Dear r-helpers,
I have two questions on multilevel binary and ordered regression models,
respectively:
1. Is there any r function (like lmer or glmer) to run multilevel ordered
regression models?
2. I used the glmer function to run a two-level binary logit model. I want
to make sure
that I did it correctly since I found differences between results from
running glmer and HLM (the commercial software)
Here is my model:
level 1: y*_ij = beta_0j + beta_1j*x1_ij + beta_2j*x2_ij + beta_3j*x3_ij +
epsilon_ij
where y* is a latent continuous variable and y is an observed binary
dependent variable.
y = 1 if y* >=0, otherwise y = 0, as how a binary regression model is set
up using
the latent variable approach.
x's are predictors at level 1
beta's are regression coefficients at level 1
epsilon's are error terms in level 1 equations
level 2 Eq1: beta_0j = gamma_00 + gamma_01*w1_j + gamma_02*w2_j + mu_0j
Level 2 Eq2: beta_1j = gamma_10 + gamma_11*w1_j + gamma_02*w2_j + mu_1j
w's are level 2 predictors
beta's are regression coefficients at level 1
gamma's are regression coefficients at level 2
mu's are level 2 error terms
Here are my r codes to run the model:
glmer(y ~ x1 + x2 + x3 + w1 + w2 + w1:x1 + w2:x2 + (1 + x1 | group), data
mydata, family = binomial)
Thanks!
Jun Xu, PhD
Associate Professor
Department of Sociology
Ball State University
Muncie, IN47306
[[alternative HTML version deleted]]