[R] GLMM using lme4
Daniel Malter
daniel at umd.edu
Wed Dec 3 19:14:08 CET 2008
Try length(yourvariablename) for all variables in the model. The first error
message indicates that the vectors you are using are not of the same length
(i.e. they do not contain the same number of observations).
The modeling question: Chickens are nested in nest (haha), but you don't
have a nest indicator for each chicken in your data. Yr|ID won't pick up the
dependence structure for chickens that are from the same nest.
1. Can chickens from different nests differ in mortality? - assume yes ->
nest random effect
2. Are chickens nested in nests? - clearly yes - nested affect
3. Can circumstances in a year affect mortality? - assume yes (imagine bird
flue or whatever you like) - This, to me, suggests a year fixed effect.
4. Are the same nests (or chicken mothers) measured in year 1 and year 2 -
assume no
Given the assumption above, I would try:
y=b0+b1*HO+b2*(yr==1)+nest(i)+HO(ij)+e
which would look like:
model1 <- lmer(y~HO+factor(yr)+(1|nest)+(1|nest:HO),family=binomial,1)
But ultimately you will have to answer the modeling question yourself and
anyway test the model of whether you controll sufficiently for the
dependence structure in your data. I suggest you to pick up Pinheiro and
Bates's book and Faraway's "Extending the linear model with R".
Cheers,
Daniel
-------------------------
cuncta stricte discussurus
-------------------------
-----Ursprüngliche Nachricht-----
Von: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] Im
Auftrag von Luciano La Sala
Gesendet: Wednesday, December 03, 2008 11:22 AM
An: r help
Betreff: [R] GLMM using lme4
Dear R-experts,
I am running R version 2.7.1 on Windows Vista. I have a small dataset which
consists of “chick ID”, “year (0, 1)”, “hatching order [HO, defined as
first, second and third-hatched chick]”, and the binary outcome of interest
“death (0, 1)”. So a subset of my dataset looks like this on a txt file:
y ID Yr HO
1 1 1 First
0 2 1 First
0 3 1 Second
0 4 1 First
1 5 1 First
0 6 1 Third
0 7 1 First
0 8 1 Third
0 9 1 First
0 10 1 First
1 11 1 Third
0 12 1 First
0 13 1 First
0 14 1 First
0 15 1 First
0 17 1 First
1 19 1 First
1 20 1 First
............
n ni nj nk
I need to run a GLMM using Year (Yr) and chick ID (ID) as random effects in
order to account for lack of independence at the nest level (many chicks are
siblings) and eliminate “year effect” incase a significant difference exists
between years.
Using lmer, I specified my model as follows:
model1 <- lmer(y~HO+(Yr|ID),family=binomial,1)
summary(model1)
Output:
Error en model.frame.default(data = 1, formula = y ~ HO + (Yr + ID),
drop.unused.levels = TRUE) :
las longitudes de las variables son diferentes (encontrada para 'HO')
Q1: What does this error message mean and how can I solve the problema?
Q2: Is my model correctly specified?
I look forward to hearing from you guys, always so helpful.
Lucho
______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
More information about the R-help
mailing list