Hi Dave,
When I run your code, I get the following error:
repeated t() values within lowest-level panels
r(459);
Thus, xtmixed doesn't know that the repeated measures are within id.
Thus, you need to change your xtmixed code to:
xi: xtmixed y group i.time i.group*i.time || id: , nocons
residuals(unstructured, t(time))
The || id: tells xtmixed that there are repeated observations within
id and nocons makes it so you don't estimate this random intercept.
This should be equivalent to the REPEATED statement in SAS. I don't
have the book you reference, so I can't check whether the results
match up or not.
Best,
Scott
On Fri, Jan 29, 2010 at 10:55 AM, Amado David Quezada Sanchez
<[email protected]> wrote:
> Hello,
>
> I have tried to replicate an example from Fitzmaurice/Laird/Ware book "Applied Longitudinal Analysis". The code presented by the authors was written for SAS:
>
> PROC MIXED;
> CLASS id group time;
> MODEL y=group time group*time/S CHISQ;
> REPEATED time/TYPE=UN SUBJECT=id R RCORR;
>
> According to the authors "The MODEL statement specifies the response varriable and the fixed effects" and "The SUBJECT option ... is used to denote a variable that distinguishes clusters of correlated responses"..."TYPE=UN specifies an unstructured covariance matrix"
>
> So, after setting the data into long format, I tried to run the following code:
>
> /* Setting the data */
> clear
> infile id str1 grp y0 y1 y4 y6 using http://users.uoa.gr/~fsiannis/data/longitudinal/tlcdata.txt
> reshape long y, i(id) j(time)
> gen group=0 if grp=="P"
> replace group=1 if grp=="A"
> /* Fitting the model, option residuals only found in STATA11 */
> xi: xtmixed y group i.time i.group*i.time, residuals(unstructured, t(time))
>
>
> I receive an error message. What could I do to replicate the SAS code?
>
> Thank you!
> Dave
>
>
>
>
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
