R help - Jan 2011 - HLM Model

Hi

I am trying to convert SAS codes to R, but some of the result are quite
different from SAS. 

When I ran proc mixed, I have an option ddfm=bw followed by the model. How
can I show this method in R?(I am thinking that this maybe the reason that I
can't get the similar results)

below is my SAS codes:

proc mixed data=test covtest empirical;
class pair grade team school;
	model score = trt pair grade school/ solution covb ddfm=bw ;
	random int /  sub=team solution type=un;
run;

I have tried both lmer and hglm, but non of them works.

Could anyone tell me how can I covert this SAS codes to R? Thanks
-- 
View this message in context:
http://r.789695.n4.nabble.com/HLM-Model-tp3242999p3242999.html
Sent from the R help mailing list archive at Nabble.com.

I think it should be

fm <- lmer(score ~ trt + pair + grade + school + (1|team), test)

The unstructured covariance matrix you use in proc mixed is not available in Rs
function for mixed models
> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at
r-project.org] On
> Behalf Of Belle
> Sent: Thursday, January 27, 2011 2:44 PM
> To: r-help at r-project.org
> Subject: [R] HLM Model
> 
> 
> Hi
> 
> I am trying to convert SAS codes to R, but some of the result are quite
> different from SAS.
> 
> When I ran proc mixed, I have an option ddfm=bw followed by the model. How
> can I show this method in R?(I am thinking that this maybe the reason that
I
> can't get the similar results)
> 
> below is my SAS codes:
> 
> proc mixed data=test covtest empirical;
> class pair grade team school;
> 	model score = trt pair grade school/ solution covb ddfm=bw ;
> 	random int /  sub=team solution type=un;
> run;
> 
> I have tried both lmer and hglm, but non of them works.
> 
> Could anyone tell me how can I covert this SAS codes to R? Thanks
> --
> View this message in context: http://r.789695.n4.nabble.com/HLM-Model-
> tp3242999p3242999.html
> Sent from the R help mailing list archive at Nabble.com.
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

Hi Harold:

Yes, this was the R code that I tried, and got different result from SAS. 

Is that mean I cannot actually use R to run unstructured covariance matrix?
How can I solve this problem if I need an unstructured covariance matrix
method?

Thanks for the help.
-- 
View this message in context:
http://r.789695.n4.nabble.com/HLM-Model-tp3242999p3243158.html
Sent from the R help mailing list archive at Nabble.com.

Hi Belle,

try this:

SAS:
proc mixed data=test noclprint noinfo covtest noitprint 
method=reml;
class pair grade team school;
model score = trt pair grade school / solution ddfm=bw 
notest;
random int /  sub=team solution type=un r;
run;

R:
require(nlme)
unstruct <- gls(score~trt+pair+grade+school, test,
                correlation=corSymm(form = ~ 1 |id),
                weights=varIdent(form = ~ 1|team), 
method="REML")
summary(unstruct)

--------------------------------------
Silvano Cesar da Costa
Departamento de Estat?stica
Universidade Estadual de Londrina
Fone: 3371-4346
--------------------------------------
----- Original Message ----- 
From: "Belle" <ping.yi at gmail.com>
To: <r-help at r-project.org>
Sent: Thursday, January 27, 2011 5:43 PM
Subject: [R] HLM Model

>
> Hi
>
> I am trying to convert SAS codes to R, but some of the 
> result are quite
> different from SAS.
>
> When I ran proc mixed, I have an option ddfm=bw followed 
> by the model. How
> can I show this method in R?(I am thinking that this maybe 
> the reason that I
> can't get the similar results)
>
> below is my SAS codes:
>
> proc mixed data=test covtest empirical;
> class pair grade team school;
> model score = trt pair grade school/ solution covb ddfm=bw 
> ;
> random int /  sub=team solution type=un;
> run;
>
> I have tried both lmer and hglm, but non of them works.
>
> Could anyone tell me how can I covert this SAS codes to R? 
> Thanks
> -- 
> View this message in context: 
> http://r.789695.n4.nabble.com/HLM-Model-tp3242999p3242999.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide 
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, 
> reproducible code.
>

Hi Silvano:

Could you tell me what "correlation=corSymm(form = ~ 1 |id)"
represents?  In
our case, team is random effect, trt, pairs, grade, school are fixed effect,
and each team is within school. 

I still got the different results from both SAS and R. 
> unstruct <- gls(score~trt+pairs+grade+school,
> test,correlation=corSymm(form = ~ 1 |StudentID),
+          weights=varIdent(form=~1|team), method="REML")

I tried school instead of StudentID, but I got error.
> unstruct <- gls(score~trt+pairs+grade+school,
> test,correlation=corSymm(form = ~ 1 |school),
+          weights=varIdent(form=~1|team), method="REML")
Error in vector("double", length) : vector size specified is too large

Thanks for the help

-- 
View this message in context:
http://r.789695.n4.nabble.com/HLM-Model-tp3242999p3249518.html
Sent from the R help mailing list archive at Nabble.com.