SPSS and R discrepancy

#1
Hi,
I'm trying to figure out why SPSS and lmer in R are giving me different outputs.

Continuous variables (in both models):
Hippo, Age, BMI, ICV, PTEDUCAT, time, PC4

factorial:
PTGENDER, APOEGrp, BL_Supp_Omega3_1

SPSS:
MIXED Hippo BY PTGENDER APOEGrp BL_Supp_Omega3_1 WITH Age BMI ICV PTEDUCAT PC4 time
/CRITERIA=DFMETHOD(SATTERTHWAITE) CIN(95) MXITER(100) MXSTEP(10) SCORING(1)
SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE)
/FIXED=PTGENDER APOEGrp Age BMI ICV PTEDUCAT PC4 time BL_Supp_Omega3_1 | NOINT SSTYPE(3)
/METHOD=REML
/PRINT=SOLUTION
/RANDOM=INTERCEPT time | SUBJECT(SubjID) COVTYPE(VC)

R:
test <- lmer(Hippo~PC4+time+PTEDUCAT+ICV+APOEGrp+BMI+PTGENDER+Age+BL_Supp_Omega3_1+(time|SubjID), data=File)

##################
After investigation, I think it might be the covariance type used in SPSS compared to lmer. I can't find this answer - is there a way to used VC in lme4?
 

hlsmith

Not a robit
#2
How big is your discrepancy? I wonder too if you do get both covariance structures to align, you still may have a slight discrepancy due to say starting values for model convergence criteria.