I tried running a double mediation model but it did not seem to converge, outputting the following error:
THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES MAY NOT BE TRUSTWORTHY FOR SOME PARAMETERS DUE TO A NON-POSITIVE DEFINITE FIRST-ORDER DERIVATIVE PRODUCT MATRIX. THIS MAY BE DUE TO THE STARTING VALUES BUT MAY ALSO BE AN INDICATION OF MODEL NONIDENTIFICATION. THE CONDITION NUMBER IS 0.229D-11. PROBLEM INVOLVING PARAMETER 14. MODIFICATION INDICES COULD NOT BE COMPUTED.THE MODEL MAY NOT BE IDENTIFIED.
I located the problematic variable, and I do not get the error when I remove it from the model.
The scales across the variables are similar. The variance on the problematic variable was high, so to reduce the variance, I tried including the problematic variable but using the mean score for that variable rather than the sum. Even though the variance for the mean scores was low and not too discrepant from that of the other variables, the model still failed to converge.
I also tried increased the the number of iterations to 10000, and used start values but the error persists.
What do you think is the source of the problem and how would you advise to proceed?