Interrupted time series using proc mised

[sophiec]

Hello, 

I am trying to run an interrupted time series analysis with a control group and stratified by race/ethnicity. The intervention (a health policy change) occurred at the state level, so I have 2 states (one that did the intervention and a control state). The outcome is mortality, which was derived from CDC data. I created four separate groups (based on race/ethnicity and state) and have repeated measures over a 10 year period. 

I used code from the 2020 SAS Global Forum Paper 4674-2020

proc mixed data=work.state method=ml 
	plots(maxpoints=60000)=(residualpanel(unpack) vcirypanel(unpack)); 
	class group exposed intervention; 
	model mortality= year intervention intervention*year exposed exposed*year exposed*intervention exposed*intervention*year/s;
	repeated / subject = class type=un R;
run; 

When I try to run the code, I get this error message. Any suggestions? 

Thanks, 

NOTE: PROCEDURE MIXED used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

ERROR: Variable class not found.
NOTE: The SAS System stopped processing this step because of errors.

[sophiec]

I found my error "subject = class" should have been "subject = group".

Unfortunately I receive the warning: unable to make Hessian positive definite. That seems to be a separate problem

Apologies that I can't figure out how to delete the original post.

[sbxkoenk]

---

@sophiec wrote:
Unfortunately I receive the warning: unable to make Hessian positive definite. That seems to be a separate problem.

---

That's a separate problem indeed.

See 

Home > Analytics > Statistical Procedures >
Proc Mixed with warning "unable to make hessian positive definite"
https://communities.sas.com/t5/Statistical-Procedures/Proc-Mixed-with-warning-quot-unable-to-make-he...

In that post @PaigeMiller refers to
https://www.theanalysisfactor.com/wacky-hessian-matrix/

Koen

[jiltao]

what happens if you change type=un to type=cs?

Thanks,

Jill