01-09-2017 11:42 AM
I am analyzing some data about the hazard likelihood of children being investigated for child abuse to be placed in foster care by the different child protection agencies in a state. As we suppose the agencies differ in their rate of child placement, I try to use the PHREG procedure with the agencies as clusters using the random statement.
The following is one of the SAS syntax I use:
proc phreg data=here.inv_spell;
ods output FitStatistics=f2 ParameterEstimates=p2 solutionR=r2;
class genderf (ref=First) inv_age (ref=First) ethnic_inv (ref='WH') disp_new (ref='Substantiated')
perp_new (ref=First) iy (ref=FIRST) agency /PARAM=REFERENCE;
iy genderf inv_age ethnic_inv perp_new disp_new ;
random agency /solution;
The results indicate strong clustering effect of the agencies. I would like to further investigate whether the effect of child age or other covariates differs among the agencies. I wonder what the best way to specify that in the model. I tried to use “random invagecat /subject=dcs_agency solution”, but it does not work. I would be glad if you can give me some suggestions. I am using SAS 9.4 for Unix.