I am new in SEM and Mplus. Recently, I have estimated a simple path model with one measurement and a regression component. The measurement model measures a latent variable (L) from five observed variables. The regression compoonent shows the effect of two other observed variables (say X1 and X2) upon L. In my model, all L, X1 and X2 are considered endogenous, which means, X1 and X2 can affect L which can also affect X1 and X2. L is continuous while X1 and X2 are categorical.
I have used WLSMV estimator in Mplus to estimate the model. The estimation looks fine except one (probably) small issue which cant understand. Usually, if there is five categories, in general regression we expect 5-1=4 coefficients with respect to a base cagtegory. But here Mplus reports, only one coefficient. I dont understand how to interepret this coefficient.
Does Mplus consider the categorical variable as continuous? Or it is something to interpret with threshold values? Please respond.
Since it is not quite clear what you are referring to. you migth add the relevant part of the output file.
Thanks Jan for your interest, I have got an answer from Mplus forum. Mplus considers categorical variables as continuous unless we create n-1 dummies from the categorical variables. In Stata we can define categorical variables as i.X, so Stata automatically split those categories to give coefficients for X2......Xn. In Mplus, we have to manually create X2......Xn dummies.
You just have to define you variables as categorical (using CATEGORICAL ARE) in order for Mplus to treat them as such.
I did that before, but it did not give different coefficents for each category, then Prof Muthen advised me to create dummies.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling