something with standardized scores

[@Honest_13]

Hello, thank you for this video, I'm French, I apologize in advance for my poor English! I would like to know something. I have several variables that I would like to compare because I have different groups. For example, I would like to know if the degree of autonomy influences psychological distress at work. Problem is that the variables have sub-scales...and I don't know how to do it! Because I have hypotheses but I don't know if I should average the variable for example "psychological distress" without taking into account the sub-scales such as "anxiety", "depression", etc. I averaged all the sub-variables but I'm not going to link all the sub-variables between them, that would be a lot !!! All my variables have 3 sub-scales so it makes things complicated, if I had only one dimensional scales it would be easier ! I hope I have been understood 😕

[@AlexanderSwan]

Hi--it sounds like you either want to run a regular simultaneous regression, with each subscale as a separate predictor, with psychological distress as your Y/outcome variable. It's hard to answer your question with the limited info and not being able to see the full methodology and the way you have your data set up. Apologies I can't be more help!

[@Honest_13]

Alexander Swan oh thanks anyway 😬

[@Anynymoooo12]

Hi Alexander, thank you for the explanation! I have a binary outcome (whether someone has committed fraud) and want to use two variables as predictors and a third as a mediator. However, in order to use the fraud variable as outcome, I must change it to a scale variable. Is that appropriate? Does the outcome make sense? One mediated relationship is significant and one is not.

[@AlexanderSwan]

Binary outcomes require logistic regression. Take a look down the playlist -- I have a recent JASP tutorial on that very analysis/test!

[@Anynymoooo12]

@@AlexanderSwan I have watched that video already, it was very useful! However, you don't talk about mediation (excuse my previous confusion, I wrote moderation instead of mediation).

[@AlexanderSwan]

@@Anynymoooo12 aha! Great. No worries, they go together. As far as I know, I don't think JASP can do a mediated logistic regression, unfortunately.

[@Anynymoooo12]

@@AlexanderSwan That is a shame. I guess I will have to look into R or Jasp perhaps. Do you know whether Lavaan in R allows binary outcome mediation analysis?

[@AlexanderSwan]

I am unsure about lavaan, but there might be another package out there

[@ERFL1988]

Thanks for your amazing videos! I have question, how could I calculate the effect size for this type analysis?

[@AlexanderSwan]

The effect sizes are the r values you get for each effect.

[@ERFL1988]

@@AlexanderSwan Thank you so much.

[@lala-bp3wh]

im trying to create a mediation analysis model where i use BMI as the predictor (BMI on scale), gender as a mediator (gender on nominal) and body image as the outcome (body image on scale) but for some reason i can't put gender as a mediator. it won't let me do it, can you help dear sir? i will wish you to always have toilet paper when you go to the bathroom, thank you.😄

[@AlexanderSwan]

Lol thanks for the last line…. Anyway, gender is a nominal variable and shouldn’t/can’t be used as a mediator. It can be a moderator, but not a mediator.