Interpretation of R squared in growth curve models (especially for the effect of time-invariant variables)
Deni
Hello everyone,
I have a question regarding the interpretation of R squared in longitudinal growth curve models and I hope, someone of you can help.
About my analysis:
-4 time points (2,4,6,8; equal intervals); regarding change in a psychological variable over time
time-invariant variable/group variable: dichotome variable consisting of two groups
I have a significant effect of group on the intercept of the psychological variable, but it is very small: standardized estimate = - .06. Now I would like to find out the R² value for the effect of the group on the psychological variable, but unfortunately I do not know which one it is to interprete the growth curve analysis.
This is my model:
model <- '
i =~ 1*v2 + 1*v4 + 1*v6 + 1*v8
s =~ 0* v2 + 1* v4 + 2* v6 + 3* v8
i ~ 1
s ~ 1
i ~~ i
s ~~ s
i ~~ s
i + s ~ group
'
fitmodel <- growth(model, data=dat, estimator="MLR")
summary(fitmodel, fit.measures=TRUE, standardized=TRUE, rsquare=TRUE)
with “ i + s ~ group in the model “, my result regarding
R squared is:
R-Square:
Estimate
i 0.004
s 0.000
v2 0.693
v4 0.665
v6 0.672
v8 0.723
What I found out: The R-squared statistics regarding v2-v8 show
how much the growth factors i and s account
for the percentage of the variance in the repeated measure (psychological
variable) over the four time points. -->
Please correct me, if this is a wrong interpretation.
Now, I would like to know, if the R² values for i and s are the effect of the
group on the average initial intercept and average slope. So in my case, this would
mean, that group accounts for 0,6 % of
the variance in the average initial value of the psychological variable. Is
that correct or is there a different analysis needed? And if correct, can this interpretation
be transferred to the calculation of the r square value of group on the slope?
I am thankful for any advice and hope to hear from someone. Many thanks in advance!
Best Regards,
Deni
