Hi,

Please I need help to understand problems on CFA using lavaan package.

When I run the following code:

#Specify the CFA model

ModelCFA <- ' Factor 1=~ COMM1 + COMM2 + COMM3

Factor 2=~ CLOS1 + CLOS2 + CLOS3 + CLOS4

Factor 3=~ COMP1 + COMP2 + COMP3 + COMP4 '

#Get model fit

fit <- cfa(ModelCFA, data=DavAuraMarR)

Then the Warning message indicates:

Warning message:

In lav_object_post_check(object) :

lavaan WARNING: covariance matrix of latent variables

is not positive definite;

use lavInspect(fit, "cov.lv") to investigate.

When I check covariances and correlations using lavInspect(fit, "cor.lv") and lavInspect(fit, "cor.lv"), I obtain:

lavInspect(fit, "cov.lv")

Fact 1 Fact 2 Fact 3

Factor 1 1.363

Factor 2 1.409 1.677

Factor 3 1.240 1.434 1.117

lavInspect(fit, "cor.lv")

Fact 1 Fact 2 Fact 3

Factor 1 1.000

Factor 2 0.932 1.000

Factor 3 1.005 1.048 1.000

How is it possible to see correlations greater than 1?

What can I do?

The model fits good but I have this warning message, and overall I can't see the modification indices.

When I try t see it, the second warning message tells me:

Warning messages:

1: In lav_start_check_cov(lavpartable = lavpartable, start = START) :

lavaan WARNING: starting values imply a correlation larger than 1;

variables involved are: Factor 1 Factor 3

2: In lav_start_check_cov(lavpartable = lavpartable, start = START) :

lavaan WARNING: starting values imply a correlation larger than 1;

variables involved are: Factor 2 Factor 3

I have a great population (over 1000)...

Thank you for your answers