Dear all

I am trying to fit a cross lagged panel model in lavaan, but it return the following error message:

Fejl i modindices(result1) :

lavaan ERROR: could not compute modification indices; information matrix is singular

In addition: Advarselsbeskeder:

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

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

variables involved are: A1 G1

2: I sqrt(var.lhs.value * var.rhs.value) : NaNs produced

3: I lav_start_check_cov(lavpartable = lavpartable, start = START) :

lavaan WARNING: starting values imply NaN for a correlation value;

variables involved are: A2 G2

4: I lav_start_check_cov(lavpartable = lavpartable, start = START) :

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

variables involved are: A3 G3

5: I sqrt(var.lhs.value * var.rhs.value) : NaNs produced

6: I lav_start_check_cov(lavpartable = lavpartable, start = START) :

lavaan WARNING: starting values imply NaN for a correlation value;

variables involved are: AA GG

Does anyone know what went wrong?

'''

#### RI-CLPM, for simplicity only considering the first three sessions:

## define latent variables (A1,A2,A3,G1,G2,G3) from observed ones (a1,a2,a3,g1,g2,g3)

model1 <- '

A1 =~ 1*anx1
A2 =~ 1*anx2

A3 =~ 1

*anx3*

G1 =~ 1gq1

G1 =~ 1

G2 =~ 1

*gq2*

G3 =~ 1gq3

G3 =~ 1

## define stable individual trait variables

AA =~ 1*anx1 + 1*anx2 + 1*anx3
GG =~ 1*gq1 + 1

*gq2 + 1*gq3

## set up clpm for latent variables

G3 + A3 ~ A2 + G2

G2 + A2 ~ A1 + G1

## include residual correlations in clpm

A1 ~~ G1

A2 ~~ G2

A3 ~~ G3

## allow correlation btw individual trait variables

AA ~~ GG

## disallow correlations btw latent variables and trait-variables

A1 + A2 + A3 + G1 + G2 + G3 ~~ 0*AA + 0*GG'

'''

Best wishes Morten