Hello. I am fitting a simple linear model to data
with a 3-level hierarchy (species, plants within species, times within
individuals). The model is:
Random effects:
Formula: ~ 1 + time | Sp
Structure: General positive-definite
StdDev Corr
(Intercept) 1.05723143 (Inter
time 0.03824994 -0.53
Formula: ~ 1 + time | plante %in% Sp
Structure: General positive-definite
StdDev Corr
(Intercept) 0.18103949 (Inter
time 0.01453066 1
Residual 0.48031388
Fixed effects: log(FM) ~ time
Value Std.Error DF t-value p-value
(Intercept) -4.213142 0.1933546 823 -21.78972 <.0001
time 0.250802 0.0072350 823 34.66499
<.0001
Correlation:
(Intr)
time -0.537
QUESTION: I note that the correlation between the
intercepts and slopes (time) for plants within species is 1.0. This indicates
(I think) that I should not allow both to randomly vary at this level.
However, the equivalent correlation at the level of species is -0.53 and I do
what both the slopes and intercepts to vary at this species level. How can I
specify this? Specifically, that both slopes and intercepts can vary at one
level but only one of the two can vary at a lower level?
Thanks for any help.
Bill Shipley
Subject Matter Editor, Ecology
North American Editor, Annals of Botany
Département de biologie, Université de Sherbrooke,
Sherbrooke (Québec) J1K 2R1 CANADA
Bill.Shipley@USherbrooke.ca
http://callisto.si.usherb.ca:8080/bshipley/