Moderated-Mediation with SEM

[Thomas Trebilco]

I'm attempting to run the following moderated-mediation model in lavaan

Y is an ordinal outcome variable with two levels. Samples are either classified as 0's or 1's. X, M and W are all continuous/numeric. All variables are observed variables.

After having written and run my code, I get the following output with an error:

For search engines, the error is: "Error in lav_data_full(data = data, group = group, cluster = cluster, : lavaan ERROR: some variables have no values (only missings) or no variance"

Based on the output, the issue looks to be with the third factor, the interaction between W and M - for some reason I cannot understand it is not being indexed in the model. 

When I check for missing values and variance estimates on this variable, it shows no missing values and shows that it has substantial variance.

It doesn't matter whether I create this interaction term within lavaan (i.e. "term := W:M") or within R (i.e., "Interaction <- W*M), the same error comes up.

The code I wrote for this can be found here:

R Code Lavaan SEM Moderated-Mediation (github.com)

Please note that I have intentionally changed the variable names.

[Thomas Trebilco]

I seem to be also getting this same error in a simple moderation model.

[Terrence Jorgensen]

Without data, it will be difficult to track down the problem.  But in your GitHub syntax, you do not add McxWc to the data.frame  ModMed_Data.  Perhaps lavaan does not find the variable?

Terrence D. Jorgensen

Assistant Professor, Methods and Statistics

Research Institute for Child Development and Education, the University of Amsterdam

http://www.uva.nl/profile/t.d.jorgensen