I am fitting a mediation model and get the following error.
WARNING: THE LATENT VARIABLE COVARIANCE MATRIX (PSI) IS NOT POSITIVE DEFINITE. THIS COULD INDICATE A NEGATIVE VARIANCE/RESIDUAL VARIANCE FOR A LATENT VARIABLE, A CORRELATION GREATER OR EQUAL TO ONE BETWEEN TWO LATENT VARIABLES, OR A LINEAR DEPENDENCY AMONG MORE THAN TWO LATENT VARIABLES. CHECK THE TECH4 OUTPUT FOR MORE INFORMATION. PROBLEM INVOLVING VARIABLE c1.
THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL. PROBLEM INVOLVING THE FOLLOWING PARAMETER: Parameter 99, c1 WITH c2
THE CONDITION NUMBER IS -0.645D-10.
Both c1 and c2 are independent variables (covariates). c1 is ordinal and c2 is binary. Both are declared categorical.
I checked the TECH4 output and found the following between c1 and c2:
ESTIMATED COVARIANCE MATRIX FOR THE LATENT VARIABLES: -0.401
ESTIMATED CORRELATION MATRIX FOR THE LATENT VARIABLES: -0.975
I examined the data and there is not any coding issues. Is there a way resolve the issue?
We need to see your output - please send to Support along with your license number.
Jenny posted on Monday, February 12, 2018 - 8:36 am
Hi Dr. Muthen,
I'm running a five-factor hierarchical CFA and received the not positive definite message involving one of the first order factors. I saw that it has a small negative residual variance (non-sig) and correlation with the second-order factor of greater than 1. I fixed the negative residual variance to 0 and the model ran fine but the correlation between this factor and the second-order factor is 1. What do you suggest in this instance?