Hi everyone,
I have a question with reference to "glm.cluster" from the package
"miceadds" and hope that someone can help me. I am trying to calculate
cluster-robust standard errors for a glm-model with multiply imputed datasets.
Everything works just fine with glm.cluster but in the end I just get an output
with estimates, standard errors and confidence intervals. But I need to report
the significance level and I cannot figure out how to obtain this.
Here is the code I ran:
mod <- lapply(impulist, FUN=function(imp1){
glm.cluster(data=imp1, formula=AV ~ p123 + ISEI,
family=binomial("logit"), cluster=imp1$ID_s)
})
# extract parameters and covariance matrix
betas <- lapply( mod , FUN = function(rr){ coef(rr) } )
vars <- lapply( mod , FUN = function(rr){ vcov(rr) } )
# conduct statistical inference
summary( mitools::MIcombine(betas,vars) )
And that's what the output looks like:
Multiple imputation results:
MIcombine.default(betas, vars)
results se (lower upper) missInfo
(Intercept) -2.785309e+00 1.344773e-01 -3.053180e+00 -2.5174383210 25 %
P123 8.273687e-05 2.834233e-05 1.797328e-05 0.0001475005 74 %
ISEI 3.788065e-02 2.347762e-03 3.270191e-02 0.0430593934 67 %
Thanks for your help!
Best,
Jen
---
Jennifer Lorenz, M.A.
Georg-August-Universit?t G?ttingen
Sozialwissenschaftliche Fakult?t
Institut f?r Erziehungswissenschaft
Lehrstuhl Schulp?dagogik / Empirische Schulforschung
e-mail: jlorenz at uni-goettingen.de<mailto:jlorenz at uni-goettingen.de>
phone: 0551-39-21411
adress: Waldweg 26, 37073 G?ttingen
room: 8.106
[[alternative HTML version deleted]]