Skip to content
Prev 5869 / 20628 Next

level 1 variance-covariance structure

Thierry,
I can run this model... but what does it mean?
The correlation structure that I get is:

Correlation Structure: ARMA(1,0)
  Formula: ~age13 | id
  Parameter estimate(s):
Phi1
    0

What does zero mean? I would expect get some positive number there...

  m3a <- lme(attit ~ 1 +  age13 , data, random= ~ 0 + factor(age13)| id, 
correlation = corAR1(form =  ~ age13 | id))
Linear mixed-effects model fit by REML
  Data: data
         AIC       BIC   logLik
   -324.2096 -229.5528 181.1048

Random effects:
  Formula: ~0 + factor(age13) | id
  Structure: General positive-definite, Log-Cholesky parametrization
                 StdDev     Corr
factor(age13)-2 0.17219431 f(13)-2 f(13)-1 f(13)0 f(13)1
factor(age13)-1 0.19789254 0.493
factor(age13)0  0.25942941 0.425   0.544
factor(age13)1  0.28354459 0.442   0.442   0.723
factor(age13)2  0.29097081 0.498   0.474   0.639  0.808
Residual        0.07457025

Correlation Structure: ARMA(1,0)
  Formula: ~age13 | id
  Parameter estimate(s):
Phi1
    0
Fixed effects: attit ~ 1 + age13
                 Value   Std.Error  DF  t-value p-value
(Intercept) 0.3210558 0.012832840 839 25.01829       0
age13       0.0593529 0.004716984 839 12.58282       0
  Correlation:
       (Intr)
age13 0.504

Standardized Within-Group Residuals:
         Min          Q1         Med          Q3         Max
-1.46371874 -0.27170442 -0.04080688  0.26239551  1.69883907

Number of Observations: 1079
Number of Groups: 239
On 4/12/2011 10:21 AM, ONKELINX, Thierry wrote: