pcor_sum {BGGM} | R Documentation |
Partial Correlation Sum
Description
Compute and test partial correlation sums either within or between GGMs (e.g., different groups), resulting in a posterior distribution.
Usage
pcor_sum(..., iter = NULL, relations)
Arguments
... |
An object of class |
iter |
Number of iterations (posterior samples; defaults to the number in the object). |
relations |
Character string. Which partial correlations should be summed? |
Details
Some care must be taken when writing the string for partial_sum
. Below are several examples
Just a Sum: Perhaps a sum is of interest, and not necessarily the difference of two sums. This can be written as
-
partial_sum <- c("A1--A2 + A1--A3 + A1--A4")
which will sum those relations.
Comparing Sums:
When comparing sums, each must be seperated by ";
". For example,
-
partial_sum <- c("A1--A2 + A1--A3; A1--A2 + A1--A4")
which will sum both and compute the difference. Note that there cannot be more than two sums, such
that c("A1--A2 + A1--A3; A1--A2 + A1--A4; A1--A2 + A1--A5")
will result in an error.
Comparing Groups:
When more than one fitted object is suppled to object
it is assumed that the groups
should be compared for the same sum. Hence, in this case, only the sum needs to be written.
-
partial_sum <- c("A1--A2 + A1--A3 + A1--A4")
The above results in that sum being computed for each group and then compared.
Value
An object of class posterior_sum
, including the sum and possibly the difference for
two sums.
Examples
# data
Y <- bfi
# males
Y_males <- subset(Y, gender == 1, select = -c(education, gender))[,1:5]
# females
Y_females <- subset(Y, gender == 2, select = -c(education, gender))[,1:5]
# males
fit_males <- estimate(Y_males, seed = 1,
progress = FALSE)
# fit females
fit_females <- estimate(Y_females, seed = 2,
progress = FALSE)
sums <- pcor_sum(fit_males,
fit_females,
relations = "A1--A2 + A1--A3")
# print
sums
# plot difference
plot(sums)[[3]]