Model without SNP effects is not identified (df=18) while being identified with SNP effects

[Thanh Le]

Hello everyone,

Using 8 traits, I arrived at a hierarchical model with a good fit of CFI being 0.9879857 and SRMR being 0.04539556. While the degree of freedom is 18, I received that this warning for the model without SNP effects "The unstandardized model produced the following warning: lavaan->lav_model_vcov():

Could not compute standard errors! The information matrix could not be inverted. This may be a symptom that the model is not identified."

There are no warnings with multivariate GWAS. I thought that this is because the SNP effects provide more information that allows each model to be identified. However, given that the degree of freedom in the model without SNP is 18, I don't know why this model is not identified and if there is a reason for concerning.

Thank you so much in advance!

Best regards,

Thanh Le