I am running a small mediation model but would like to account for the clustering associated with site, which is a categorical variable comprised of 5 sites. When running the model I get the following error using the TYPE=COMPLEX procedure.
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.
THIS IS MOST LIKELY DUE TO HAVING MORE PARAMETERS THAN THE NUMBEROF CLUSTERS MINUS THE NUMBER OF STRATA WITH MORE THAN ONE CLUSTER.
Should I be concerned about this and are there any workarounds for this.
Five clusters is not enough for TYPE=COMPLEX. A minimum of 20 and from 30-50 is recommended. You can create four dummy variables to represent the five sites and use those as covariates in your model to account for non-independence of observations.