path model with dichotomous outcome

[Sandy Schumann]

Hi everyone,

I would like to test a path model where one variable (choice) is dichotomous (0,1). I wrote the following code:

mediatedmodel12<-'##regressions

              choice~fight+age+gender_recoded+education

              fight~RATaccuracy + WCSTaccuracy

              ##covariance

              RATaccuracy ~~ WCSTaccuracy

              age~~education

              age~~gender_recoded

              education~~gender_recoded

              ##intercepts

              choice~1

              fight~1'

fitmediated12 <- sem(mediatedmodel12, data = fulldataset, ordered = c("choice"), estimator = "WLSMV")

summary(fitmediated12, fit.measures = TRUE, standardized = TRUE)

I got the following warnings:

3: In lav_model_vcov(lavmodel = lavmodel, lavsamplestats = lavsamplestats,  :

  lavaan WARNING:

    Could not compute standard errors! The information matrix could

    not be inverted. This may be a symptom that the model is not

    identified.

4: In lav_test_satorra_bentler(lavobject = NULL, lavsamplestats = lavsamplestats,  :

  lavaan WARNING: could not invert information matrix needed for robust test statistic

as well as 

In cbind(c1, c2, c3, deparse.level = 0) :

  number of rows of result is not a multiple of vector length (arg 3)

However, the output also states that lavaan ended normally after 72 iterations.

Can I use the robust estimates from the output? Or is it not safe to do so. (This is a replication of a study, and I follow the modelling of the original author, which is why I am not testing indirect effects.)

Thank you for your help,

Sandy

[Terrence Jorgensen]

  lavaan WARNING: could not invert information matrix needed for robust test statistic

Can I use the robust estimates from the output?

Not according to the warning message.  You could try another estimator, like ULSMV or PML.

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

[Sandy Schumann]

Thank you so much. I used the PML estimator and the warning is no longer reported. 

Best,

Sandy