1 year ago
#256396
Chris T.
Converting btergm confidence interval to standard error (in odds-ratio terms)
I'm working on R's btergm and would like to convert the coefficients as well as their standard errors in odds-ratio terms for the ease of interpretation. However, the btergm's output format differs from that of ergm in the sense that the former does not provide s.e. (instead, only confidence intervals are provided) and needs to subset relevant statistics using @ instead of $. This is not a big deal for converting the coefficients to odds-ratio form because one only needs to apply exp() on the output's coef, but cannot use the same trick on confidence interval, it will not work.
I've referenced this page, but again, the recommended method only works for standard glm or ergm output format.
I put a btergm example output at below and hope someone could shed some light on this. Any help will be appreciated.
btergm.fit <- readRDS(url("https://www.dropbox.com/s/z43t55q0cr59vpb/btergm.fit.rds?dl=1"))
summary(btergm.fit)
Estimate 2.5% 97.5%
edges -5.0498497 -5.1876 -4.9032
gwesp.fixed.0 1.6242771 1.5634 1.6900
edgecov.LSP[[i]] 2.1065906 1.8466 2.3998
edgecov.warNet[[i]] 0.2866346 0.2010 0.3804
nodecov.polity -0.0050764 -0.0132 -0.0001
nodecov.cinc 8.3308516 2.1716 12.3954
absdiff.polity -0.1257948 -0.1443 -0.1160
absdiff.cinc -4.8494093 -8.9411 0.4529
edgecov.contigMat[[i]] 3.2145679 3.1007 3.3061
#### generated using the following code ####
library(statnet)
library(btergm)
set.seed(10)
data("alliances", package = "xergm.common")
btergm.fit <- btergm(allyNet ~ edges + gwesp(0, fixed = TRUE) +
edgecov(LSP) + edgecov(warNet) + nodecov("polity") +
nodecov("cinc") + absdiff("polity") + absdiff("cinc") +
edgecov(contigMat), R = 50, parallel = "snow", ncpus = 2)r
confidence-interval
exponentiation
standard-error
0 Answers
Your Answer