Dear
R-helpers,
I have time
series data from 16 subjects and 2 treatment groups. The seasonal variation can
be best described by two harmonics, I called the frequencies
omega and omega2.
I now want
to test whether (1) the seasonal pattern differs between the treatments (some
kind of overall test). If this is the case, (2) I want to conduct tests to find
out whether the amplitude of the fist harmonic, the amplitude of the second
harmonic,
the phase of the first harmonic and the phase of the second harmonic differs
between the treatment groups.
My attempt
was to create a lme containing the two oscillations and the treatment group:
omegaDay=omega*Day # there is probably a more
elegant solution
omega2Day=omega2*Day
m=lme(y~(cos(omegaDay)+sin(omegaDay)+cos(omega2Day)+sin(omega2Day))*treatment,
random=~1|subject)
summary(m)
The summary
table tells me that the respective elements are significant (except for
cos(omegaDay):treatmentTRUE).
I know how to calculate the amplitudes and phases for both treatment groups
from the model estimates, but how do I know if they are significantly
different?
Thanks to
anyone who takes the time to help me with this problem!
Karin
_________________________________________________________________
Hotmail: Vertrauenswürdige E-Mails dank leistungsstarkem SPAM-Schutz.
https://signup.live.com/signup.aspx?id=60969
[[alternative HTML version deleted]]
