Mediation analysis
Hello everyone
For my thesis I have to run a mediation analysis. The set-up of the data is like this:
1) Independent variable: nominal variable --> group 1 & group 2
2) mediators: Continuous --> shame, guilt and pride
3) outcome variable: ordinal --> impact
Can anyone tell me how I can run this analysis?
I tried with SEM, but I keep getting this error notification:
Already a big thanks!!
Kind regards
Marie-Laure
Comments
Hi Marie-Laure,
It seems the error message says that the IT (impact?) variable needs to be either binary or continuous, but it is ordinal? Changing the scale to continuous may make the analysis run (not a perfect solution of course!). I am wondering why the analysis requires this -- did you use lavaan code?
Cheers,
E.J.
Hey EJ
Thanks for your reply! I tried a dummy coding and now the problem seems solved.
However, my dependent variable is a ordinal one. People had to say if the independent one would influence their choice on a scale from "totaly not" to "very much" --> 5 levels.
Putting in this variable, I get this error:
Just changing it to a continuous one would not be sufficient I guess.
Do you have an idea? :-)
Kind regards!
Hey MarieLaure,
you have some missing values, and the default full information maximum likelihood (FIML) handling of those missing values is only available with continuous endogenous variables. You can set the missing value handling to listwise deletion if you think this is appropriate for your problem.
Erik-Jan