Error with mediation analysis in lavaan in R

Question

I am beginner to run mediation analysis using lavaan package in r and have some issues to run the models and understand errors.

Here is my syntax.

# Ordinal NP with covariates     

model1 <- ' # outcome
            CogTest ~ c*EV +b*NP + Sex + AgeatTest + TesttoDeath + Race + Edu*Occu
            
            # mediator
            NP ~ a*EV + Sex + AgeatDeath + Gen
              
            # indirect effect (a*b)
            ab := a*b
            
            # dirdct effect
            direct := c
              
            # total effect
            total := c + (a*b)
         '

NP : ordinal, mediator EV : continuous, predictor CogTest : continuous, outcome Sex, AgeatTest, TesttoDeath, Race, Edu*Occu (interaction of Edu and Occu) : covariates for the outcome path Sex, AgeatDeath, and Gen : covariates for the mediator path.

fit.model1.delta <- sem(model1, data = df, ordered = c("NP"))
summary(fit.model1.delta)

When I ran the model, I got an error message.

Error in if (!all(x == cache$theta)) { :
missing value where TRUE/FALSE needed

But, if one of the two covariates, AgeatTest and TesttoDeath, was excluded from the outcome path, for example

# outcome
CogTest ~ c*EV +b*NP + Sex + TesttoDeath + Race + Edu*Occu

or 

CogTest ~ c*EV +b*NP + Sex + AgeatTest + Race + Edu*Occu

The model was fitted and results were released. Does anyone know what was wrong with my analysis?

Thanks in advance!

Answer 1

You can't put an interaction term in the model in lavaan. Create the interaction term first.