Hi, I tried to run a mediation analysis in which the DV is made up of test scores after a training in the experimental group VS the control group, and in which the mediator is measured by a Likert-scale based questionnaire.
The aim is to understand whether the mediator has had a different impact on performances following the two types of training, but it seems I cannot run a mediation analysis with samples with different numerosity.
How can I bypass this problem?
Thanks in advance for any help with this!