topic Re: Model with Clustering, Absorption, Year/individual fixed effects, and linear combination testing in SAS Procedures

Model with Clustering, Absorption, Year/individual fixed effects, and linear combination testing

oshih14 — Thu, 14 Nov 2024 19:00:53 GMT

Hello, I am looking for a procedure which achieves all of the listed: error clustering, year and individual fixed effects with individual fixed effects absorbed (since I have a lot of individuals), and the ability to test linear combinations of variables post-estimation. In Stata, the code would look something like

areg y x1 x2 x3 i.year, absorb(individual) cluster(individual)

lincom x1 x2

However, the only procedure I am able to find with absorption is glm, which does not support clustering. Alternatives I've considered are genmod, mixed, panel, and surveyreg, but they do not support absorption. Are there any procedures which suit my needs?

Re: Model with Clustering, Absorption, Year/individual fixed effects, and linear combination testing

ChrisNZ — Mon, 18 Nov 2024 21:45:39 GMT

@Rick_SAS Can you help?

Re: Model with Clustering, Absorption, Year/individual fixed effects, and linear combination testing

Ksharp — Tue, 19 Nov 2024 00:39:05 GMT

You could try PROC HPMIXED which is for high performance and big table.

Re: Model with Clustering, Absorption, Year/individual fixed effects, and linear combination testing

Rick_SAS — Tue, 19 Nov 2024 10:52:32 GMT

I do not know the answer. Perhaps @StatDave or @StatsMan might know.

Re: Model with Clustering, Absorption, Year/individual fixed effects, and linear combination testing

StatDave — Sat, 07 Dec 2024 17:20:46 GMT

The GEE method available in PROC GEE (and GENMOD, but GEE is the recommended procedure) via the REPEATED statement is effectively absorption and clustering. It allows estimation of the effects specified in the MODEL statement adjusted for the repeated (clustered) measurements within subjects and avoids the need to estimate parameters for the individual subjects. Using the example from the Details:Absorption section of the GLM documentation, the following statements produce essentially the same results:

proc glm;
absorb herd cow;
class treatment;
model y = treatment/solution;
run;
proc gee;
class herd cow treatment;
model y=treatment/type3 wald;
repeated subject=cow(herd);
run;