What type of standard errors are produced in proc mixed?

[KMun]

I have been reading SAS manuals and am getting conflicting results. I am reading a lot of information about how to get different types of standard errors with additional lines of code. I am not sure what the default being produced here. Based on a very old project, I remember these being robust standard errors or errors clustered on country. Can anyone help me with identifying these (and should I be adding code to report something else)?

proc mixed method = ml covtest ic;
class country;
model q5= age white gender income q5 country_q5
/solution ddfm=bw notest; weight dweight;
random int income/subject = country G TYPE = VC;
run;

As an aside, there are a small number of countries in the dataset. This will effect the significance of group-level variables, right?

[ChrisNZ]

Moved to stats community where you are more likely to get a reply.

[SteveDenham]

While you can get various sorts of standard errors in PROC MIXED, this code would not enable that.  It will produce standard errors considering country, country by income and residual sources of error.  See the documentation for how these arise (but be prepared to consider matrix equations)  There may be some "G matrix not positive definite" warnings in the output that could indicate you have overfit the random effects. In the usual sense of the word, none of the mixed model procedures produce robust standard errors.

SteveDenham

[KMun]

Thank you. Would these accurate be described as standard errors clustered on country?

[SteveDenham]

Yes.  That would be an accurate statement.

SteveDenham