It seems that in the summary method any NCT call that did not specify binary.data leads to a header stating the network is Gaussian:
|
if(object$info$call$binary.data){ |
|
data_type <- "BINARY" |
|
} else { |
|
data_type <- "GAUSSIAN" |
|
} |
This will lead to unexpected behavior when using the package in combination with bootnet:
# Load data:
library("psychonetrics")
data("Jonas")
# Make binary:
JonasBinary <- Jonas[,1:10]
JonasBinary[JonasBinary==-1] <- 0
# Make groups:
Jonas1 <- JonasBinary[Jonas$familiar == "I know Jonas (very) well!",]
Jonas2 <- JonasBinary[Jonas$familiar != "I know Jonas (very) well!",]
# Fit bootnet:
library("bootnet")
net1 <- estimateNetwork(Jonas1, default = "IsingFit")
net2 <- estimateNetwork(Jonas2, default = "IsingFit")
# Run NCT:
library("NetworkComparisonTest")
nct <- NCT(net1, net2)
# Print summary:
summary(nct)
States GAUSSIAN NETWORK COMPARISON even though the networks are Ising models. I'd just remove the data type in this header.
It seems that in the summary method any NCT call that did not specify
binary.dataleads to a header stating the network is Gaussian:NetworkComparisonTest/R/s3Methods.R
Lines 35 to 39 in 4dc74c7
This will lead to unexpected behavior when using the package in combination with bootnet:
States
GAUSSIAN NETWORK COMPARISONeven though the networks are Ising models. I'd just remove the data type in this header.