[R-sig-ME] estimating AR1 parameters of level one error using lme

[Asher Strauss]

Hi all,
I am new to using R for mixed effects (have been using SPSS until now),
please for give me if this is a trivial question.

I am trying to understand how to estimate AR1 parameters of level one error
using lme (I have understood that specifying level one variance-covariance
matrix is not easily possible in lmer, is this true?).

When using SPSS one estimates two parameters: AR1 diagonal and AR1 rho. I
am searching for an equivalent command in R.

for example using the Glucose data from the nlme package:

data(Glucose)

fGlucose<-filter(Glucose,Meal=="10am")

summary(
lme(
fixed=conc~Time,
random=~1+Time|Subject,
method="REML",
data=fGlucose,
na.action="na.omit",
correlation=corAR1(form=~1+Time|Subject))
)


I get an out put of:

Correlation Structure: ARMA(1,0)
 Formula: ~1 + Time | Subject
 Parameter estimate(s):
     Phi1
0.4334469

is Phi1 equivalent to Rho?  I do not believe so, since when estimating AR1
diagonal and AR1 rho using SPSS I received 1.349 and -0.942 respectively.

here is the SPSS syntax I am using:
COMPUTE filter_$=(Meal="10am").
VARIABLE LABELS filter_$ 'Meal="10am" (FILTER)'.
VALUE LABELS filter_$ 0 'Not Selected' 1 'Selected'.
FORMATS filter_$ (f1.0).
FILTER BY filter_$.
EXECUTE.

MIXED conc WITH Time
 /CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1)
SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE)
PCONVERGE(0.000001, ABSOLUTE)
  /FIXED=INTERCEPT Time | SSTYPE(3)
  /METHOD=REML
  /PRINT=SOLUTION TESTCOV
  /RANDOM=INTERCEPT Time | SUBJECT(Subject)
  /REPEATED=Time | SUBJECT(Subject)COVTYPE(AR1).

Thank you!
Asher

	[[alternative HTML version deleted]]