Hello Statalist,


I have some questions regarding to my mediation analysis as I'm very new to Stata and not sure if I got this right.
My DV is an ordinal categorical variable (7-point Likert scale). However, my department is not very strict on treating this as a continuous variable so I could run OLS. My IV is a categorical variable with 3 conditions (2 treatments and 1 control; coded as 0,1,2). My mediator is a continuous variable (index scores that I compute from 12 items questionnaire).

I have followed this link; https://stats.idre.ucla.edu/stata/fa...e-sem-command/
And run the mediation analysis shown below;
*where pur_int = DV, condition = IV and Mediator_idx = MV


sem (mediator_idx <- condition) (pur_int <- mediator_idx condition)
(79 observations with missing values excluded)

Endogenous variables

Observed: mediator_idx pur_int

Exogenous variables

Observed: condition

Fitting target model:

Iteration 0: log likelihood = -2901.0187
Iteration 1: log likelihood = -2901.0187
Array

I am aware that my IV has 3 conditions. Therefore, I tried to put i.prefix on it. However, it doesn't work and I got this response from Stata;

sem (mediator_idx <- i.condition) (pur_int <- mediator_idx i.condition)
factor variables not allowed
r(101);

Therefore, I would like to ask Statalist on what command should I use when my IV has 3 conditions and how to interpret them.
Thank you very much in advance.