Hello all-
I'm fairly new to S-plus and I've been trying to implement a generalized
linear mixed model with negative binomial errors (from some insect count
data). My variables are as follows:
"count" = insect count data from branch samples
"mass" = offset to account for different sizes of samples
"type" = fixed effect with 2 levels
"collection" = 2 collection periods on same trees (repeated measure)
"site" = random effect
I used the GlmmPQL frunction from the MASS library. I estimated theta
using the "glm.nb" function, here is the call:
glmmPQL(count~ type*collection + offset(log(mass)),
random=~collection|site, family=negative.binomial(theta=1.17))
I've attached the output summary below, but my questions are:
1. Does this seem like the right approach for these data?
2. Do I have the syntax right, particularly with regards to the repeated
measure (collection)?
Any thoughts, criticisms, advice would be much appreciated!
--Kyle Apigian
kapigian@nature.berkeley.edu
-----------------------------------------------------------
Linear mixed-effects model fit by maximum likelihood
Data: NULL
AIC BIC logLik
905.5136 934.7615 -444.7568
Random effects:
Formula: ~ collection | site
Structure: General positive-definite
StdDev Corr
(Intercept) 1.0083883 (Inter
collection 0.6298937 -0.987
Residual 0.8824405
Variance function:
Structure: fixed weights
Formula: ~ invwt
Fixed effects: count ~ type * collection + offset(log(mass))
Value Std.Error DF t-value p-value
(Intercept) 3.394008 0.4086388 275 8.305642 <.0001
type -0.566635 0.1952115 275 -2.902671 0.0050
collection -1.887850 0.2638352 275 -7.155412 <.0001
type:collection 0.311640 0.1384645 275 2.250682 0.0131
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-1.160909 -0.7231607 -0.2896666 0.5418532 5.702982
Number of Observations: 286
Number of Groups: 8
---------------------------------------------------------------
|