Hi there,

I am interested in empirically validating a theory through Structural Equation Modeling. This theory suggests that the relationship between X and Y is mediated by a sequence of M1, M2, and M3. In my study, X is a multi-categorical variable that has three levels (e.g., X1 = close, X2 = distant, and X3 = none). My study will also have one covariate, CV, that is expected to be negatively related to Y (i.e., when CV increases, Y decreases).

The alternative to having X as a multi-categorical variable would have been to utilise a continuous moderator, W. In the alternative model, W would have moderated the relationship between X and M1, as well as X and Y.

With this in mind, here are my questions:

  • Does the attached diagram accurately depict my description above (involving a multi-categorical X)?
  • If so, is running a path analysis (e.g., through PROCESS or lavaan) appropriate for analysing this model? If not, what is/are the appropriate analysis/es to consider?
  • My study will recruit a mixed clinical sample. How do I determine the appropriate sample size for this model to have sufficient power? There is no certainty within the field in relation to effect sizes within the observed variables in this model.
  • Thank you for your time and input!

    Warm regards,

    George

    More George Saadé's questions See All
    Similar questions and discussions