Question

I’ve estimated confidence intervals for intercepts for each group in a dataset like this:

d <- data.frame(
    g=sample(letters[1:5], 100, replace=TRUE),
    y=runif(100)
)

library(lme4)

fm <- lmList(y ~ 1 | g, data=d)
ci.fm <- confint(fm)
ci.fm
# An object of class "lmList.confint"
# , , (Intercept)
# 
#       2.5 %    97.5 %
# a 0.4253236 0.6668585
# b 0.4860047 0.6794625
# c 0.4071186 0.6822713
# d 0.2785848 0.5992378
# e 0.4110070 0.6890399

There must be a way to then index each of the columns separately, but I can’t figure it out. I’ve tried looking at str(ci.fm) and attributes(ci.fm) and attempted things like attr(ci.fm, dimnames[[2]]$"2.5 %") among other similar things, all of which didn’t work. I’d be grateful for a pointer in the right direction. Thanks.

Answer 1

It is an S4 object: you can have the list of “slots” with slotNames,
and access them with @ (instead of $).
In this case, the only slot is .Data, with a leading dot,
so ci.fm@.Data contains the actual data.

str(ci.fm)
# Formal class 'lmList.confint' [package "lme4"] with 1 slots
#   ..@ .Data: num [1:5, 1:2, 1] 0.35 0.348 0.315 0.143 0.388 ...
#   .. ..- attr(*, "dimnames")=List of 3
#   .. .. ..$ : chr [1:5] "a" "b" "c" "d" ...
#   .. .. ..$ : chr [1:2] "2.5 %" "97.5 %"
#   .. .. ..$ : chr "(Intercept)"

slotNames(ci.fm)
# [1] ".Data"

str( ci.fm@.Data )  # 3-dimensional array
#  num [1:5, 1:2, 1] 0.35 0.348 0.315 0.143 0.388 ...
#  - attr(*, "dimnames")=List of 3
#   ..$ : chr [1:5] "a" "b" "c" "d" ...
#   ..$ : chr [1:2] "2.5 %" "97.5 %"
#   ..$ : chr "(Intercept)"

ci.fm@.Data[,1,1]  # First column
#         a         b         c         d         e 
# 0.3496147 0.3482908 0.3147435 0.1434213 0.3876549