Proc mixed G matrix (random intercept)

[mahmood]

Hello,

Thanks to everyone. I have a question on Proc mixed: I am using longitudinal/panel data. I have a random intercept model and the G matrix I am getting shows (only) row one for the first ID. Is this correct? How can I explain this?

 Estimated G Matrix

                            Row    Effect       NEW_PIN         Col1

                              1    Intercept       10001      0.1448

[Rick_SAS]

Please post the code that you are using. 

Is your output from a table or from a data set?

You say "first.ID", is this for a computation that includes BY group processing? Does the log provide any information, such as the model did not converge except for the first BY group?

[mahmood]

Hi, following is the code. x1, x2, x3 are time-variant; x1_mean x2_mean x3_mean are individual mean over time; z are time-invariant.

proc mixed data = reg_fin_1 noitprint PLOTS(MAXPOINTS= 20000000000) ;

class ID ;

by _imputation_ ;

model Y_i,t = x1_i,t x2_i,t x3_i,t x1_mean x2_mean x3_mean z1 z2 z3

/solution ddfm = bw residual ;

random intercept / subject = ID type = un solution g ;

ods output solutionF=mxparms3 ;

run;

[mahmood]

converged well; no warning or notice on the log

[mahmood]

NEW_PIN is the ID

[Rick_SAS]

I don't understand your syntax. What are the commas doing in the MODEL statement?

model Y_i,t = x1_i,t x2_i,t x3_i,t

[mahmood]

sorry, the actual syntax:
model depression =

financial_stress personal_rel_stress
other_stress VPA age
with_both_parents

fin_stress_M perso_stress_M other_stress_M VPA_M age_M

female mother_university ln_density_house
self_esteem parent_born_canada

/solution ddfm = bw residual ;

[mahmood]

on previous syntax: i is individual, and t = time

[mahmood]

Hello,

It seems that the G matrix only prints for the first observation. However, for the R matrix we can select which observations.

Thanks,

Rahid Mahmood